This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Update ExtUtils-MakeMaker to CPAN version 7.42
[perl5.git] / handy.h
CommitLineData
a0d0e21e 1/* handy.h
a687059c 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
da5d8dbb 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
a687059c 5 *
6e21c824
LW
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.
8d063cd8 8 *
8d063cd8
LW
9 */
10
4650c663
KW
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
6a5bc5ac
KW
14#ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */
15#define PERL_HANDY_H_
9d745869 16
24792b8d
NC
17#ifndef PERL_CORE
18# define Null(type) ((type)NULL)
954c1994
GS
19
20/*
ccfc67b7 21=head1 Handy Values
954c1994 22
78342678 23=for apidoc AmnU||Nullch
72d33970
FC
24Null character pointer. (No longer available when C<PERL_CORE> is
25defined.)
2307c6d0 26
78342678 27=for apidoc AmnU||Nullsv
72d33970 28Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
29
30=cut
31*/
32
24792b8d
NC
33# define Nullch Null(char*)
34# define Nullfp Null(PerlIO*)
35# define Nullsv Null(SV*)
36#endif
8d063cd8 37
641d3f0b
PP
38#ifdef TRUE
39#undef TRUE
40#endif
41#ifdef FALSE
42#undef FALSE
43#endif
44#define TRUE (1)
45#define FALSE (0)
46
cf3f0ffb
DM
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;
49 * e.g.:
50 *
51 * const SV *sv = ...;
52 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
53 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
54 */
55
b1bc3f34 56#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
f065df04 57# define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
b1bc3f34
NC
58#else
59# define MUTABLE_PTR(p) ((void *) (p))
60#endif
61
a062e10d 62#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 63#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 64#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 65#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 66#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 67#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 68
f789f6a4 69#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
70# include <stdbool.h>
71# ifndef HAS_BOOL
72# define HAS_BOOL 1
73# endif
74#endif
75
8e84507e 76/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
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
8e84507e 79 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
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
5d94fbed 83*/
3609ea0d 84#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 85# ifndef PERL_BOOL_AS_CHAR
5d94fbed 86# ifndef HAS_BOOL
c1d22f6b 87# define HAS_BOOL 1
5d94fbed 88# endif
f789f6a4 89# endif
5d94fbed
AD
90#endif
91
92#ifndef HAS_BOOL
f789f6a4
FC
93# ifdef bool
94# undef bool
95# endif
70d5cb32 96# define bool char
c1d22f6b 97# define HAS_BOOL 1
a687059c 98#endif
0d3e774c 99
25ba28ce
KW
100/*
101=for apidoc Am|bool|cBOOL|bool expr
102
103Cast-to-bool. A simple S<C<(bool) I<expr>>> cast may not do the right thing:
104if C<bool> is defined as C<char>, for example, then the cast from C<int> is
105implementation-defined.
106
107C<(bool)!!(cbool)> in a ternary triggers a bug in xlc on AIX
108
109=cut
110*/
18f5643b 111#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 112
46c6c7e2 113/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
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. */
46c6c7e2
JH
117#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
118# define FUNCTION__ __func__
7adf2470 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). */
07798b17 120# define FUNCTION__ ""
46c6c7e2 121#else
07798b17 122# define FUNCTION__ __FUNCTION__ /* Common extension. */
46c6c7e2
JH
123#endif
124
27d4fb96
PP
125/* XXX A note on the perl source internal type system. The
126 original intent was that I32 be *exactly* 32 bits.
127
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
24fef2a7 134
dc45a647
MB
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.
693762b4 138
dc45a647
MB
139 Similarly, there is no guarantee that I16 and U16 have exactly 16
140 bits.
693762b4 141
8e84507e
NIS
142 For dealing with issues that may arise from various 32/64-bit
143 systems, we will ask Configure to check out
8175356b 144
3609ea0d
JH
145 SHORTSIZE == sizeof(short)
146 INTSIZE == sizeof(int)
147 LONGSIZE == sizeof(long)
dc45a647 148 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 149 PTRSIZE == sizeof(void *)
dc45a647
MB
150 DOUBLESIZE == sizeof(double)
151 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 152
27d4fb96
PP
153*/
154
69512466
JH
155#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
156# include <inttypes.h>
dd0eed91
JH
157# ifdef INT32_MIN_BROKEN
158# undef INT32_MIN
159# define INT32_MIN (-2147483647-1)
160# endif
161# ifdef INT64_MIN_BROKEN
162# undef INT64_MIN
163# define INT64_MIN (-9223372036854775807LL-1)
164# endif
69512466
JH
165#endif
166
8175356b
JH
167typedef I8TYPE I8;
168typedef U8TYPE U8;
169typedef I16TYPE I16;
170typedef U16TYPE U16;
171typedef I32TYPE I32;
172typedef U32TYPE U32;
16d89be8 173
74b807c7 174#ifdef QUADKIND
8175356b
JH
175typedef I64TYPE I64;
176typedef U64TYPE U64;
16d89be8 177#endif
8175356b 178
d8668976 179#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 180
5ff3f7a4
GS
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
185
5ff3f7a4
GS
186#define I16_MAX INT16_MAX
187#define I16_MIN INT16_MIN
188#define U16_MAX UINT16_MAX
189#define U16_MIN UINT16_MIN
190
5ff3f7a4
GS
191#define I32_MAX INT32_MAX
192#define I32_MIN INT32_MIN
0e983133
GS
193#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
194# define U32_MAX UINT32_MAX
195#else
196# define U32_MAX 4294967295U
197#endif
5ff3f7a4
GS
198#define U32_MIN UINT32_MIN
199
200#else
201
5c9fa16e
KA
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. */
27d4fb96
PP
204#define U8_MAX PERL_UCHAR_MAX
205#define U8_MIN PERL_UCHAR_MIN
79072805 206
27d4fb96
PP
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
79072805 211
c4f23d77 212#if LONGSIZE > 4
27d4fb96
PP
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
79072805 217#else
27d4fb96
PP
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
79072805
LW
222#endif
223
5ff3f7a4
GS
224#endif
225
247cee9f
KW
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.
229 *
230 * Restrict these to core for now until we are more certain this is a good
231 * idea. */
232#if defined(PERL_CORE) || defined(PERL_EXT)
233# ifdef I_STDINT
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;
238# else
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;
243# endif
244#endif
245
464decb6
KW
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 */
248#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
fc36a67e
PP
249#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
250#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
251
88794300 252/* Unused by core; should be deprecated */
ff68c719 253#define Ctl(ch) ((ch) & 037)
8d063cd8 254
98fce2a4
KW
255#if defined(PERL_CORE) || defined(PERL_EXT)
256# ifndef MIN
257# define MIN(a,b) ((a) < (b) ? (a) : (b))
258# endif
259# ifndef MAX
260# define MAX(a,b) ((a) > (b) ? (a) : (b))
261# endif
262#endif
263
84ff4fa9
KW
264/* Returns a boolean as to whether the input unsigned number is a power of 2
265 * (2**0, 2**1, etc). In other words if it has just a single bit set.
266 * If not, subtracting 1 would leave the uppermost bit set, so the & would
267 * yield non-zero */
268#if defined(PERL_CORE) || defined(PERL_EXT)
011b1419 269# define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
84ff4fa9
KW
270#endif
271
8d9433eb
KW
272/*
273=for apidoc Am|void|__ASSERT_|bool expr
274
275This is a helper macro to avoid preprocessor issues, replaced by nothing
276unless under DEBUGGING, where it expands to an assert of its argument,
277followed by a comma (hence the comma operator). If we just used a straight
278assert(), we would get a comma with nothing before it when not DEBUGGING.
279
280=cut
281
282We also use empty definition under Coverity since the __ASSERT__
283checks often check for things that Really Cannot Happen, and Coverity
284detects that and gets all excited. */
3e94db23 285
e7ae132e
KW
286#if defined(DEBUGGING) && !defined(__COVERITY__) \
287 && ! defined(PERL_SMALL_MACRO_BUFFER)
0f092d08
KW
288# define __ASSERT_(statement) assert(statement),
289#else
290# define __ASSERT_(statement)
291#endif
292
3fe05580 293/*
a4ee4fb5 294=head1 SV Manipulation Functions
3fe05580 295
3bb9fd01 296=for apidoc Ama|SV*|newSVpvs|"literal string"
1568d13a 297Like C<newSVpvn>, but takes a literal string instead of a
30a15352 298string/length pair.
3fe05580 299
3bb9fd01 300=for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
1568d13a 301Like C<newSVpvn_flags>, but takes a literal string instead of
30a15352 302a string/length pair.
84bafc02 303
3bb9fd01 304=for apidoc Ama|SV*|newSVpvs_share|"literal string"
1568d13a 305Like C<newSVpvn_share>, but takes a literal string instead of
30a15352 306a string/length pair and omits the hash parameter.
3fe05580 307
3bb9fd01 308=for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
1568d13a 309Like C<sv_catpvn_flags>, but takes a literal string instead
30a15352 310of a string/length pair.
9dcc53ea 311
3bb9fd01 312=for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
1568d13a 313Like C<sv_catpvn_nomg>, but takes a literal string instead of
0c395ea5 314a string/length pair.
9dcc53ea 315
3bb9fd01 316=for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
1568d13a 317Like C<sv_catpvn>, but takes a literal string instead of a
0c395ea5 318string/length pair.
3fe05580 319
3bb9fd01 320=for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
1568d13a 321Like C<sv_catpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
322string/length pair.
323
3bb9fd01 324=for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
1568d13a 325Like C<sv_setpvn>, but takes a literal string instead of a
0c395ea5 326string/length pair.
3fe05580 327
3bb9fd01 328=for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
1568d13a 329Like C<sv_setpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
330string/length pair.
331
3bb9fd01 332=for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
1568d13a 333Like C<sv_setref_pvn>, but takes a literal string instead of
0c395ea5 334a string/length pair.
9dcc53ea 335
3fe05580
MHM
336=head1 Memory Management
337
3bb9fd01 338=for apidoc Ama|char*|savepvs|"literal string"
1568d13a 339Like C<savepvn>, but takes a literal string instead of a
30a15352 340string/length pair.
3fe05580 341
3bb9fd01 342=for apidoc Ama|char*|savesharedpvs|"literal string"
9dcc53ea
Z
343A version of C<savepvs()> which allocates the duplicate string in memory
344which is shared between threads.
345
3fe05580
MHM
346=head1 GV Functions
347
3bb9fd01 348=for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
1568d13a 349Like C<gv_stashpvn>, but takes a literal string instead of a
0c395ea5 350string/length pair.
3fe05580
MHM
351
352=head1 Hash Manipulation Functions
353
3bb9fd01 354=for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
1568d13a 355Like C<hv_fetch>, but takes a literal string instead of a
0c395ea5 356string/length pair.
3fe05580 357
3bb9fd01 358=for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
1568d13a 359Like C<hv_store>, but takes a literal string instead of a
0c395ea5 360string/length pair
3fe05580
MHM
361and omits the hash parameter.
362
510966aa
Z
363=head1 Lexer interface
364
3bb9fd01 365=for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
510966aa 366
1568d13a 367Like L</lex_stuff_pvn>, but takes a literal string instead of
0c395ea5 368a string/length pair.
510966aa 369
3fe05580
MHM
370=cut
371*/
372
a34e53fc
KW
373/*
374=head1 Handy Values
2efa8cc7 375
a34e53fc
KW
376=for apidoc Amu|pair|STR_WITH_LEN|"literal string"
377
378Returns two comma separated tokens of the input literal string, and its length.
379This is convenience macro which helps out in some API calls.
380Note that it can't be used as an argument to macros or functions that under
381some configurations might be macros, which means that it requires the full
382Perl_xxx(aTHX_ ...) form for any API calls where it's used.
383
384=cut
385*/
386
387
388#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
ba3a79e7
GA
389
390/* STR_WITH_LEN() shortcuts */
391#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
392#define newSVpvs_flags(str,flags) \
393 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 394#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
395#define sv_catpvs_flags(sv, str, flags) \
396 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
397#define sv_catpvs_nomg(sv, str) \
398 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
399#define sv_catpvs(sv, str) \
400 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
401#define sv_catpvs_mg(sv, str) \
402 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 403#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
404#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
405#define sv_setref_pvs(rv, classname, str) \
406 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 407#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
408#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
409#define gv_stashpvs(str, create) \
410 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
411#define gv_fetchpvs(namebeg, add, sv_type) \
412 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
413#define gv_fetchpvn(namebeg, len, add, sv_type) \
414 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
415#define sv_catxmlpvs(dsv, str, utf8) \
416 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 417
ba3a79e7 418
510966aa
Z
419#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
420
b96d8cd9
NC
421#define get_cvs(str, flags) \
422 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 423
954c1994 424/*
ccfc67b7
JH
425=head1 Miscellaneous Functions
426
954c1994 427=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
428Test two C<NUL>-terminated strings to see if they are different. Returns true
429or false.
