This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Change return type of regcurly to bool
[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
4aada8b9
KW
481=for apidoc Am|bool|memCHRs|"list"|char c
482Returns the position of the first occurence of the byte C<c> in the literal
483string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
484treated as unsigned char. Thus this macro can be used to determine if C<c> is
485in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
486is C<NUL> (and the set doesn't include C<NUL>).
487
954c1994 488=cut
fc169e00
KW
489
490New macros should use the following conventions for their names (which are
491based on the underlying C library functions):
492
493 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
494
495 Each has two main parameters, string-like operands that are compared
496 against each other, as specified by the macro name. Some macros may
497 additionally have one or potentially even two length parameters. If a length
498 parameter applies to both string parameters, it will be positioned third;
499 otherwise any length parameter immediately follows the string parameter it
500 applies to.
501
502 If the prefix to the name is 'str', the string parameter is a pointer to a C
503 language string. Such a string does not contain embedded NUL bytes; its
504 length may be unknown, but can be calculated by C<strlen()>, since it is
505 terminated by a NUL, which isn't included in its length.
506
507 The optional 'n' following 'str' means that that there is a third parameter,
508 giving the maximum number of bytes to look at in each string. Even if both
509 strings are longer than the length parameter, those extra bytes will be
510 unexamined.
511
512 The 's' suffix means that the 2nd byte string parameter is a literal C
513 double-quoted string. Its length will automatically be calculated by the
514 macro, so no length parameter will ever be needed for it.
515
516 If the prefix is 'mem', the string parameters don't have to be C strings;
517 they may contain embedded NUL bytes, do not necessarily have a terminating
518 NUL, and their lengths can be known only through other means, which in
519 practice are additional parameter(s) passed to the function. All 'mem'
520 functions have at least one length parameter. Barring any 'l' or 's' suffix,
521 there is a single length parameter, in position 3, which applies to both
522 string parameters. The 's' suffix means, as described above, that the 2nd
523 string is a literal double-quoted C string (hence its length is calculated by
524 the macro, and the length parameter to the function applies just to the first
525 string parameter, and hence is positioned just after it). An 'l' suffix
526 means that the 2nd string parameter has its own length parameter, and the
527 signature will look like memFOOl(s1, l1, s2, l2).
528
529 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
530 substring of the 1st string. 'P' if present indicates that the substring
531 must be a "proper" one in tha mathematical sense that the first one must be
532 strictly larger than the 2nd.
533
954c1994
GS
534*/
535
62946e08 536
75400963
KW
537#define strNE(s1,s2) (strcmp(s1,s2) != 0)
538#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
539#define strLT(s1,s2) (strcmp(s1,s2) < 0)
540#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
541#define strGT(s1,s2) (strcmp(s1,s2) > 0)
542#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 543
75400963
KW
544#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
545#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 546
9d3980bc
KW
547#define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
548#define memNE(s1,s2,l) (! memEQ(s1,s2,l))
36477c24 549
085b7534 550/* memEQ and memNE where second comparand is a string constant */
568a785a 551#define memEQs(s1, l, s2) \
777fa2cb 552 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 553#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 554
fdbb9a7c
KW
555/* Keep these private until we decide it was a good idea */
556#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
557
558#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
559
bdb7e3f0 560#define memBEGINs(s1, l, s2) \
30a6480c 561 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 562 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158 563#define memBEGINPs(s1, l, s2) \
30a6480c 564 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
de627158 565 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0 566#define memENDs(s1, l, s2) \
30a6480c 567 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 568 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424 569#define memENDPs(s1, l, s2) \
30a6480c 570 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
b80f8424 571 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 572#endif /* End of making macros private */
bdb7e3f0 573
062b6850
KW
574#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
575#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
576#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
577#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
578
4aada8b9
KW
579#define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
580
bbce6d69
PP
581/*
582 * Character classes.
583 *
584 * Unfortunately, the introduction of locales means that we
585 * can't trust isupper(), etc. to tell the truth. And when
586 * it comes to /\w+/ with tainting enabled, we *must* be able
587 * to trust our character classes.