954c1994
GS
430
431=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
432Test two C<NUL>-terminated strings to see if they are equal. Returns true or
433false.
954c1994
GS
434
435=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
436Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
437second, C<s2>. Returns true or false.
954c1994
GS
438
439=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
440Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
441equal to the second, C<s2>. Returns true or false.
954c1994
GS
442
443=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
444Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
445the second, C<s2>. Returns true or false.
954c1994
GS
446
447=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
448Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
449or equal to the second, C<s2>. Returns true or false.
954c1994
GS
450
451=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
452Test two C<NUL>-terminated strings to see if they are different. The C<len>
453parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
454wrapper for C<strncmp>).
455
456=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
457Test two C<NUL>-terminated strings to see if they are equal. The C<len>
458parameter indicates the number of bytes to compare. Returns true or false. (A
459wrapper for C<strncmp>).
954c1994 460
bd18bd40
KW
461=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
462Test two buffers (which may contain embedded C<NUL> characters, to see if they
463are equal. The C<len> parameter indicates the number of bytes to compare.
464Returns zero if equal, or non-zero if non-equal.
465
3bb9fd01 466=for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
467Like L</memEQ>, but the second string is a literal enclosed in double quotes,
468C<l1> gives the number of bytes in C<s1>.
469Returns zero if equal, or non-zero if non-equal.
470
bd18bd40
KW
471=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
472Test two buffers (which may contain embedded C<NUL> characters, to see if they
473are not equal. The C<len> parameter indicates the number of bytes to compare.
474Returns zero if non-equal, or non-zero if equal.
475
3bb9fd01 476=for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
477Like L</memNE>, but the second string is a literal enclosed in double quotes,
478C<l1> gives the number of bytes in C<s1>.
479Returns zero if non-equal, or zero if non-equal.
480
954c1994 481=cut
fc169e00
KW
482
483New macros should use the following conventions for their names (which are
484based on the underlying C library functions):
485
486 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
487
488 Each has two main parameters, string-like operands that are compared
489 against each other, as specified by the macro name. Some macros may
490 additionally have one or potentially even two length parameters. If a length
491 parameter applies to both string parameters, it will be positioned third;
492 otherwise any length parameter immediately follows the string parameter it
493 applies to.
494
495 If the prefix to the name is 'str', the string parameter is a pointer to a C
496 language string. Such a string does not contain embedded NUL bytes; its
497 length may be unknown, but can be calculated by C<strlen()>, since it is
498 terminated by a NUL, which isn't included in its length.
499
500 The optional 'n' following 'str' means that that there is a third parameter,
501 giving the maximum number of bytes to look at in each string. Even if both
502 strings are longer than the length parameter, those extra bytes will be
503 unexamined.
504
505 The 's' suffix means that the 2nd byte string parameter is a literal C
506 double-quoted string. Its length will automatically be calculated by the
507 macro, so no length parameter will ever be needed for it.
508
509 If the prefix is 'mem', the string parameters don't have to be C strings;
510 they may contain embedded NUL bytes, do not necessarily have a terminating
511 NUL, and their lengths can be known only through other means, which in
512 practice are additional parameter(s) passed to the function. All 'mem'
513 functions have at least one length parameter. Barring any 'l' or 's' suffix,
514 there is a single length parameter, in position 3, which applies to both
515 string parameters. The 's' suffix means, as described above, that the 2nd
516 string is a literal double-quoted C string (hence its length is calculated by
517 the macro, and the length parameter to the function applies just to the first
518 string parameter, and hence is positioned just after it). An 'l' suffix
519 means that the 2nd string parameter has its own length parameter, and the
520 signature will look like memFOOl(s1, l1, s2, l2).
521
522 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
523 substring of the 1st string. 'P' if present indicates that the substring
524 must be a "proper" one in tha mathematical sense that the first one must be
525 strictly larger than the 2nd.
526
954c1994
GS
527*/
528
62946e08 529
75400963
KW
530#define strNE(s1,s2) (strcmp(s1,s2) != 0)
531#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
532#define strLT(s1,s2) (strcmp(s1,s2) < 0)
533#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
534#define strGT(s1,s2) (strcmp(s1,s2) > 0)
535#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 536
75400963
KW
537#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
538#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 539
9d3980bc
KW
540#define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
541#define memNE(s1,s2,l) (! memEQ(s1,s2,l))
36477c24 542
085b7534 543/* memEQ and memNE where second comparand is a string constant */
568a785a 544#define memEQs(s1, l, s2) \
777fa2cb 545 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 546#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 547
fdbb9a7c
KW
548/* Keep these private until we decide it was a good idea */
549#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
550
551#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
552
bdb7e3f0 553#define memBEGINs(s1, l, s2) \
30a6480c 554 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 555 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158 556#define memBEGINPs(s1, l, s2) \
30a6480c 557 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
de627158 558 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0 559#define memENDs(s1, l, s2) \
30a6480c 560 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 561 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424 562#define memENDPs(s1, l, s2) \
30a6480c 563 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
b80f8424 564 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 565#endif /* End of making macros private */
bdb7e3f0 566
062b6850
KW
567#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
568#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
569#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
570#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
571
bbce6d69
PP
572/*
573 * Character classes.
574 *
575 * Unfortunately, the introduction of locales means that we
576 * can't trust isupper(), etc. to tell the truth. And when
577 * it comes to /\w+/ with tainting enabled, we *must* be able
578 * to trust our character classes.
579 *
580 * Therefore, the default tests in the text of Perl will be
581 * independent of locale. Any code that wants to depend on
582 * the current locale will use the tests that begin with "lc".
583 */
584
2304df62
AD
585#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
586# ifndef CTYPE256
587# define CTYPE256
588# endif
589#endif
590
954c1994 591/*
ccfc67b7 592
dcccc8ff 593=head1 Character classification
243effed
KW
594This section is about functions (really macros) that classify characters
595into types, such as punctuation versus alphabetic, etc. Most of these are
596analogous to regular expression character classes. (See
597L<perlrecharclass/POSIX Character Classes>.) There are several variants for
598each class. (Not all macros have all variants; each item below lists the
599ones valid for it.) None are affected by C<use bytes>, and only the ones
600with C<LC> in the name are affected by the current locale.
601
d713f9d9
KW
602The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
603treating it as a code point, and returns a boolean as to whether or not the
604character represented by it is (or on non-ASCII platforms, corresponds to) an
6aff1f14
KW
605ASCII character in the named class based on platform, Unicode, and Perl rules.
606If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 607
c98722a4 608Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
609with no suffix C<"_A">. This variant is used to emphasize by its name that
610only ASCII-range characters can return TRUE.
4b9734bf 611
d60679e1 612Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
613onto the platform. That is, the code points that are ASCII are unaffected,
614since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
615as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
616true when called with the code point 0xDF, which is a word character in both
4650c663 617ASCII and EBCDIC (though it represents different characters in each).
d713f9d9
KW
618If the input is a number that doesn't fit in an octet, FALSE is returned.
619(Perl's documentation uses a colloquial definition of Latin-1, to include all
620code points below 256.)
243effed 621
d713f9d9
KW
622Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
623inputs below 256, but if the code point is larger than 255, Unicode rules are
624used to determine if it is in the character class. For example,
d0da05db 625C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 626WITH MACRON in Unicode, and is a word character.
243effed 627
059703b0
KW
628Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
629but are used for UTF-8 encoded strings. The two forms are different names for
630the same thing. Each call to one of these classifies the first character of
631the string starting at C<p>. The second parameter, C<e>, points to anywhere in
632the string beyond the first character, up to one byte past the end of the
633entire string. Although both variants are identical, the suffix C<_safe> in
634one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
635provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
636C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
637some way, the program may croak, or the function may return FALSE, at the
638discretion of the implementation, and subject to change in future releases.
243effed 639
d713f9d9
KW
640Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
641but the result is based on the current locale, which is what C<LC> in the name
642stands for. If Perl can determine that the current locale is a UTF-8 locale,
643it uses the published Unicode rules; otherwise, it uses the C library function
644that gives the named classification. For example, C<isDIGIT_LC()> when not in
645a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
646returned if the input won't fit into an octet. On some platforms where the C
647library function is known to be defective, Perl changes its result to follow
648the POSIX standard's rules.
243effed 649
d713f9d9
KW
650Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
651than 256, but for larger ones it returns the Unicode classification of the code
652point.
243effed 653
059703b0
KW
654Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
655C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
656are different names for the same thing. Each call to one of these classifies
657the first character of the string starting at C<p>. The second parameter,
658C<e>, points to anywhere in the string beyond the first character, up to one
659byte past the end of the entire string. Although both variants are identical,
660the suffix C<_safe> in one name emphasizes that it will not attempt to read
661beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
662is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
663character is malformed in some way, the program may croak, or the function may
664return FALSE, at the discretion of the implementation, and subject to change in
665future releases.
ccfc67b7 666
d713f9d9
KW
667=for apidoc Am|bool|isALPHA|int ch
668Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
669analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
670See the L<top of this section|/Character classification> for an explanation of
671variants
059703b0
KW
672C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>,
673C<isALPHA_utf8_safe>, C<isALPHA_LC>, C<isALPHA_LC_uvchr>, C<isALPHA_LC_utf8>,
674and C<isALPHA_LC_utf8_safe>.
8a58bdcf 675
f16858ed
KW
676=cut
677
678Here and below, we add the protoypes of these macros for downstream programs
679that would be interested in them, such as Devel::PPPort
680
681=for apidoc Amh|bool|isALPHA_A|int ch
682=for apidoc Amh|bool|isALPHA_L1|int ch
683=for apidoc Amh|bool|isALPHA_uvchr|int ch
684=for apidoc Amh|bool|isALPHA_utf8_safe|U8 * s|U8 * end
d23c7e08 685=for apidoc Amh|bool|isALPHA_utf8|U8 * s|U8 * end
f16858ed
KW
686=for apidoc Amh|bool|isALPHA_LC|int ch
687=for apidoc Amh|bool|isALPHA_LC_uvchr|int ch
688=for apidoc Amh|bool|isALPHA_LC_utf8_safe|U8 * s| U8 *end
689
d713f9d9
KW
690=for apidoc Am|bool|isALPHANUMERIC|int ch
691Returns a boolean indicating whether the specified character is one of
692C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
693See the L<top of this section|/Character classification> for an explanation of
694variants
d0da05db 695C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
059703b0
KW
696C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>,
697C<isALPHANUMERIC_LC_uvchr>, C<isALPHANUMERIC_LC_utf8>, and
698C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 699
255b632a
KW
700A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
701this corresponds to the C language alphanumeric definition). Also
702there are the variants
703C<isALNUMC_A>, C<isALNUMC_L1>
704C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
705
f16858ed
KW
706=for apidoc Amh|bool|isALPHANUMERIC_A|int ch
707=for apidoc Amh|bool|isALPHANUMERIC_L1|int ch
708=for apidoc Amh|bool|isALPHANUMERIC_uvchr|int ch
709=for apidoc Amh|bool|isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
d23c7e08 710=for apidoc Amh|bool|isALPHANUMERIC_utf8|U8 * s|U8 * end
f16858ed
KW
711=for apidoc Amh|bool|isALPHANUMERIC_LC|int ch
712=for apidoc Amh|bool|isALPHANUMERIC_LC_uvchr|int ch
713=for apidoc Amh|bool|isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
714=for apidoc Amh|bool|isALNUMC|int ch
715=for apidoc Amh|bool|isALNUMC_A|int ch
716=for apidoc Amh|bool|isALNUMC_L1|int ch
717=for apidoc Amh|bool|isALNUMC_LC|int ch
718=for apidoc Amh|bool|isALNUMC_LC_uvchr|int ch
719
d713f9d9 720=for apidoc Am|bool|isASCII|int ch
8a58bdcf 721Returns a boolean indicating whether the specified character is one of the 128
243effed 722characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 723On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
724character corresponds to an ASCII character. Variants C<isASCII_A()> and
725C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
726See the L<top of this section|/Character classification> for an explanation of
727variants
059703b0
KW
728C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_utf8_safe>, C<isASCII_LC>,
729C<isASCII_LC_uvchr>, C<isASCII_LC_utf8>, and C<isASCII_LC_utf8_safe>.
730Note, however, that some platforms do not have the C library routine
731C<isascii()>. In these cases, the variants whose names contain C<LC> are the
732same as the corresponding ones without.
243effed 733
f16858ed
KW
734=for apidoc Amh|bool|isASCII_A|int ch
735=for apidoc Amh|bool|isASCII_L1|int ch
736=for apidoc Amh|bool|isASCII_uvchr|int ch
737=for apidoc Amh|bool|isASCII_utf8_safe|U8 * s|U8 * end
d23c7e08 738=for apidoc Amh|bool|isASCII_utf8|U8 * s|U8 * end
f16858ed
KW
739=for apidoc Amh|bool|isASCII_LC|int ch
740=for apidoc Amh|bool|isASCII_LC_uvchr|int ch
741=for apidoc Amh|bool|isASCII_LC_utf8_safe|U8 * s| U8 *end
742
d98532ea
KW
743Also note, that because all ASCII characters are UTF-8 invariant (meaning they
744have the exact same representation (always a single byte) whether encoded in
745UTF-8 or not), C<isASCII> will give the correct results when called with any
059703b0
KW
746byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
747C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
d98532ea 748
243effed
KW
749=for apidoc Am|bool|isBLANK|char ch
750Returns a boolean indicating whether the specified character is a
6aff1f14 751character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
752See the L<top of this section|/Character classification> for an explanation of
753variants
059703b0
KW
754C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>,
755C<isBLANK_utf8_safe>, C<isBLANK_LC>, C<isBLANK_LC_uvchr>, C<isBLANK_LC_utf8>,
756and C<isBLANK_LC_utf8_safe>. Note,
da8c1a98
KW
757however, that some platforms do not have the C library routine
758C<isblank()>. In these cases, the variants whose names contain C<LC> are
759the same as the corresponding ones without.
243effed 760
f16858ed
KW
761=for apidoc Amh|bool|isBLANK_A|int ch
762=for apidoc Amh|bool|isBLANK_L1|int ch
763=for apidoc Amh|bool|isBLANK_uvchr|int ch
764=for apidoc Amh|bool|isBLANK_utf8_safe|U8 * s|U8 * end
d23c7e08 765=for apidoc Amh|bool|isBLANK_utf8|U8 * s|U8 * end
f16858ed
KW
766=for apidoc Amh|bool|isBLANK_LC|int ch
767=for apidoc Amh|bool|isBLANK_LC_uvchr|int ch
768=for apidoc Amh|bool|isBLANK_LC_utf8_safe|U8 * s| U8 *end
769
243effed
KW
770=for apidoc Am|bool|isCNTRL|char ch
771Returns a boolean indicating whether the specified character is a
6aff1f14 772control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
773See the L<top of this section|/Character classification> for an explanation of
774variants
059703b0
KW
775C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>,
776C<isCNTRL_utf8_safe>, C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, C<isCNTRL_LC_utf8>
777and C<isCNTRL_LC_utf8_safe>. On EBCDIC
da8c1a98 778platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 779
f16858ed
KW
780=for apidoc Amh|bool|isCNTRL_A|int ch
781=for apidoc Amh|bool|isCNTRL_L1|int ch
782=for apidoc Amh|bool|isCNTRL_uvchr|int ch
783=for apidoc Amh|bool|isCNTRL_utf8_safe|U8 * s|U8 * end
d23c7e08 784=for apidoc Amh|bool|isCNTRL_utf8|U8 * s|U8 * end
f16858ed
KW
785=for apidoc Amh|bool|isCNTRL_LC|int ch
786=for apidoc Amh|bool|isCNTRL_LC_uvchr|int ch
787=for apidoc Amh|bool|isCNTRL_LC_utf8_safe|U8 * s| U8 *end
788
954c1994 789=for apidoc Am|bool|isDIGIT|char ch
2787a470 790Returns a boolean indicating whether the specified character is a
6aff1f14 791digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 792Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
793See the L<top of this section|/Character classification> for an explanation of
794variants
059703b0
KW
795C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>,
796C<isDIGIT_LC_uvchr>, C<isDIGIT_LC_utf8>, and C<isDIGIT_LC_utf8_safe>.
243effed 797
f16858ed
KW
798=for apidoc Amh|bool|isDIGIT_A|int ch
799=for apidoc Amh|bool|isDIGIT_L1|int ch
800=for apidoc Amh|bool|isDIGIT_uvchr|int ch
801=for apidoc Amh|bool|isDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 802=for apidoc Amh|bool|isDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
803=for apidoc Amh|bool|isDIGIT_LC|int ch
804=for apidoc Amh|bool|isDIGIT_LC_uvchr|int ch
805=for apidoc Amh|bool|isDIGIT_LC_utf8_safe|U8 * s| U8 *end
806
243effed
KW
807=for apidoc Am|bool|isGRAPH|char ch
808Returns a boolean indicating whether the specified character is a
6aff1f14 809graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 810See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
811variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>,
812C<isGRAPH_utf8_safe>, C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>,
813C<isGRAPH_LC_utf8_safe>, and C<isGRAPH_LC_utf8_safe>.
954c1994 814
f16858ed
KW
815=for apidoc Amh|bool|isGRAPH_A|int ch
816=for apidoc Amh|bool|isGRAPH_L1|int ch
817=for apidoc Amh|bool|isGRAPH_uvchr|int ch
818=for apidoc Amh|bool|isGRAPH_utf8_safe|U8 * s|U8 * end
d23c7e08 819=for apidoc Amh|bool|isGRAPH_utf8|U8 * s|U8 * end
f16858ed
KW
820=for apidoc Amh|bool|isGRAPH_LC|int ch
821=for apidoc Amh|bool|isGRAPH_LC_uvchr|int ch
822=for apidoc Amh|bool|isGRAPH_LC_utf8_safe|U8 * s| U8 *end
823
0c82b6df 824=for apidoc Am|bool|isLOWER|char ch
2787a470 825Returns a boolean indicating whether the specified character is a
6aff1f14 826lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
827See the L<top of this section|/Character classification> for an explanation of
828variants
059703b0
KW
829C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>,
830C<isLOWER_utf8_safe>, C<isLOWER_LC>, C<isLOWER_LC_uvchr>, C<isLOWER_LC_utf8>,
831and C<isLOWER_LC_utf8_safe>.
0c82b6df 832
f16858ed
KW
833=for apidoc Amh|bool|isLOWER_A|int ch
834=for apidoc Amh|bool|isLOWER_L1|int ch
835=for apidoc Amh|bool|isLOWER_uvchr|int ch
836=for apidoc Amh|bool|isLOWER_utf8_safe|U8 * s|U8 * end
d23c7e08 837=for apidoc Amh|bool|isLOWER_utf8|U8 * s|U8 * end
f16858ed
KW
838=for apidoc Amh|bool|isLOWER_LC|int ch
839=for apidoc Amh|bool|isLOWER_LC_uvchr|int ch
840=for apidoc Amh|bool|isLOWER_LC_utf8_safe|U8 * s| U8 *end
841
c99e91e9 842=for apidoc Am|bool|isOCTAL|char ch
2787a470 843Returns a boolean indicating whether the specified character is an
6aff1f14 844octal digit, [0-7].
243effed
KW
845The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
846C<isOCTAL>.
847
f16858ed
KW
848=for apidoc Amh|bool|isOCTAL_A|int ch
849=for apidoc Amh|bool|isOCTAL_L1|int ch
850
243effed
KW
851=for apidoc Am|bool|isPUNCT|char ch
852Returns a boolean indicating whether the specified character is a
6aff1f14
KW
853punctuation character, analogous to C<m/[[:punct:]]/>.
854Note that the definition of what is punctuation isn't as
243effed
KW
855straightforward as one might desire. See L<perlrecharclass/POSIX Character
856Classes> for details.
dcccc8ff 857See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
858variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>,
859C<isPUNCT_utf8_safe>, C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, C<isPUNCT_LC_utf8>,
860and C<isPUNCT_LC_utf8_safe>.
c99e91e9 861
f16858ed
KW
862=for apidoc Amh|bool|isPUNCT_A|int ch
863=for apidoc Amh|bool|isPUNCT_L1|int ch
864=for apidoc Amh|bool|isPUNCT_uvchr|int ch
865=for apidoc Amh|bool|isPUNCT_utf8_safe|U8 * s|U8 * end
d23c7e08 866=for apidoc Amh|bool|isPUNCT_utf8|U8 * s|U8 * end
f16858ed
KW
867=for apidoc Amh|bool|isPUNCT_LC|int ch
868=for apidoc Amh|bool|isPUNCT_LC_uvchr|int ch
869=for apidoc Amh|bool|isPUNCT_LC_utf8_safe|U8 * s| U8 *end
870
0c82b6df 871=for apidoc Am|bool|isSPACE|char ch
2787a470 872Returns a boolean indicating whether the specified character is a
6aff1f14 873whitespace character. This is analogous
398d098a 874to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 875this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
876locale forms of this macro (the ones with C<LC> in their names) matched
877precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
878in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
879(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
880See the L<top of this section|/Character classification> for an explanation of
881variants
059703b0
KW
882C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>,
883C<isSPACE_utf8_safe>, C<isSPACE_LC>, C<isSPACE_LC_uvchr>, C<isSPACE_LC_utf8>,
884and C<isSPACE_LC_utf8_safe>.
0c82b6df 885
f16858ed
KW
886=for apidoc Amh|bool|isSPACE_A|int ch
887=for apidoc Amh|bool|isSPACE_L1|int ch
888=for apidoc Amh|bool|isSPACE_uvchr|int ch
889=for apidoc Amh|bool|isSPACE_utf8_safe|U8 * s|U8 * end
d23c7e08 890=for apidoc Amh|bool|isSPACE_utf8|U8 * s|U8 * end
f16858ed
KW
891=for apidoc Amh|bool|isSPACE_LC|int ch
892=for apidoc Amh|bool|isSPACE_LC_uvchr|int ch
893=for apidoc Amh|bool|isSPACE_LC_utf8_safe|U8 * s| U8 *end
894
398d098a
KW
895=for apidoc Am|bool|isPSXSPC|char ch
896(short for Posix Space)
779cf272
KW
897Starting in 5.18, this is identical in all its forms to the
898corresponding C<isSPACE()> macros.
398d098a
KW
899The locale forms of this macro are identical to their corresponding
900C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
901non-locale forms differ from their C<isSPACE()> forms only in that the
902C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
903Otherwise they are identical. Thus this macro is analogous to what
904C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 905See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
906variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
907C<isPSXSPC_utf8_safe>, C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>,
908C<isPSXSPC_LC_utf8>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 909
f16858ed
KW
910=for apidoc Amh|bool|isPSXSPC_A|int ch
911=for apidoc Amh|bool|isPSXSPC_L1|int ch
912=for apidoc Amh|bool|isPSXSPC_uvchr|int ch
913=for apidoc Amh|bool|isPSXSPC_utf8_safe|U8 * s|U8 * end
d23c7e08 914=for apidoc Amh|bool|isPSXSPC_utf8|U8 * s|U8 * end
f16858ed
KW
915=for apidoc Amh|bool|isPSXSPC_LC|int ch
916=for apidoc Amh|bool|isPSXSPC_LC_uvchr|int ch
917=for apidoc Amh|bool|isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
918
954c1994 919=for apidoc Am|bool|isUPPER|char ch
2787a470 920Returns a boolean indicating whether the specified character is an
6aff1f14 921uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 922See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
923variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>,
924C<isUPPER_utf8_safe>, C<isUPPER_LC>, C<isUPPER_LC_uvchr>, C<isUPPER_LC_utf8>,
925and C<isUPPER_LC_utf8_safe>.
954c1994 926
f16858ed
KW
927=for apidoc Amh|bool|isUPPER_A|int ch
928=for apidoc Amh|bool|isUPPER_L1|int ch
929=for apidoc Amh|bool|isUPPER_uvchr|int ch
930=for apidoc Amh|bool|isUPPER_utf8_safe|U8 * s|U8 * end
d23c7e08 931=for apidoc Amh|bool|isUPPER_utf8|U8 * s|U8 * end
f16858ed
KW
932=for apidoc Amh|bool|isUPPER_LC|int ch
933=for apidoc Amh|bool|isUPPER_LC_uvchr|int ch
934=for apidoc Amh|bool|isUPPER_LC_utf8_safe|U8 * s| U8 *end
935
243effed 936=for apidoc Am|bool|isPRINT|char ch
8eea39dd 937Returns a boolean indicating whether the specified character is a
6aff1f14 938printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
939See the L<top of this section|/Character classification> for an explanation of
940variants
059703b0
KW
941C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>,
942C<isPRINT_utf8_safe>, C<isPRINT_LC>, C<isPRINT_LC_uvchr>, C<isPRINT_LC_utf8>,
943and C<isPRINT_LC_utf8_safe>.