588 *
589 * Therefore, the default tests in the text of Perl will be
590 * independent of locale. Any code that wants to depend on
591 * the current locale will use the tests that begin with "lc".
592 */
593
2304df62
AD
594#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
595# ifndef CTYPE256
596# define CTYPE256
597# endif
598#endif
599
954c1994 600/*
ccfc67b7 601
dcccc8ff 602=head1 Character classification
243effed
KW
603This section is about functions (really macros) that classify characters
604into types, such as punctuation versus alphabetic, etc. Most of these are
605analogous to regular expression character classes. (See
606L<perlrecharclass/POSIX Character Classes>.) There are several variants for
607each class. (Not all macros have all variants; each item below lists the
608ones valid for it.) None are affected by C<use bytes>, and only the ones
609with C<LC> in the name are affected by the current locale.
610
d713f9d9
KW
611The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
612treating it as a code point, and returns a boolean as to whether or not the
613character represented by it is (or on non-ASCII platforms, corresponds to) an
6aff1f14
KW
614ASCII character in the named class based on platform, Unicode, and Perl rules.
615If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 616
c98722a4 617Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
618with no suffix C<"_A">. This variant is used to emphasize by its name that
619only ASCII-range characters can return TRUE.
4b9734bf 620
d60679e1 621Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
622onto the platform. That is, the code points that are ASCII are unaffected,
623since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
624as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
625true when called with the code point 0xDF, which is a word character in both
4650c663 626ASCII and EBCDIC (though it represents different characters in each).
d713f9d9
KW
627If the input is a number that doesn't fit in an octet, FALSE is returned.
628(Perl's documentation uses a colloquial definition of Latin-1, to include all
629code points below 256.)
243effed 630
d713f9d9
KW
631Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
632inputs below 256, but if the code point is larger than 255, Unicode rules are
633used to determine if it is in the character class. For example,
d0da05db 634C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 635WITH MACRON in Unicode, and is a word character.
243effed 636
059703b0
KW
637Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
638but are used for UTF-8 encoded strings. The two forms are different names for
639the same thing. Each call to one of these classifies the first character of
640the string starting at C<p>. The second parameter, C<e>, points to anywhere in
641the string beyond the first character, up to one byte past the end of the
642entire string. Although both variants are identical, the suffix C<_safe> in
643one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
644provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
645C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
646some way, the program may croak, or the function may return FALSE, at the
647discretion of the implementation, and subject to change in future releases.
243effed 648
d713f9d9
KW
649Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
650but the result is based on the current locale, which is what C<LC> in the name
651stands for. If Perl can determine that the current locale is a UTF-8 locale,
652it uses the published Unicode rules; otherwise, it uses the C library function
653that gives the named classification. For example, C<isDIGIT_LC()> when not in
654a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
655returned if the input won't fit into an octet. On some platforms where the C
656library function is known to be defective, Perl changes its result to follow
657the POSIX standard's rules.
243effed 658
d713f9d9
KW
659Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
660than 256, but for larger ones it returns the Unicode classification of the code
661point.
243effed 662
059703b0
KW
663Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
664C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
665are different names for the same thing. Each call to one of these classifies
666the first character of the string starting at C<p>. The second parameter,
667C<e>, points to anywhere in the string beyond the first character, up to one
668byte past the end of the entire string. Although both variants are identical,
669the suffix C<_safe> in one name emphasizes that it will not attempt to read
670beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
671is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
672character is malformed in some way, the program may croak, or the function may
673return FALSE, at the discretion of the implementation, and subject to change in
674future releases.
ccfc67b7 675
d713f9d9
KW
676=for apidoc Am|bool|isALPHA|int ch
677Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
678analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
679See the L<top of this section|/Character classification> for an explanation of
680variants
059703b0
KW
681C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>,
682C<isALPHA_utf8_safe>, C<isALPHA_LC>, C<isALPHA_LC_uvchr>, C<isALPHA_LC_utf8>,
683and C<isALPHA_LC_utf8_safe>.