243effed 944
f16858ed
KW
945=for apidoc Amh|bool|isPRINT_A|int ch
946=for apidoc Amh|bool|isPRINT_L1|int ch
947=for apidoc Amh|bool|isPRINT_uvchr|int ch
948=for apidoc Amh|bool|isPRINT_utf8_safe|U8 * s|U8 * end
d23c7e08 949=for apidoc Amh|bool|isPRINT_utf8|U8 * s|U8 * end
f16858ed
KW
950=for apidoc Amh|bool|isPRINT_LC|int ch
951=for apidoc Amh|bool|isPRINT_LC_uvchr|int ch
952=for apidoc Amh|bool|isPRINT_LC_utf8_safe|U8 * s| U8 *end
953
243effed
KW
954=for apidoc Am|bool|isWORDCHAR|char ch
955Returns a boolean indicating whether the specified character is a character
956that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
957in a regular expression. A word character is an alphabetic character, a
958decimal digit, a connecting punctuation character (such as an underscore), or
959a "mark" character that attaches to one of those (like some sort of accent).
960C<isALNUM()> is a synonym provided for backward compatibility, even though a
961word character includes more than the standard C language meaning of
962alphanumeric.
dcccc8ff 963See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
964variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
965C<isWORDCHAR_utf8>, and C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>,
966C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and C<isWORDCHAR_LC_utf8_safe>
967are also as described there, but additionally include the platform's native
968underscore.
8a58bdcf 969
f16858ed
KW
970=for apidoc Amh|bool|isWORDCHAR_A|int ch
971=for apidoc Amh|bool|isWORDCHAR_L1|int ch
972=for apidoc Amh|bool|isWORDCHAR_uvchr|int ch
973=for apidoc Amh|bool|isWORDCHAR_utf8_safe|U8 * s|U8 * end
d23c7e08 974=for apidoc Amh|bool|isWORDCHAR_utf8|U8 * s|U8 * end
f16858ed
KW
975=for apidoc Amh|bool|isWORDCHAR_LC|int ch
976=for apidoc Amh|bool|isWORDCHAR_LC_uvchr|int ch
977=for apidoc Amh|bool|isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
978=for apidoc Amh|bool|isALNUM|int ch
979=for apidoc Amh|bool|isALNUM_A|int ch
980=for apidoc Amh|bool|isALNUM_LC|int ch
981=for apidoc Amh|bool|isALNUM_LC_uvchr|int ch
982
8a58bdcf
KW
983=for apidoc Am|bool|isXDIGIT|char ch
984Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
985digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
986and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
987See the L<top of this section|/Character classification> for an explanation of
988variants
059703b0
KW
989C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>,
990C<isXDIGIT_LC_uvchr>, C<isXDIGIT_LC_utf8>, and C<isXDIGIT_LC_utf8_safe>.
243effed 991
f16858ed
KW
992=for apidoc Amh|bool|isXDIGIT_A|int ch
993=for apidoc Amh|bool|isXDIGIT_L1|int ch
994=for apidoc Amh|bool|isXDIGIT_uvchr|int ch
995=for apidoc Amh|bool|isXDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 996=for apidoc Amh|bool|isXDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
997=for apidoc Amh|bool|isXDIGIT_LC|int ch
998=for apidoc Amh|bool|isXDIGIT_LC_uvchr|int ch
999=for apidoc Amh|bool|isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1000
3c3ecf18
KW
1001=for apidoc Am|bool|isIDFIRST|char ch
1002Returns a boolean indicating whether the specified character can be the first
1003character of an identifier. This is very close to, but not quite the same as
1004the official Unicode property C<XID_Start>. The difference is that this
1005returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
1006See the L<top of this section|/Character classification> for an explanation of
1007variants
059703b0
KW
1008C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
1009C<isIDFIRST_utf8_safe>, C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>,
1010C<isIDFIRST_LC_utf8>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18 1011
f16858ed
KW
1012=for apidoc Amh|bool|isIDFIRST_A|int ch
1013=for apidoc Amh|bool|isIDFIRST_L1|int ch
1014=for apidoc Amh|bool|isIDFIRST_uvchr|int ch
1015=for apidoc Amh|bool|isIDFIRST_utf8_safe|U8 * s|U8 * end
d23c7e08 1016=for apidoc Amh|bool|isIDFIRST_utf8|U8 * s|U8 * end
f16858ed
KW
1017=for apidoc Amh|bool|isIDFIRST_LC|int ch
1018=for apidoc Amh|bool|isIDFIRST_LC_uvchr|int ch
1019=for apidoc Amh|bool|isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1020
3c3ecf18
KW
1021=for apidoc Am|bool|isIDCONT|char ch
1022Returns a boolean indicating whether the specified character can be the
1023second or succeeding character of an identifier. This is very close to, but
1024not quite the same as the official Unicode property C<XID_Continue>. The
1025difference is that this returns true only if the input character also matches
dcccc8ff 1026L</isWORDCHAR>. See the L<top of this section|/Character classification> for
059703b0
KW
1027an explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
1028C<isIDCONT_utf8>, C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>,
1029C<isIDCONT_LC_utf8>, and C<isIDCONT_LC_utf8_safe>.
3c3ecf18 1030
f16858ed
KW
1031=for apidoc Amh|bool|isIDCONT_A|int ch
1032=for apidoc Amh|bool|isIDCONT_L1|int ch
1033=for apidoc Amh|bool|isIDCONT_uvchr|int ch
1034=for apidoc Amh|bool|isIDCONT_utf8_safe|U8 * s|U8 * end
d23c7e08 1035=for apidoc Amh|bool|isIDCONT_utf8|U8 * s|U8 * end
f16858ed
KW
1036=for apidoc Amh|bool|isIDCONT_LC|int ch
1037=for apidoc Amh|bool|isIDCONT_LC_uvchr|int ch
1038=for apidoc Amh|bool|isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1039
243effed 1040=head1 Miscellaneous Functions
8eea39dd 1041
95a59cab 1042=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 1043Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
1044Behaviour is only well defined when isXDIGIT(*str) is true.
1045
e7c1e6c1 1046=head1 Character case changing
21da7284
KW
1047Perl uses "full" Unicode case mappings. This means that converting a single
1048character to another case may result in a sequence of more than one character.
1049For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1050character sequence C<SS>. This presents some complications The lowercase of
1051all characters in the range 0..255 is a single character, and thus
1052C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1053return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1054an API that does allow every possible legal result to be returned.) Likewise
1055no other function that is crippled by not being able to give the correct
1056results for the full range of possible inputs has been implemented here.
e7c1e6c1 1057
d713f9d9 1058=for apidoc Am|U8|toUPPER|int ch
1f607577
KW
1059Converts the specified character to uppercase. If the input is anything but an
1060ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 1061C<toUPPER_A> is equivalent.
954c1994 1062
d0da05db
KW
1063=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1064Converts the code point C<cp> to its uppercase version, and
1065stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1066point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1067that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1068bytes since the uppercase version may be longer than the original character.
1069
1070The first code point of the uppercased version is returned
21da7284
KW
1071(but note, as explained at L<the top of this section|/Character case
1072changing>, that there may be more.)
1f607577 1073
059703b0 1074=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1075Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1076extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
1077stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1078that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1079bytes since the uppercase version may be longer than the original character.
1080
1081The first code point of the uppercased version is returned
21da7284
KW
1082(but note, as explained at L<the top of this section|/Character case
1083changing>, that there may be more).
1f607577 1084
059703b0
KW
1085It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1086S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1087the UTF-8 for the input character is malformed in some way, the program may
1088croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1089of the implementation, and subject to change in future releases.
a239b1e2 1090
059703b0
KW
1091=for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1092Same as L</toUPPER_utf8>.
1f607577 1093
25200305
KW
1094=for apidoc Am|U8|toFOLD|U8 ch
1095Converts the specified character to foldcase. If the input is anything but an
1096ASCII uppercase character, that input character itself is returned. Variant
1097C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 1098Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 1099
d0da05db
KW
1100=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1101Converts the code point C<cp> to its foldcase version, and
1102stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1103point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1104that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1105bytes since the foldcase version may be longer than the original character.
1106
1107The first code point of the foldcased version is returned
21da7284
KW
1108(but note, as explained at L<the top of this section|/Character case
1109changing>, that there may be more).
1f607577 1110
059703b0 1111=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1112Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1113extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
1114stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1115that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1116bytes since the foldcase version may be longer than the original character.
1117
1118The first code point of the foldcased version is returned
21da7284
KW
1119(but note, as explained at L<the top of this section|/Character case
1120changing>, that there may be more).
1f607577 1121
059703b0 1122It will not attempt
a239b1e2
KW
1123to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1124true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1125input character is malformed in some way, the program may croak, or the
1126function may return the REPLACEMENT CHARACTER, at the discretion of the
1127implementation, and subject to change in future releases.
1128
059703b0
KW
1129=for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1130Same as L</toFOLD_utf8>.
1f607577
KW
1131
1132=for apidoc Am|U8|toLOWER|U8 ch
1133Converts the specified character to lowercase. If the input is anything but an
1134ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 1135C<toLOWER_A> is equivalent.
954c1994 1136
1f607577 1137=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
1138Converts the specified Latin1 character to lowercase. The results are
1139undefined if the input doesn't fit in a byte.
1f607577
KW
1140
1141=for apidoc Am|U8|toLOWER_LC|U8 ch
1142Converts the specified character to lowercase using the current locale's rules,
1143if possible; otherwise returns the input character itself.
1144
d0da05db
KW
1145=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1146Converts the code point C<cp> to its lowercase version, and
1147stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1148point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1149that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1150bytes since the lowercase version may be longer than the original character.
1151
1152The first code point of the lowercased version is returned
21da7284
KW
1153(but note, as explained at L<the top of this section|/Character case
1154changing>, that there may be more).
1f607577 1155
bd350c85 1156=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1157Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1158extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
1159stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1160that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1161bytes since the lowercase version may be longer than the original character.
1162
1163The first code point of the lowercased version is returned
21da7284
KW
1164(but note, as explained at L<the top of this section|/Character case
1165changing>, that there may be more).
059703b0
KW
1166It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1167S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1168the UTF-8 for the input character is malformed in some way, the program may
1169croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1170of the implementation, and subject to change in future releases.
1f607577 1171
059703b0
KW
1172=for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1173Same as L</toLOWER_utf8>.
1f607577 1174
25200305
KW
1175=for apidoc Am|U8|toTITLE|U8 ch
1176Converts the specified character to titlecase. If the input is anything but an
1177ASCII lowercase character, that input character itself is returned. Variant
b7d90381 1178C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 1179range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 1180not a concept used in locale handling, so there is no functionality for that.)
25200305 1181
d0da05db
KW
1182=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1183Converts the code point C<cp> to its titlecase version, and
1184stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1185point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1186that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1187bytes since the titlecase version may be longer than the original character.
1188
1189The first code point of the titlecased version is returned
21da7284
KW
1190(but note, as explained at L<the top of this section|/Character case
1191changing>, that there may be more).
1f607577 1192
059703b0 1193=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1194Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1195extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1196stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1197that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1198bytes since the titlecase version may be longer than the original character.
1199
1200The first code point of the titlecased version is returned
21da7284
KW
1201(but note, as explained at L<the top of this section|/Character case
1202changing>, that there may be more).
1f607577 1203
059703b0 1204It will not attempt
a239b1e2
KW
1205to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1206true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1207input character is malformed in some way, the program may croak, or the
1208function may return the REPLACEMENT CHARACTER, at the discretion of the
1209implementation, and subject to change in future releases.
1210
059703b0
KW
1211=for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1212Same as L</toTITLE_utf8>.
1f607577 1213
954c1994 1214=cut
353c9b6f 1215
d0da05db 1216XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1217really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1218and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1219
8a58bdcf 1220Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1221patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1222
954c1994
GS
1223*/
1224
8f5283f4
KW
1225/*
1226 void below because that's the best fit, and works for Devel::PPPort
1227=for apidoc AmnU|void|WIDEST_UTYPE
1228
1229Yields the widest unsigned integer type on the platform, currently either
1230C<U32> or C<64>. This can be used in declarations such as
1231
1232 WIDEST_UTYPE my_uv;
1233
1234or casts
1235
1236 my_uv = (WIDEST_UTYPE) val;
1237
1238=cut
1239
1240*/
de9e2639
KW
1241#ifdef QUADKIND
1242# define WIDEST_UTYPE U64
7c062697
KW
1243#else
1244# define WIDEST_UTYPE U32
1245#endif
1246
3912bc88
KW
1247/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1248 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1249 * bits of information are lost even on a 64-bit machine, but to get the
1250 * compiler to optimize it out if possible. This is because Configure makes
1251 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1252 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1253 *
1254 * For Coverity, be always true, because otherwise Coverity thinks
1255 * it finds several expressions that are always true, independent
1256 * of operands. Well, they are, but that is kind of the point.
220c71bf 1257 */
7e75d1a1 1258#ifndef __COVERITY__
6c5b02ac
KW
1259 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1260 * pointer) */
1261#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1262 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1263#else
1264#define FITS_IN_8_BITS(c) (1)
1265#endif
cf301eb7 1266
45f4bb73 1267/* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
833b0f46 1268 * Written this way so that after optimization, only one conditional test is
76d3ad4c
KW
1269 * needed. (The NV casts stop any warnings about comparison always being true
1270 * if called with an unsigned. The cast preserves the sign, which is all we
1271 * care about.) */
1272#define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1273 __ASSERT_((NV) (n) >= 0) \
1274 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
833b0f46 1275
94250c4f
KW
1276/* Returns true if c is in the range l..u, where 'l' is non-negative
1277 * Written this way so that after optimization, only one conditional test is
4758c20d 1278 * needed. */
1eaefa6e 1279#define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
4758c20d 1280 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
4758c20d
KW
1281 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1282 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
45f4bb73 1283 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
305fe86e 1284
41f43cc2 1285#ifdef EBCDIC
b6340bd0 1286# ifndef _ALL_SOURCE
0852beac
KW
1287 /* The native libc isascii() et.al. functions return the wrong results
1288 * on at least z/OS unless this is defined. */
b6340bd0
KW
1289# error _ALL_SOURCE should probably be defined
1290# endif
41f43cc2 1291#else
0852beac
KW
1292 /* There is a simple definition of ASCII for ASCII platforms. But the
1293 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1294 * other macros below.
3f3c579d
KW
1295 *
1296 * The cast here is used instead of '(c) >= 0', because some compilers emit
1297 * a warning that that test is always true when the parameter is an
1298 * unsigned type. khw supposes that it could be written as
1299 * && ((c) == '\0' || (c) > 0)
1300 * to avoid the message, but the cast will likely avoid extra branches even
1301 * with stupid compilers.
1302 *
1303 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1304 * a pointer) */
9c903d59 1305# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1306#endif
1307
38694112
KW
1308/* Take the eight possible bit patterns of the lower 3 bits and you get the
1309 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1310 * can be ignored. If the rest match '0', we have an octal */
1311#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1312
9fb1bf9d 1313#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1314
a500dc72
KW
1315/* Character class numbers. For internal core Perl use only. The ones less
1316 * than 32 are used in PL_charclass[] and the ones up through the one that
1317 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1318 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1319 * their actual definitions are here. If that file has a name not used here,
1320 * it won't compile.
1709d539
KW
1321 *
1322 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561 1323 * frequency of their use. This gives a slight edge to exiting a loop earlier
58a3ba2c
KW
1324 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1325 * real Posix character class, and some (small) inefficiencies in regular
1326 * expression handling would be introduced by putting it in the middle of those
1327 * that are. Also, cntrl and ascii come after the others as it may be useful
1328 * to group these which have no members that match above Latin1, (or above
1329 * ASCII in the latter case) */
1330
1709d539
KW
1331# define _CC_WORDCHAR 0 /* \w and [:word:] */
1332# define _CC_DIGIT 1 /* \d and [:digit:] */
1333# define _CC_ALPHA 2 /* [:alpha:] */
1334# define _CC_LOWER 3 /* [:lower:] */
1335# define _CC_UPPER 4 /* [:upper:] */
1336# define _CC_PUNCT 5 /* [:punct:] */
1337# define _CC_PRINT 6 /* [:print:] */
15861f94 1338# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1339# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1340# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
779cf272 1341# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1342# define _CC_BLANK 11 /* [:blank:] */
1343# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1344# define _CC_CNTRL 13 /* [:cntrl:] */
1345# define _CC_ASCII 14 /* [:ascii:] */
1346# define _CC_VERTSPACE 15 /* \v */
1709d539 1347
a0947d7b
KW
1348# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1349
1709d539 1350/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1351 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1352# define _CC_IDFIRST 16
1353# define _CC_CHARNAME_CONT 17
1354# define _CC_NONLATIN1_FOLD 18
1355# define _CC_NONLATIN1_SIMPLE_FOLD 19
1356# define _CC_QUOTEMETA 20
1357# define _CC_NON_FINAL_FOLD 21
1358# define _CC_IS_IN_SOME_FOLD 22
1359# define _CC_MNEMONIC_CNTRL 23
073c22b3
KW
1360
1361/* This next group is only used on EBCDIC platforms, so theoretically could be
1362 * shared with something entirely different that's only on ASCII platforms */
abb8abf6
KW
1363# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
1364/* Unused: 24-30
f4cdb42c
KW
1365 * If more bits are needed, one could add a second word for non-64bit
1366 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1367 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1368 * is used only for optimization (as of this writing), and differs in the
1369 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1370 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1371 * would just cause /i regexes which match them to run less efficiently.
1372 * Similarly the EBCDIC-only bits are used just for speed, and could be
1373 * replaced by other means */
96ac0975 1374
3a371f2f
KW
1375#if defined(PERL_CORE) || defined(PERL_EXT)
1376/* An enum version of the character class numbers, to help compilers
1377 * optimize */
1378typedef enum {
3a371f2f 1379 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1380 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1381 _CC_ENUM_ASCII = _CC_ASCII,
1382 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1383 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1384 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1385 _CC_ENUM_DIGIT = _CC_DIGIT,
1386 _CC_ENUM_GRAPH = _CC_GRAPH,
1387 _CC_ENUM_LOWER = _CC_LOWER,
1388 _CC_ENUM_PRINT = _CC_PRINT,
1389 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1390 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1391 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1392 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1393 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1394 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1395} _char_class_number;
1396#endif
1397
86f72d56 1398#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1399
6635f04f 1400START_EXTERN_C
96ac0975
NC
1401# ifdef DOINIT
1402EXTCONST U32 PL_charclass[] = {
1403# include "l1_char_class_tab.h"
1404};
1405
1406# else /* ! DOINIT */
1407EXTCONST U32 PL_charclass[];
1408# endif
6635f04f 1409END_EXTERN_C
96ac0975 1410
265c1f46 1411 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1412# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1413
1414 /* For internal core Perl use only: the base macro for defining macros like
1415 * isALPHA */
ff7ecfc3 1416# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1417 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1418
f4cdb42c
KW
1419 /* The mask for the _A versions of the macros; it just adds in the bit for
1420 * ASCII. */
1421# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1422
4650c663
KW
1423 /* For internal core Perl use only: the base macro for defining macros like
1424 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1425 * the ASCII bit are present */
b7d90381
KW
1426# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1427 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1428 == _CC_mask_A(classnum)))
f4cdb42c 1429
26c1d9d8
KW
1430/* On ASCII platforms certain classes form a single range. It's faster to
1431 * special case these. isDIGIT is a single range on all platforms */
b877c1ff
KW
1432# ifdef EBCDIC
1433# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1434# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1435# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1436# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1437# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1438# else
26c1d9d8 1439 /* By folding the upper and lowercase, we can use a single range */
b877c1ff 1440# define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
26c1d9d8 1441# define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
b877c1ff
KW
1442# define isLOWER_A(c) inRANGE(c, 'a', 'z')
1443# define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1444# define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1445# endif
15861f94 1446# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1447# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1448# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
b877c1ff 1449# define isDIGIT_A(c) inRANGE(c, '0', '9')
f4cdb42c
KW
1450# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1451# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
f4cdb42c 1452# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1453# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1454 */
d95f8b6a 1455# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1456# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1457# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1458# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1459
1460 /* continuation character for legal NAME in \N{NAME} */
1461# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1462
1463# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1464# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1465# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1466# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1467# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1468# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1469# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1470# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1471# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1472# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1473
0852beac
KW
1474# ifdef EBCDIC
1475# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1476# endif
1477
f12c0118
KW
1478 /* Participates in a single-character fold with a character above 255 */
1479# 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)))
1480
1481 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1482# 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)))
430b7c70 1483
4eeeb416 1484# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1485# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1486 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1487# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1488 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1489# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1490 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1491#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1492
1493 /* If we don't have perl.h, we are compiling a utility program. Below we
1494 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1495 * to it. Most are coded based on first principles. These are written to
74665a89 1496 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
182c4ace 1497# define isDIGIT_A(c) inRANGE(c, '0', '9')
0852beac 1498# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1499# define isSPACE_A(c) (isBLANK_A(c) \
1500 || (c) == '\n' \
1501 || (c) == '\r' \
1502 || (c) == '\v' \
0852beac 1503 || (c) == '\f')
74665a89
KW
1504 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1505 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1506 * performance (if optimization isn't on), and allow the same code to be
1507 * used for both platform types */
182c4ace
KW
1508# define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1509 || inRANGE((c), 'j', 'r') \
1510 || inRANGE((c), 's', 'z')
1511# define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1512 || inRANGE((c), 'J', 'R') \
1513 || inRANGE((c), 'S', 'Z')
a4d7a999
KW
1514# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1515# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1516# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1517# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
182c4ace
KW
1518# define isXDIGIT_A(c) ( isDIGIT_A(c) \
1519 || inRANGE((c), 'a', 'f') \
1520 || inRANGE((c), 'A', 'F')
74665a89
KW
1521# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1522 || (c) == '#' || (c) == '$' || (c) == '%' \
1523 || (c) == '&' || (c) == '\'' || (c) == '(' \
1524 || (c) == ')' || (c) == '*' || (c) == '+' \
1525 || (c) == ',' || (c) == '.' || (c) == '/' \
1526 || (c) == ':' || (c) == ';' || (c) == '<' \
1527 || (c) == '=' || (c) == '>' || (c) == '?' \
1528 || (c) == '@' || (c) == '[' || (c) == '\\' \
1529 || (c) == ']' || (c) == '^' || (c) == '_' \
1530 || (c) == '`' || (c) == '{' || (c) == '|' \
1531 || (c) == '}' || (c) == '~')
1532# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1533# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1534
0852beac 1535# ifdef EBCDIC
74665a89
KW
1536 /* The below is accurate for the 3 EBCDIC code pages traditionally
1537 * supported by perl. The only difference between them in the controls
1538 * is the position of \n, and that is represented symbolically below */
1539# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1540 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1541 || (c) == '\t' || (c) == '\v' \
182c4ace 1542 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
8ec0a736 1543 || (c) == 7F /* U+7F DEL */ \
182c4ace
KW
1544 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1545 DC[1-3] */ \
74665a89
KW
1546 || (c) == 0x18 /* U+18 CAN */ \
1547 || (c) == 0x19 /* U+19 EOM */ \
182c4ace 1548 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
74665a89
KW
1549 || (c) == 0x26 /* U+17 ETB */ \
1550 || (c) == 0x27 /* U+1B ESC */ \
1551 || (c) == 0x2D /* U+05 ENQ */ \
1552 || (c) == 0x2E /* U+06 ACK */ \
1553 || (c) == 0x32 /* U+16 SYN */ \
1554 || (c) == 0x37 /* U+04 EOT */ \
1555 || (c) == 0x3C /* U+14 DC4 */ \
1556 || (c) == 0x3D /* U+15 NAK */ \
1557 || (c) == 0x3F)/* U+1A SUB */
0852beac 1558# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1559# else /* isASCII is already defined for ASCII platforms, so can use that to
1560 define isCNTRL */
1561# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1562# endif
1563
3ffc8c70 1564 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1565 * during debugging, and it seems best to keep them. We may be called
1566 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1567 * do anything anyway, so make it not a problem */
1568# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1569# define NATIVE_TO_LATIN1(ch) (ch)
1570# endif
3ded5eb0
KW
1571# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1572# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1573# define isBLANK_L1(c) (isBLANK_A(c) \
1574 || (FITS_IN_8_BITS(c) \
1575 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1576# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1577# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1578# define isLOWER_L1(c) (isLOWER_A(c) \
1579 || (FITS_IN_8_BITS(c) \
ae683a5f 1580 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1581 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1582 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1583 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1584 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1585# define isPRINT_L1(c) (isPRINT_A(c) \
1586 || (FITS_IN_8_BITS(c) \
1587 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1588# define isPUNCT_L1(c) (isPUNCT_A(c) \
1589 || (FITS_IN_8_BITS(c) \
ae683a5f 1590 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1591 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1592 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1593 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1594 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1595 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1596 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1597# define isSPACE_L1(c) (isSPACE_A(c) \
1598 || (FITS_IN_8_BITS(c) \
ae683a5f 1599 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1600 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1601# define isUPPER_L1(c) (isUPPER_A(c) \
1602 || (FITS_IN_8_BITS(c) \
182c4ace
KW
1603 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1604 0xC0, 0xDE) \
3ded5eb0
KW
1605 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1606# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1607# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1608# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1609 || isBLANK_L1(c) \
1610 || (c) == '-' \
1611 || (c) == '(' \
1612 || (c) == ')')
1613 /* The following are not fully accurate in the above-ASCII range. I (khw)
1614 * don't think it's necessary to be so for the purposes where this gets
1615 * compiled */
1616# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1617# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1618
1619 /* And these aren't accurate at all. They are useful only for above
1620 * Latin1, which utilities and bootstrapping don't deal with */
1621# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1622# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1623# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1624
1625 /* Many of the macros later in this file are defined in terms of these. By
1626 * implementing them with a function, which converts the class number into
1627 * a call to the desired macro, all of the later ones work. However, that
1628 * function won't be actually defined when building a utility program (no
1629 * perl.h), and so a compiler error will be generated if one is attempted
1630 * to be used. And the above-Latin1 code points require Unicode tables to
1631 * be present, something unlikely to be the case when bootstrapping */
1632# define _generic_isCC(c, classnum) \
1633 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1634# define _generic_isCC_A(c, classnum) \
1635 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1636#endif /* End of no perl.h H_PERL */
8a58bdcf 1637
e66b99e9
KW
1638#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1639#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1640#define isASCII_A(c) isASCII(c)
1641#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1642#define isBLANK(c) isBLANK_A(c)
1643#define isCNTRL(c) isCNTRL_A(c)
1644#define isDIGIT(c) isDIGIT_A(c)
1645#define isGRAPH(c) isGRAPH_A(c)
1646#define isIDFIRST(c) isIDFIRST_A(c)
1647#define isLOWER(c) isLOWER_A(c)
1648#define isPRINT(c) isPRINT_A(c)
779cf272 1649#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1650#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1651#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1652#define isPUNCT(c) isPUNCT_A(c)
1653#define isSPACE(c) isSPACE_A(c)
1654#define isUPPER(c) isUPPER_A(c)
1655#define isWORDCHAR(c) isWORDCHAR_A(c)
1656#define isXDIGIT(c) isXDIGIT_A(c)
1657
1658/* ASCII casing. These could also be written as
1659 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1660 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1661 which uses table lookup and mask instead of subtraction. (This would
c5e9991e
KW
1662 work because the _MOD does not apply in the ASCII range).