8a58bdcf 684
f16858ed
KW
685=cut
686
687Here and below, we add the protoypes of these macros for downstream programs
688that would be interested in them, such as Devel::PPPort
689
690=for apidoc Amh|bool|isALPHA_A|int ch
691=for apidoc Amh|bool|isALPHA_L1|int ch
692=for apidoc Amh|bool|isALPHA_uvchr|int ch
693=for apidoc Amh|bool|isALPHA_utf8_safe|U8 * s|U8 * end
d23c7e08 694=for apidoc Amh|bool|isALPHA_utf8|U8 * s|U8 * end
f16858ed
KW
695=for apidoc Amh|bool|isALPHA_LC|int ch
696=for apidoc Amh|bool|isALPHA_LC_uvchr|int ch
697=for apidoc Amh|bool|isALPHA_LC_utf8_safe|U8 * s| U8 *end
698
d713f9d9
KW
699=for apidoc Am|bool|isALPHANUMERIC|int ch
700Returns a boolean indicating whether the specified character is one of
701C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
702See the L<top of this section|/Character classification> for an explanation of
703variants
d0da05db 704C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
059703b0
KW
705C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>,
706C<isALPHANUMERIC_LC_uvchr>, C<isALPHANUMERIC_LC_utf8>, and
707C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 708
255b632a
KW
709A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
710this corresponds to the C language alphanumeric definition). Also
711there are the variants
712C<isALNUMC_A>, C<isALNUMC_L1>
713C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
714
f16858ed
KW
715=for apidoc Amh|bool|isALPHANUMERIC_A|int ch
716=for apidoc Amh|bool|isALPHANUMERIC_L1|int ch
717=for apidoc Amh|bool|isALPHANUMERIC_uvchr|int ch
718=for apidoc Amh|bool|isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
d23c7e08 719=for apidoc Amh|bool|isALPHANUMERIC_utf8|U8 * s|U8 * end
f16858ed
KW
720=for apidoc Amh|bool|isALPHANUMERIC_LC|int ch
721=for apidoc Amh|bool|isALPHANUMERIC_LC_uvchr|int ch
722=for apidoc Amh|bool|isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
723=for apidoc Amh|bool|isALNUMC|int ch
724=for apidoc Amh|bool|isALNUMC_A|int ch
725=for apidoc Amh|bool|isALNUMC_L1|int ch
726=for apidoc Amh|bool|isALNUMC_LC|int ch
727=for apidoc Amh|bool|isALNUMC_LC_uvchr|int ch
728
d713f9d9 729=for apidoc Am|bool|isASCII|int ch
8a58bdcf 730Returns a boolean indicating whether the specified character is one of the 128
243effed 731characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 732On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
733character corresponds to an ASCII character. Variants C<isASCII_A()> and
734C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
735See the L<top of this section|/Character classification> for an explanation of
736variants
059703b0
KW
737C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_utf8_safe>, C<isASCII_LC>,
738C<isASCII_LC_uvchr>, C<isASCII_LC_utf8>, and C<isASCII_LC_utf8_safe>.
739Note, however, that some platforms do not have the C library routine
740C<isascii()>. In these cases, the variants whose names contain C<LC> are the
741same as the corresponding ones without.
243effed 742
f16858ed
KW
743=for apidoc Amh|bool|isASCII_A|int ch
744=for apidoc Amh|bool|isASCII_L1|int ch
745=for apidoc Amh|bool|isASCII_uvchr|int ch
746=for apidoc Amh|bool|isASCII_utf8_safe|U8 * s|U8 * end
d23c7e08 747=for apidoc Amh|bool|isASCII_utf8|U8 * s|U8 * end
f16858ed
KW
748=for apidoc Amh|bool|isASCII_LC|int ch
749=for apidoc Amh|bool|isASCII_LC_uvchr|int ch
750=for apidoc Amh|bool|isASCII_LC_utf8_safe|U8 * s| U8 *end
751
d98532ea
KW
752Also note, that because all ASCII characters are UTF-8 invariant (meaning they
753have the exact same representation (always a single byte) whether encoded in
754UTF-8 or not), C<isASCII> will give the correct results when called with any
059703b0
KW
755byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
756C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
d98532ea 757
243effed
KW
758=for apidoc Am|bool|isBLANK|char ch
759Returns a boolean indicating whether the specified character is a
6aff1f14 760character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
761See the L<top of this section|/Character classification> for an explanation of
762variants
059703b0
KW
763C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>,
764C<isBLANK_utf8_safe>, C<isBLANK_LC>, C<isBLANK_LC_uvchr>, C<isBLANK_LC_utf8>,
765and C<isBLANK_LC_utf8_safe>. Note,
da8c1a98
KW
766however, that some platforms do not have the C library routine
767C<isblank()>. In these cases, the variants whose names contain C<LC> are
768the same as the corresponding ones without.