1663
1664 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1665 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1666 there non-ASCII invariants, and all of them are controls.) */
68067e4e
DM
1667#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1668#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1669
25200305
KW
1670/* In the ASCII range, these are equivalent to what they're here defined to be.
1671 * But by creating these definitions, other code doesn't have to be aware of
c5e9991e
KW
1672 * this detail. Actually this works for all UTF-8 invariants, not just the
1673 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
25200305 1674#define toFOLD(c) toLOWER(c)
25200305
KW
1675#define toTITLE(c) toUPPER(c)
1676
c753c8d3
KW
1677#define toLOWER_A(c) toLOWER(c)
1678#define toUPPER_A(c) toUPPER(c)
25200305
KW
1679#define toFOLD_A(c) toFOLD(c)
1680#define toTITLE_A(c) toTITLE(c)
1a0901db 1681
4650c663 1682/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1683#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1684 ? (c) \
f4cd282c 1685 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1686#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1687
1a0901db 1688/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1689 * all mapped to one of them, and these need special handling; returns the
1690 * input itself if is out-of-range */
b2bf251f 1691#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1692 ? (c) \
f4cd282c 1693 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1694#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1695
beab9ebe
KW
1696/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1697
1698/* For internal core Perl use only: the base macro for defining macros like
1699 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1700 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1701 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1702 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1703 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1704 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1705 * code points under locale rules, use the macros based on _generic_LC_uvchr
1706 * instead of this. */
beab9ebe
KW
1707#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1708 (! FITS_IN_8_BITS(c) \
1709 ? 0 \
31f05a37
KW
1710 : IN_UTF8_CTYPE_LOCALE \
1711 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1712 : cBOOL(non_utf8))
1713
1714/* For internal core Perl use only: a helper macro for defining macros like
1715 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1716 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1717 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1718 * _generic_LC_base for more info */
1719#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1720 _generic_LC_base(c,utf8_locale_classnum, \
1721 non_utf8_func( (_LC_CAST) (c)))
1722
1723/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1724 * 'c' is the platform's native underscore character */
1725#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1726 _generic_LC_base(c, utf8_locale_classnum, \
1727 (non_utf8_func( (_LC_CAST) (c)) \
1728 || (char)(c) == '_'))
1729
1730/* These next three are also for internal core Perl use only: case-change
247985d4
KW
1731 * helper macros. The reason for using the PL_latin arrays is in case the
1732 * system function is defective; it ensures uniform results that conform to the
b257a28c 1733 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
beab9ebe
KW
1734#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1735 ? (c) \
31f05a37
KW
1736 : (IN_UTF8_CTYPE_LOCALE) \
1737 ? PL_latin1_lc[ (U8) (c) ] \
5a10328c 1738 : (cast)function((cast)(c)))
beab9ebe 1739
31f05a37
KW
1740/* Note that the result can be larger than a byte in a UTF-8 locale. It
1741 * returns a single value, so can't adequately return the upper case of LATIN
1742 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1743 * values "SS"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1744 * otherwise returns its input. It does not handle the anomalies in UTF-8
1745 * Turkic locales. */
beab9ebe
KW
1746#define _generic_toUPPER_LC(c, function, cast) \
1747 (! FITS_IN_8_BITS(c) \
1748 ? (c) \
31f05a37 1749 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1750 ? (cast)function((cast)(c)) \
31f05a37
KW
1751 : ((((U8)(c)) == MICRO_SIGN) \
1752 ? GREEK_CAPITAL_LETTER_MU \
1753 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1754 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1755 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1756 ? (__ASSERT_(0) (c)) \
1757 : PL_mod_latin1_uc[ (U8) (c) ])))))
1758
1759/* Note that the result can be larger than a byte in a UTF-8 locale. It
1760 * returns a single value, so can't adequately return the fold case of LATIN
1761 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1762 * values "ss"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1763 * otherwise returns its input. It does not handle the anomalies in UTF-8
1764 * Turkic locales */
beab9ebe 1765#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1766 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1767 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1768 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1769 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1770 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1771
84061b6a 1772/* Use the libc versions for these if available. */
f05550c0 1773#if defined(HAS_ISASCII)
84061b6a
KW
1774# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1775#else
1776# define isASCII_LC(c) isASCII(c)
1777#endif
1778
f05550c0 1779#if defined(HAS_ISBLANK)
84061b6a 1780# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1781#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1782# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1783#endif
1784
f05550c0 1785#define _LC_CAST U8
bbce6d69 1786
f05550c0 1787#ifdef WIN32
375f5f06
KW
1788 /* The Windows functions don't bother to follow the POSIX standard, which
1789 * for example says that something can't both be a printable and a control.
1790 * But Windows treats the \t control as a printable, and does such things
1791 * as making superscripts into both digits and punctuation. This tames
1792 * these flaws by assuming that the definitions of both controls and space
1793 * are correct, and then making sure that other definitions don't have
1794 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1795 * Not all possible weirdnesses are checked for, just the ones that were
1796 * detected on actual Microsoft code pages */
1797
b7d90381
KW
1798# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1799# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1800
1801# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1802 && isALPHANUMERIC_LC(c))
1803# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1804 ! isPUNCT_LC(c))
1805# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1806 isALPHANUMERIC_LC(c))
1807# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1808# define isIDFIRST_LC(c) (((c) == '_') \
1809 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1810# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1811# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1812# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1813# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1814# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1815# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1816 && isALPHANUMERIC_LC(c))
f05550c0
BF
1817
1818# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1819# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1820# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1821
1822#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1823 /* For most other platforms */
beab9ebe 1824
f05550c0
BF
1825# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1826# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1827# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1828# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1829# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1830# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1831# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1832# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1833# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1834# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1835# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1836# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1837# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1838
1839
1840# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1841# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1842# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1843
1844#else /* The final fallback position */
1845
b7d90381
KW
1846# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1847# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1848# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1849# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1850# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1851# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1852# define isLOWER_LC(c) (isascii(c) && islower(c))
1853# define isPRINT_LC(c) (isascii(c) && isprint(c))
1854# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1855# define isSPACE_LC(c) (isascii(c) && isspace(c))
1856# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1857# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1858# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1859
1860# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1861# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1862# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1863
f05550c0 1864#endif
55204971 1865
eba68aa0
KW
1866#define isIDCONT(c) isWORDCHAR(c)
1867#define isIDCONT_A(c) isWORDCHAR_A(c)
1868#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1869#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1870#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1871
4650c663 1872/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1873 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1874 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1875 * classes where there is a macro or function 'above_latin1' that takes the
1876 * single argument 'c' and returns the desired value. These exist for those
2366ba44
KW
1877 * classes which have simple definitions, avoiding the overhead of an inversion
1878 * list binary search. _generic_invlist_uvchr() can be used
4650c663 1879 * for classes where that overhead is faster than a direct lookup.