243effed 769
f16858ed
KW
770=for apidoc Amh|bool|isBLANK_A|int ch
771=for apidoc Amh|bool|isBLANK_L1|int ch
772=for apidoc Amh|bool|isBLANK_uvchr|int ch
773=for apidoc Amh|bool|isBLANK_utf8_safe|U8 * s|U8 * end
d23c7e08 774=for apidoc Amh|bool|isBLANK_utf8|U8 * s|U8 * end
f16858ed
KW
775=for apidoc Amh|bool|isBLANK_LC|int ch
776=for apidoc Amh|bool|isBLANK_LC_uvchr|int ch
777=for apidoc Amh|bool|isBLANK_LC_utf8_safe|U8 * s| U8 *end
778
243effed
KW
779=for apidoc Am|bool|isCNTRL|char ch
780Returns a boolean indicating whether the specified character is a
6aff1f14 781control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
782See the L<top of this section|/Character classification> for an explanation of
783variants
059703b0
KW
784C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>,
785C<isCNTRL_utf8_safe>, C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, C<isCNTRL_LC_utf8>
786and C<isCNTRL_LC_utf8_safe>. On EBCDIC
da8c1a98 787platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 788
f16858ed
KW
789=for apidoc Amh|bool|isCNTRL_A|int ch
790=for apidoc Amh|bool|isCNTRL_L1|int ch
791=for apidoc Amh|bool|isCNTRL_uvchr|int ch
792=for apidoc Amh|bool|isCNTRL_utf8_safe|U8 * s|U8 * end
d23c7e08 793=for apidoc Amh|bool|isCNTRL_utf8|U8 * s|U8 * end
f16858ed
KW
794=for apidoc Amh|bool|isCNTRL_LC|int ch
795=for apidoc Amh|bool|isCNTRL_LC_uvchr|int ch
796=for apidoc Amh|bool|isCNTRL_LC_utf8_safe|U8 * s| U8 *end
797
954c1994 798=for apidoc Am|bool|isDIGIT|char ch
2787a470 799Returns a boolean indicating whether the specified character is a
6aff1f14 800digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 801Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
802See the L<top of this section|/Character classification> for an explanation of
803variants
059703b0
KW
804C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>,
805C<isDIGIT_LC_uvchr>, C<isDIGIT_LC_utf8>, and C<isDIGIT_LC_utf8_safe>.
243effed 806
f16858ed
KW
807=for apidoc Amh|bool|isDIGIT_A|int ch
808=for apidoc Amh|bool|isDIGIT_L1|int ch
809=for apidoc Amh|bool|isDIGIT_uvchr|int ch
810=for apidoc Amh|bool|isDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 811=for apidoc Amh|bool|isDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
812=for apidoc Amh|bool|isDIGIT_LC|int ch
813=for apidoc Amh|bool|isDIGIT_LC_uvchr|int ch
814=for apidoc Amh|bool|isDIGIT_LC_utf8_safe|U8 * s| U8 *end
815
243effed
KW
816=for apidoc Am|bool|isGRAPH|char ch
817Returns a boolean indicating whether the specified character is a
6aff1f14 818graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 819See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
820variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>,
821C<isGRAPH_utf8_safe>, C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>,
822C<isGRAPH_LC_utf8_safe>, and C<isGRAPH_LC_utf8_safe>.