d0da05db 1880 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1881 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1882 * have duplicate checks here, so could create versions of the macros that
1883 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1884
1885/* Note that all ignore 'use bytes' */
1e222e4f
KW
1886#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1887 ? _generic_isCC(c, classnum) \
cd500f2f 1888 : above_latin1(c))
2366ba44 1889#define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
1e222e4f 1890 ? _generic_isCC(c, classnum) \
922e8cb4 1891 : _is_uni_FOO(classnum, c))
2366ba44
KW
1892#define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
1893#define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
d0da05db
KW
1894#define isASCII_uvchr(c) isASCII(c)
1895#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1896#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2366ba44
KW
1897#define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
1898#define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1899#define isIDCONT_uvchr(c) \
1900 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1901#define isIDFIRST_uvchr(c) \
1902 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2366ba44
KW
1903#define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
1904#define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
d0da05db 1905
2366ba44 1906#define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
d0da05db
KW
1907#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1908#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1909
2366ba44 1910#define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
d0da05db 1911#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2366ba44 1912#define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
d0da05db
KW
1913#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1914
1915#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1916#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1917#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1918#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1919
1920/* For backwards compatibility, even though '_uni' should mean official Unicode
1921 * code points, in Perl it means native for those below 256 */
1922#define isALPHA_uni(c) isALPHA_uvchr(c)
1923#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1924#define isASCII_uni(c) isASCII_uvchr(c)
1925#define isBLANK_uni(c) isBLANK_uvchr(c)
1926#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1927#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1928#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1929#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1930#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1931#define isLOWER_uni(c) isLOWER_uvchr(c)
1932#define isPRINT_uni(c) isPRINT_uvchr(c)
1933#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1934#define isSPACE_uni(c) isSPACE_uvchr(c)
1935#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1936#define isUPPER_uni(c) isUPPER_uvchr(c)
1937#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1938#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1939#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1940#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1941#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1942#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1943#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1944
4650c663
KW
1945/* For internal core Perl use only: the base macros for defining macros like
1946 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1947 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1948 * simple class definitions; the other for more complex. These are like
d0da05db 1949 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1950#define _generic_LC_uvchr(latin1, above_latin1, c) \
1951 (c < 256 ? latin1(c) : above_latin1(c))
2366ba44 1952#define _generic_LC_invlist_uvchr(latin1, classnum, c) \
cd500f2f
KW
1953 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1954
2366ba44
KW
1955#define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
1956#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
cd500f2f 1957 _CC_ALPHANUMERIC, c)
b7d90381
KW
1958#define isASCII_LC_uvchr(c) isASCII_LC(c)
1959#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1960 is_HORIZWS_cp_high, c)
feeab5a9 1961#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2366ba44
KW
1962#define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1963#define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1964#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1965 _is_uni_perl_idcont, c)
b7d90381 1966#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f 1967 _is_uni_perl_idstart, c)
2366ba44
KW
1968#define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
1969#define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1970#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2366ba44 1971#define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1972#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1973 is_XPERLSPACE_cp_high, c)
2366ba44
KW
1974#define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
1975#define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
cd500f2f 1976 _CC_WORDCHAR, c)
b7d90381
KW
1977#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1978 is_XDIGIT_cp_high, c)
e712593e 1979
b7d90381 1980#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1981
da8c1a98
KW
1982/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
1983 * they don't otherwise go out of their way to look for malformed UTF-8. If
1984 * they can return accurate results without knowing if the input is otherwise
1985 * malformed, they do so. For example isASCII is accurate in spite of any
1986 * non-length malformations because it looks only at a single byte. Likewise
1987 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
1988 * variant ones return FALSE. But, if the input has to be well-formed in order
1989 * for the results to be accurate, the macros will test and if malformed will
1990 * call a routine to die
1991 *
1992 * Except for toke.c, the macros do assume that e > p, asserting that on
1993 * DEBUGGING builds. Much code that calls these depends on this being true,
1994 * for other reasons. toke.c is treated specially as using the regular
1995 * assertion breaks it in many ways. All strings that these operate on there
1996 * are supposed to have an extra NUL character at the end, so that *e = \0. A
1997 * bunch of code in toke.c assumes that this is true, so the assertion allows
1998 * for that */
1999#ifdef PERL_IN_TOKE_C
2000# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2001#else
2002# define _utf8_safe_assert(p,e) ((e) > (p))
2003#endif
2004
2005#define _generic_utf8_safe(classnum, p, e, above_latin1) \
c81b3562
KW
2006 ((! _utf8_safe_assert(p, e)) \
2007 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2008 : (UTF8_IS_INVARIANT(*(p))) \
da8c1a98
KW
2009 ? _generic_isCC(*(p), classnum) \
2010 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2011 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2012 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2013 classnum) \
2014 : (_force_out_malformed_utf8_message( \
2015 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2016 : above_latin1))
b7d90381
KW
2017/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2018 * 'above_latin1' can be a macro */
da8c1a98
KW
2019#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2020 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2366ba44 2021#define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2022 _generic_utf8_safe(classnum, p, e, \
2023 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2024 ? (_force_out_malformed_utf8_message( \
2025 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2026 : above_latin1(p)))
2366ba44
KW
2027/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2028 * 'above_latin1' parameter */
2029#define _generic_invlist_utf8_safe(classnum, p, e) \
2030 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
922e8cb4 2031
cc8ab7c0 2032/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
2033 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2034 * class is TRUE for. Hence it can skip the tests for this range.
2035 * 'above_latin1' should include its arguments */
da8c1a98
KW
2036#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2037 (__ASSERT_(_utf8_safe_assert(p, e)) \
2038 (UTF8_IS_INVARIANT(*(p))) \
2039 ? _generic_isCC(*(p), classnum) \
2040 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2041 ? 0 /* Note that doesn't check validity for latin1 */ \
2042 : above_latin1)
2043
84238efa 2044
059703b0
KW
2045#define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2046#define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2047#define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2048#define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2049#define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2050#define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2051#define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2052#define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2053#define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2054#define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2055#define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2056#define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2057#define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2058#define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2059#define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2060#define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2061#define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2062#define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
e8fa43e2 2063
2366ba44 2064#define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
da8c1a98 2065#define isALPHANUMERIC_utf8_safe(p, e) \
2366ba44 2066 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
da8c1a98
KW
2067#define isASCII_utf8_safe(p, e) \
2068 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2069 * works */ \
2070 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2071#define isBLANK_utf8_safe(p, e) \
2366ba44 2072 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
da8c1a98 2073
e8fa43e2
KW
2074#ifdef EBCDIC
2075 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2076 * more efficient macro instead of the more general one */
da8c1a98 2077# define isCNTRL_utf8_safe(p, e) \
56d02b8c 2078 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 2079#else
da8c1a98 2080# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
2081#endif
2082
da8c1a98
KW
2083#define isDIGIT_utf8_safe(p, e) \
2084 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2366ba44
KW
2085 _is_utf8_FOO(_CC_DIGIT, p, e))
2086#define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
da8c1a98 2087#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
dd1a3ba7 2088 _is_utf8_perl_idcont, p, e)
e5dcd934 2089
c11ff943
KW
2090/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2091 * IDFIRST is an alnum. See
8034715d 2092 * https://github.com/Perl/perl5/issues/10275 for more detail than you
f91dcd13
KW
2093 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2094 * This used to be not the XID version, but we decided to go with the more
2095 * modern Unicode definition */
da8c1a98
KW
2096#define isIDFIRST_utf8_safe(p, e) \
2097 _generic_func_utf8_safe(_CC_IDFIRST, \
dd1a3ba7 2098 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
da8c1a98 2099
2366ba44
KW
2100#define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2101#define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
da8c1a98 2102#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2366ba44 2103#define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
da8c1a98 2104#define isSPACE_utf8_safe(p, e) \
2366ba44
KW
2105 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2106#define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
da8c1a98 2107#define isVERTWS_utf8_safe(p, e) \
2366ba44 2108 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
da8c1a98 2109#define isWORDCHAR_utf8_safe(p, e) \
2366ba44 2110 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
da8c1a98
KW
2111#define isXDIGIT_utf8_safe(p, e) \
2112 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2113 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2114 ? (_force_out_malformed_utf8_message( \
2115 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2116 : is_XDIGIT_high(p)))
a0ed51b3 2117
059703b0
KW
2118#define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2119#define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2120#define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2121#define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2e8adce6 2122
567b353c 2123/* For internal core use only, subject to change */
059703b0
KW
2124#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2125#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2126#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2127#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
a1a5ec35
KW
2128
2129#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2130#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2131#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2132#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 2133
059703b0
KW
2134#define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2135#define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2136#define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2137#define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2138#define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2139#define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2140#define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2141#define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2142#define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2143#define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2144#define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2145#define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2146#define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2147#define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2148#define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2149#define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2150#define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
34aeb2e9 2151
da8c1a98
KW
2152/* For internal core Perl use only: the base macros for defining macros like
2153 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2154 * point in 'p' is within the 0-255 range, it uses locale rules from the
2155 * passed-in 'macro' parameter */
2156#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2157 (__ASSERT_(_utf8_safe_assert(p, e)) \
2158 (UTF8_IS_INVARIANT(*(p))) \
2159 ? macro(*(p)) \
2160 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2161 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2162 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2163 : (_force_out_malformed_utf8_message( \
2164 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2165 : above_latin1))
2166
2366ba44
KW
2167#define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2168 _generic_LC_utf8_safe(macro, p, e, \
2169 _is_utf8_FOO(classnum, p, e))
da8c1a98
KW
2170
2171#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2172 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2173
2366ba44 2174#define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2175 _generic_LC_utf8_safe(classnum, p, e, \
2176 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2177 ? (_force_out_malformed_utf8_message( \
2178 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2179 : above_latin1(p)))
2180
2181#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2366ba44 2182 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
da8c1a98
KW
2183 _CC_ALPHANUMERIC, p, e)
2184#define isALPHA_LC_utf8_safe(p, e) \
2366ba44 2185 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
da8c1a98
KW
2186#define isASCII_LC_utf8_safe(p, e) \
2187 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2188#define isBLANK_LC_utf8_safe(p, e) \
2366ba44 2189 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
da8c1a98
KW
2190#define isCNTRL_LC_utf8_safe(p, e) \
2191 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2192#define isDIGIT_LC_utf8_safe(p, e) \
2366ba44 2193 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
da8c1a98 2194#define isGRAPH_LC_utf8_safe(p, e) \
2366ba44 2195 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
da8c1a98
KW
2196#define isIDCONT_LC_utf8_safe(p, e) \
2197 _generic_LC_func_utf8_safe(isIDCONT_LC, \
dd1a3ba7 2198 _is_utf8_perl_idcont, p, e)
da8c1a98
KW
2199#define isIDFIRST_LC_utf8_safe(p, e) \
2200 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
dd1a3ba7 2201 _is_utf8_perl_idstart, p, e)
da8c1a98 2202#define isLOWER_LC_utf8_safe(p, e) \
2366ba44 2203 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
da8c1a98 2204#define isPRINT_LC_utf8_safe(p, e) \
2366ba44 2205 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
da8c1a98
KW
2206#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2207#define isPUNCT_LC_utf8_safe(p, e) \
2366ba44 2208 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
da8c1a98 2209#define isSPACE_LC_utf8_safe(p, e) \
2366ba44 2210 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
da8c1a98 2211#define isUPPER_LC_utf8_safe(p, e) \
2366ba44 2212 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
da8c1a98 2213#define isWORDCHAR_LC_utf8_safe(p, e) \
2366ba44 2214 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
da8c1a98 2215#define isXDIGIT_LC_utf8_safe(p, e) \
2366ba44 2216 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2217
fbc19f27
KW
2218/* Macros for backwards compatibility and for completeness when the ASCII and
2219 * Latin1 values are identical */
b7d90381
KW
2220#define isALPHAU(c) isALPHA_L1(c)
2221#define isDIGIT_L1(c) isDIGIT_A(c)
2222#define isOCTAL(c) isOCTAL_A(c)
2223#define isOCTAL_L1(c) isOCTAL_A(c)
2224#define isXDIGIT_L1(c) isXDIGIT_A(c)
2225#define isALNUM(c) isWORDCHAR(c)
a377c856 2226#define isALNUM_A(c) isALNUM(c)
b7d90381
KW
2227#define isALNUMU(c) isWORDCHAR_L1(c)
2228#define isALNUM_LC(c) isWORDCHAR_LC(c)
2229#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2230#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
059703b0 2231#define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
4c1d9526 2232#define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
059703b0 2233#define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
4c1d9526 2234#define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
b7d90381
KW
2235#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2236#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2237#define isALNUMC(c) isALPHANUMERIC(c)
2238#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2239#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2240#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
059703b0 2241#define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
4c1d9526
KW
2242#define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2243#define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
fbc19f27 2244
2bd1cbf6
KW
2245/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2246 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2247 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2248 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2249 * the outlier from the block that contains the other controls, just like
2250 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2251 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2252 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2253 * etc. */
2bd1cbf6 2254#ifndef EBCDIC
75763b3a 2255# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2256#else
75763b3a
KW
2257# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2258 ((isPRINT_A(c)) \
2259 ? (UNLIKELY((c) == '?') \
2260 ? QUESTION_MARK_CTRL \
2261 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2262 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2263 ? '?' \
2264 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2265#endif
bbce6d69 2266
dea28490
JJ
2267/* Line numbers are unsigned, 32 bits. */
2268typedef U32 line_t;
e5dcd934 2269#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2270
91152fc1
DG
2271/* Helpful alias for version prescan */
2272#define is_LAX_VERSION(a,b) \
2273 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2274
2275#define is_STRICT_VERSION(a,b) \
2276 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2277
2278#define BADVERSION(a,b,c) \
2279 if (b) { \
2280 *b = c; \
2281 } \
2282 return a;
8c52afec 2283
cb27eebd
KW
2284/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2285 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2286 * which is advanced. The input is validated only by an assert() in DEBUGGING
2287 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2288 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2289#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2290 ? (c) \
2291 : ((c) + 9))))
902008b8
KW
2292#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2293 ? (*(s)++) \
2294 : (*(s)++ + 9))))
95a59cab 2295
cb27eebd
KW
2296/* Converts a character known to represent an octal digit (0-7) to its numeric
2297 * value. The input is validated only by an assert() in DEBUGGING builds. In
2298 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2299#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2300
305b8651
KW
2301/* Efficiently returns a boolean as to if two native characters are equivalent
2302 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2303 * the ALPHA in the name is to remind you of that. This is asserted() in
2304 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2305 * works (on valid input) for both non- and UTF-8-encoded bytes.
2306 *
2307 * When one of the inputs is a compile-time constant and gets folded by the
2308 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2309 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2310 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2311 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2312 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2313 * just a single 0, in the bit position where the upper- and lowercase differ.
2314 * */
305b8651
KW
2315#define isALPHA_FOLD_EQ(c1, c2) \
2316 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2317 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2318#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2319
8e84507e 2320/*
ccfc67b7
JH
2321=head1 Memory Management
2322
a02a5408 2323=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2324The XSUB-writer's interface to the C C<malloc> function.
2325
596f7718 2326Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2327
c5008215
JC
2328In 5.9.3, Newx() and friends replace the older New() API, and drops
2329the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2330themselves. This aid has been superseded by a new build option,
d10b4965 2331PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2332there for use in XS modules supporting older perls.
2333
a02a5408 2334=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2335The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2336cast. See also C<L</Newx>>.
954c1994 2337
596f7718 2338Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2339
a02a5408 2340=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2341The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2342memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2343
596f7718 2344Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2345
954c1994
GS
2346=for apidoc Am|void|Renew|void* ptr|int nitems|type
2347The XSUB-writer's interface to the C C<realloc> function.
2348
596f7718 2349Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2350
954c1994
GS
2351=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2352The XSUB-writer's interface to the C C<realloc> function, with
2353cast.
2354
596f7718 2355Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2356
49b8b560 2357=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2358The XSUB-writer's interface to the C C<free> function.
2359
596f7718 2360This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2361
954c1994
GS
2362=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2363The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2364source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2365C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2366
e90e2364 2367=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2368Like C<Move> but returns C<dest>. Useful
72d33970 2369for encouraging compilers to tail-call
e90e2364
NC
2370optimise.
2371
954c1994
GS
2372=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2373The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2374source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2375C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2376
e90e2364
NC
2377=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2378
796b6530 2379Like C<Copy> but returns C<dest>. Useful
72d33970 2380for encouraging compilers to tail-call
e90e2364
NC
2381optimise.
2382
954c1994
GS
2383=for apidoc Am|void|Zero|void* dest|int nitems|type
2384
2385The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2386destination, C<nitems> is the number of items, and C<type> is the type.
2387
e90e2364
NC
2388=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2389
72d33970
FC
2390Like C<Zero> but returns dest. Useful
2391for encouraging compilers to tail-call
e90e2364
NC
2392optimise.
2393
da5d8dbb 2394=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2395This is an architecture-independent macro to copy one structure to another.
954c1994 2396
7e337ee0
JH
2397=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2398
2399Fill up memory with a byte pattern (a byte repeated over and over
2400again) that hopefully catches attempts to access uninitialized memory.
2401
2402=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2403
2404PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2405
1c12ffb4 2406=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2407
2408PoisonWith(0xEF) for catching access to freed memory.
2409
9965345d
JH
2410=for apidoc Am|void|Poison|void* dest|int nitems|type
2411
7e337ee0 2412PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2413
2414=cut */
954c1994 2415
561b68a9
SH
2416/* Maintained for backwards-compatibility only. Use newSV() instead. */
2417#ifndef PERL_CORE
ff06c60c 2418#define NEWSV(x,len) newSV(len)
561b68a9 2419#endif
ff06c60c 2420
b7112dce 2421#define MEM_SIZE_MAX ((MEM_SIZE)-1)
19a94d75 2422
a500027b 2423#define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
e6bdf523 2424
27d5b266 2425#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2426
2427/* This expression will be constant-folded at compile time. It checks
2428 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2429 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2430 * As well as avoiding the need for a run-time check in some cases, it's
2431 * designed to avoid compiler warnings like:
2432 * comparison is always false due to limited range of data type
73e8ff00
DM
2433 * It's mathematically equivalent to
2434 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2435 */
2436
2437# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
445198b9
LM
2438 ( sizeof(MEM_SIZE) < sizeof(n) \
2439 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 2440
88f9f128 2441/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2442 * compiler warnings. We *want* to write the expression as
2443 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2444 * (for some compile-time constant C), but even when the LHS
2445 * constant-folds to false at compile-time, g++ insists on emitting
2446 * warnings about the RHS (e.g. "comparison is always false"), so instead
2447 * we write it as
e6bdf523 2448 *
d98e5cde 2449 * (cond ? n : X) > C
88f9f128 2450 *
d98e5cde
DM
2451 * where X is a constant with X > C always false. Choosing a value for X
2452 * is tricky. If 0, some compilers will complain about 0 > C always being
2453 * false; if 1, Coverity complains when n happens to be the constant value
2454 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2455 * for X and hope that nothing else whines.
e6bdf523
DM
2456 */
2457
2458# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2459 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2460 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2461
2462# define MEM_WRAP_CHECK(n,t) \
2463 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2464 && (croak_memory_wrap(),0))
2465
2466# define MEM_WRAP_CHECK_1(n,t,a) \
2467 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2468 && (Perl_croak_nocontext("%s",(a)),0))
2469
814eedc8
DD
2470/* "a" arg must be a string literal */
2471# define MEM_WRAP_CHECK_s(n,t,a) \
2472 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2473 && (Perl_croak_nocontext("" a ""),0))
2474
8b44ba4c 2475#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2476
a500027b 2477#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))
27d5b266
JH
2478#else
2479
410319be
NC
2480#define MEM_WRAP_CHECK(n,t)
2481#define MEM_WRAP_CHECK_1(n,t,a)
814eedc8 2482#define MEM_WRAP_CHECK_s(n,t,a)
8b44ba4c
NC
2483#define MEM_WRAP_CHECK_(n,t)
2484
a500027b 2485#define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
27d5b266 2486
1936d2a7 2487#endif
8b44ba4c 2488
fe4f188c 2489#ifdef PERL_MEM_LOG
46c6c7e2 2490/*
9f653bb5 2491 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2492 * go through functions, which are handy for debugging breakpoints, but
2493 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2494 * line number, and C function name if available) passed in. This info can
2495 * then be used for logging the calls, for which one gets a sample
73d1d973 2496 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2497 *
46c6c7e2 2498 * Known problems:
94e892a6 2499 * - not all memory allocs get logged, only those
46c6c7e2 2500 * that go through Newx() and derivatives (while all
94e892a6 2501 * Safefrees do get logged)
46c6c7e2
JH
2502 * - __FILE__ and __LINE__ do not work everywhere
2503 * - __func__ or __FUNCTION__ even less so
2504 * - I think more goes on after the perlio frees but
2505 * the thing is that STDERR gets closed (as do all
2506 * the file descriptors)
2507 * - no deeper calling stack than the caller of the Newx()
2508 * or the kind, but do I look like a C reflection/introspection
2509 * utility to you?
2510 * - the function prototypes for the logging functions
2511 * probably should maybe be somewhere else than handy.h
2512 * - one could consider inlining (macrofying) the logging
2513 * for speed, but I am too lazy
2514 * - one could imagine recording the allocations in a hash,
2515 * (keyed by the allocation address?), and maintain that
2516 * through reallocs and frees, but how to do that without
2517 * any News() happening...?
73d1d973 2518 * - lots of -Ddefines to get useful/controllable output
b953482e 2519 * - lots of ENV reads
46c6c7e2
JH
2520 */
2521
0b0ab801 2522# ifdef PERL_CORE
73d1d973 2523# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2524enum mem_log_type {
2525 MLT_ALLOC,
2526 MLT_REALLOC,
d7a2c63c
MHM
2527 MLT_FREE,
2528 MLT_NEW_SV,
2529 MLT_DEL_SV
0b0ab801
MHM
2530};
2531# endif
12754f92 2532# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2533void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2534void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2535# endif
0b0ab801
MHM
2536# endif
2537
fe4f188c
JH
2538#endif
2539
2540#ifdef PERL_MEM_LOG
d1401ee9
MHM
2541#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2542#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2543#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2544#endif
2545
2546#ifndef MEM_LOG_ALLOC
2547#define MEM_LOG_ALLOC(n,t,a) (a)
2548#endif
2549#ifndef MEM_LOG_REALLOC
2550#define MEM_LOG_REALLOC(n,t,v,a) (a)
2551#endif
2552#ifndef MEM_LOG_FREE
2553#define MEM_LOG_FREE(a) (a)
2554#endif
2555
d1401ee9
MHM
2556#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2557#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2558#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2559
2560#ifndef PERL_CORE
a02a5408
JC
2561/* pre 5.9.x compatibility */
2562#define New(x,v,n,t) Newx(v,n,t)
2563#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2564#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2565#endif
a02a5408 2566
ff68c719 2567#define Renew(v,n,t) \
d1401ee9 2568 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2569#define Renewc(v,n,t,c) \
d1401ee9 2570 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2571
2572#ifdef PERL_POISON
2573#define Safefree(d) \
06c0cc96 2574 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2575#else
fe4f188c 2576#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2577#endif
55497cff 2578
dbb57106
YO
2579/* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2580 * as it handles cases like constant string arguments without throwing warnings *
2581 * the cast is required, as is the inequality check, to avoid warnings */
45908e4d 2582#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2583
45908e4d
YO
2584
2585#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)))
2586#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)))
2587#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2588
bdd1531d 2589/* Like above, but returns a pointer to 'd' */
45908e4d
YO
2590#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)))
2591#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)))
45908e4d 2592#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2593
7e337ee0
JH
2594#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2595#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2596#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2597#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2598
caa674f3
DD
2599#ifdef PERL_POISON
2600# define PERL_POISON_EXPR(x) x
2601#else
2602# define PERL_POISON_EXPR(x)
2603#endif
2604
ff68c719 2605#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2606
1b7e2294
KW
2607/*
2608=head1 Handy Values
2609
2610=for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2611
2612Returns the number of elements in the input C array (so you want your
2613zero-based indices to be less than but not equal to).
2614
2615=for apidoc Am|void *|C_ARRAY_END|void *a
2616
2617Returns a pointer to one element past the final element of the input C array.
2618
2619=cut
2620
2621C_ARRAY_END is one past the last: half-open/half-closed range, not
2622last-inclusive range.
2623*/
622913ab 2624#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2625#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2626
2cc61e15
DD
2627#ifdef NEED_VA_COPY
2628# ifdef va_copy
2629# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2630# elif defined(__va_copy)
2631# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2632# else
07798b17 2633# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2634# endif
2635#endif
2636
472d47bc
SB
2637/* convenience debug macros */
2638#ifdef USE_ITHREADS
2639#define pTHX_FORMAT "Perl interpreter: 0x%p"
2640#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2641#define pTHX_VALUE_ (void *)my_perl,
2642#define pTHX_VALUE (void *)my_perl
2643#define pTHX__VALUE_ ,(void *)my_perl,
2644#define pTHX__VALUE ,(void *)my_perl
472d47bc 2645#else
3609ea0d 2646#define pTHX_FORMAT
472d47bc 2647#define pTHX__FORMAT
3609ea0d 2648#define pTHX_VALUE_
472d47bc 2649#define pTHX_VALUE
3609ea0d 2650#define pTHX__VALUE_
472d47bc
SB
2651#define pTHX__VALUE
2652#endif /* USE_ITHREADS */
3609ea0d 2653
2acdbac1
NC
2654/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2655 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2656 CPAN::Unpack show any users outside the core. */
2657#ifdef PERL_CORE
dc6e8de0
A
2658# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2659 "Use of " s " is deprecated")
c9680906
A
2660# define deprecate_disappears_in(when,message) \
2661 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2662 message ", and will disappear in Perl " when)
ac641426
A
2663# define deprecate_fatal_in(when,message) \
2664 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2665 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2666#endif
2667
dfff4baf
BF
2668/* Internal macros to deal with gids and uids */
2669#ifdef PERL_CORE
2670
2671# if Uid_t_size > IVSIZE
2672# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2673# define SvUID(sv) SvNV(sv)
07798b17
AC
2674# elif Uid_t_sign <= 0
2675# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2676# define SvUID(sv) SvIV(sv)
dfff4baf 2677# else
07798b17
AC
2678# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2679# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2680# endif /* Uid_t_size */
2681
2682# if Gid_t_size > IVSIZE
2683# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2684# define SvGID(sv) SvNV(sv)
07798b17
AC
2685# elif Gid_t_sign <= 0
2686# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2687# define SvGID(sv) SvIV(sv)
dfff4baf 2688# else
07798b17
AC
2689# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2690# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2691# endif /* Gid_t_size */
2692
2693#endif
2694
6a5bc5ac 2695#endif /* PERL_HANDY_H_ */
9d745869 2696
e9a8c099 2697/*
14d04a33 2698 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2699 */