954c1994 823
f16858ed
KW
824=for apidoc Amh|bool|isGRAPH_A|int ch
825=for apidoc Amh|bool|isGRAPH_L1|int ch
826=for apidoc Amh|bool|isGRAPH_uvchr|int ch
827=for apidoc Amh|bool|isGRAPH_utf8_safe|U8 * s|U8 * end
d23c7e08 828=for apidoc Amh|bool|isGRAPH_utf8|U8 * s|U8 * end
f16858ed
KW
829=for apidoc Amh|bool|isGRAPH_LC|int ch
830=for apidoc Amh|bool|isGRAPH_LC_uvchr|int ch
831=for apidoc Amh|bool|isGRAPH_LC_utf8_safe|U8 * s| U8 *end
832
0c82b6df 833=for apidoc Am|bool|isLOWER|char ch
2787a470 834Returns a boolean indicating whether the specified character is a
6aff1f14 835lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
836See the L<top of this section|/Character classification> for an explanation of
837variants
059703b0
KW
838C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>,
839C<isLOWER_utf8_safe>, C<isLOWER_LC>, C<isLOWER_LC_uvchr>, C<isLOWER_LC_utf8>,
840and C<isLOWER_LC_utf8_safe>.
0c82b6df 841
f16858ed
KW
842=for apidoc Amh|bool|isLOWER_A|int ch
843=for apidoc Amh|bool|isLOWER_L1|int ch
844=for apidoc Amh|bool|isLOWER_uvchr|int ch
845=for apidoc Amh|bool|isLOWER_utf8_safe|U8 * s|U8 * end
d23c7e08 846=for apidoc Amh|bool|isLOWER_utf8|U8 * s|U8 * end
f16858ed
KW
847=for apidoc Amh|bool|isLOWER_LC|int ch
848=for apidoc Amh|bool|isLOWER_LC_uvchr|int ch
849=for apidoc Amh|bool|isLOWER_LC_utf8_safe|U8 * s| U8 *end
850
c99e91e9 851=for apidoc Am|bool|isOCTAL|char ch
2787a470 852Returns a boolean indicating whether the specified character is an
6aff1f14 853octal digit, [0-7].
243effed
KW
854The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
855C<isOCTAL>.
856
f16858ed
KW
857=for apidoc Amh|bool|isOCTAL_A|int ch
858=for apidoc Amh|bool|isOCTAL_L1|int ch
859
243effed
KW
860=for apidoc Am|bool|isPUNCT|char ch
861Returns a boolean indicating whether the specified character is a
6aff1f14
KW
862punctuation character, analogous to C<m/[[:punct:]]/>.
863Note that the definition of what is punctuation isn't as
243effed
KW
864straightforward as one might desire. See L<perlrecharclass/POSIX Character
865Classes> for details.
dcccc8ff 866See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
867variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>,
868C<isPUNCT_utf8_safe>, C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, C<isPUNCT_LC_utf8>,
869and C<isPUNCT_LC_utf8_safe>.
c99e91e9 870
f16858ed
KW
871=for apidoc Amh|bool|isPUNCT_A|int ch
872=for apidoc Amh|bool|isPUNCT_L1|int ch
873=for apidoc Amh|bool|isPUNCT_uvchr|int ch
874=for apidoc Amh|bool|isPUNCT_utf8_safe|U8 * s|U8 * end
d23c7e08 875=for apidoc Amh|bool|isPUNCT_utf8|U8 * s|U8 * end
f16858ed
KW
876=for apidoc Amh|bool|isPUNCT_LC|int ch
877=for apidoc Amh|bool|isPUNCT_LC_uvchr|int ch
878=for apidoc Amh|bool|isPUNCT_LC_utf8_safe|U8 * s| U8 *end
879
0c82b6df 880=for apidoc Am|bool|isSPACE|char ch
2787a470 881Returns a boolean indicating whether the specified character is a
6aff1f14 882whitespace character. This is analogous
398d098a 883to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 884this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
885locale forms of this macro (the ones with C<LC> in their names) matched
886precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
887in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
888(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
889See the L<top of this section|/Character classification> for an explanation of
890variants
059703b0
KW
891C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>,
892C<isSPACE_utf8_safe>, C<isSPACE_LC>, C<isSPACE_LC_uvchr>, C<isSPACE_LC_utf8>,
893and C<isSPACE_LC_utf8_safe>.
0c82b6df 894
f16858ed
KW
895=for apidoc Amh|bool|isSPACE_A|int ch
896=for apidoc Amh|bool|isSPACE_L1|int ch
897=for apidoc Amh|bool|isSPACE_uvchr|int ch
898=for apidoc Amh|bool|isSPACE_utf8_safe|U8 * s|U8 * end
d23c7e08 899=for apidoc Amh|bool|isSPACE_utf8|U8 * s|U8 * end
f16858ed
KW
900=for apidoc Amh|bool|isSPACE_LC|int ch
901=for apidoc Amh|bool|isSPACE_LC_uvchr|int ch
902=for apidoc Amh|bool|isSPACE_LC_utf8_safe|U8 * s| U8 *end
903
398d098a
KW
904=for apidoc Am|bool|isPSXSPC|char ch
905(short for Posix Space)
779cf272
KW
906Starting in 5.18, this is identical in all its forms to the
907corresponding C<isSPACE()> macros.
398d098a
KW
908The locale forms of this macro are identical to their corresponding
909C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
910non-locale forms differ from their C<isSPACE()> forms only in that the
911C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
912Otherwise they are identical. Thus this macro is analogous to what
913C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 914See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
915variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
916C<isPSXSPC_utf8_safe>, C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>,
917C<isPSXSPC_LC_utf8>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 918
f16858ed
KW
919=for apidoc Amh|bool|isPSXSPC_A|int ch
920=for apidoc Amh|bool|isPSXSPC_L1|int ch
921=for apidoc Amh|bool|isPSXSPC_uvchr|int ch
922=for apidoc Amh|bool|isPSXSPC_utf8_safe|U8 * s|U8 * end
d23c7e08 923=for apidoc Amh|bool|isPSXSPC_utf8|U8 * s|U8 * end
f16858ed
KW
924=for apidoc Amh|bool|isPSXSPC_LC|int ch
925=for apidoc Amh|bool|isPSXSPC_LC_uvchr|int ch
926=for apidoc Amh|bool|isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
927
954c1994 928=for apidoc Am|bool|isUPPER|char ch
2787a470 929Returns a boolean indicating whether the specified character is an
6aff1f14 930uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 931See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
932variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>,
933C<isUPPER_utf8_safe>, C<isUPPER_LC>, C<isUPPER_LC_uvchr>, C<isUPPER_LC_utf8>,
934and C<isUPPER_LC_utf8_safe>.
954c1994 935
f16858ed
KW
936=for apidoc Amh|bool|isUPPER_A|int ch
937=for apidoc Amh|bool|isUPPER_L1|int ch
938=for apidoc Amh|bool|isUPPER_uvchr|int ch
939=for apidoc Amh|bool|isUPPER_utf8_safe|U8 * s|U8 * end
d23c7e08 940=for apidoc Amh|bool|isUPPER_utf8|U8 * s|U8 * end
f16858ed
KW
941=for apidoc Amh|bool|isUPPER_LC|int ch
942=for apidoc Amh|bool|isUPPER_LC_uvchr|int ch
943=for apidoc Amh|bool|isUPPER_LC_utf8_safe|U8 * s| U8 *end
944
243effed 945=for apidoc Am|bool|isPRINT|char ch
8eea39dd 946Returns a boolean indicating whether the specified character is a
6aff1f14 947printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
948See the L<top of this section|/Character classification> for an explanation of
949variants
059703b0
KW
950C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>,
951C<isPRINT_utf8_safe>, C<isPRINT_LC>, C<isPRINT_LC_uvchr>, C<isPRINT_LC_utf8>,
952and C<isPRINT_LC_utf8_safe>.
243effed 953
f16858ed
KW
954=for apidoc Amh|bool|isPRINT_A|int ch
955=for apidoc Amh|bool|isPRINT_L1|int ch
956=for apidoc Amh|bool|isPRINT_uvchr|int ch
957=for apidoc Amh|bool|isPRINT_utf8_safe|U8 * s|U8 * end
d23c7e08 958=for apidoc Amh|bool|isPRINT_utf8|U8 * s|U8 * end
f16858ed
KW
959=for apidoc Amh|bool|isPRINT_LC|int ch
960=for apidoc Amh|bool|isPRINT_LC_uvchr|int ch
961=for apidoc Amh|bool|isPRINT_LC_utf8_safe|U8 * s| U8 *end
962
243effed
KW
963=for apidoc Am|bool|isWORDCHAR|char ch
964Returns a boolean indicating whether the specified character is a character
965that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
966in a regular expression. A word character is an alphabetic character, a
967decimal digit, a connecting punctuation character (such as an underscore), or
968a "mark" character that attaches to one of those (like some sort of accent).
969C<isALNUM()> is a synonym provided for backward compatibility, even though a
970word character includes more than the standard C language meaning of
971alphanumeric.
dcccc8ff 972See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
973variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
974C<isWORDCHAR_utf8>, and C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>,
975C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and C<isWORDCHAR_LC_utf8_safe>
976are also as described there, but additionally include the platform's native
977underscore.
8a58bdcf 978
f16858ed
KW
979=for apidoc Amh|bool|isWORDCHAR_A|int ch
980=for apidoc Amh|bool|isWORDCHAR_L1|int ch
981=for apidoc Amh|bool|isWORDCHAR_uvchr|int ch
982=for apidoc Amh|bool|isWORDCHAR_utf8_safe|U8 * s|U8 * end
d23c7e08 983=for apidoc Amh|bool|isWORDCHAR_utf8|U8 * s|U8 * end
f16858ed
KW
984=for apidoc Amh|bool|isWORDCHAR_LC|int ch
985=for apidoc Amh|bool|isWORDCHAR_LC_uvchr|int ch
986=for apidoc Amh|bool|isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
987=for apidoc Amh|bool|isALNUM|int ch
988=for apidoc Amh|bool|isALNUM_A|int ch
989=for apidoc Amh|bool|isALNUM_LC|int ch
990=for apidoc Amh|bool|isALNUM_LC_uvchr|int ch
991
8a58bdcf
KW
992=for apidoc Am|bool|isXDIGIT|char ch
993Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
994digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
995and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
996See the L<top of this section|/Character classification> for an explanation of
997variants
059703b0
KW
998C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>,
999C<isXDIGIT_LC_uvchr>, C<isXDIGIT_LC_utf8>, and C<isXDIGIT_LC_utf8_safe>.
243effed 1000
f16858ed
KW
1001=for apidoc Amh|bool|isXDIGIT_A|int ch
1002=for apidoc Amh|bool|isXDIGIT_L1|int ch
1003=for apidoc Amh|bool|isXDIGIT_uvchr|int ch
1004=for apidoc Amh|bool|isXDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 1005=for apidoc Amh|bool|isXDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
1006=for apidoc Amh|bool|isXDIGIT_LC|int ch
1007=for apidoc Amh|bool|isXDIGIT_LC_uvchr|int ch
1008=for apidoc Amh|bool|isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1009
3c3ecf18
KW
1010=for apidoc Am|bool|isIDFIRST|char ch
1011Returns a boolean indicating whether the specified character can be the first
1012character of an identifier. This is very close to, but not quite the same as
1013the official Unicode property C<XID_Start>. The difference is that this
1014returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
1015See the L<top of this section|/Character classification> for an explanation of
1016variants
059703b0
KW
1017C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
1018C<isIDFIRST_utf8_safe>, C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>,
1019C<isIDFIRST_LC_utf8>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18 1020
f16858ed
KW
1021=for apidoc Amh|bool|isIDFIRST_A|int ch
1022=for apidoc Amh|bool|isIDFIRST_L1|int ch
1023=for apidoc Amh|bool|isIDFIRST_uvchr|int ch
1024=for apidoc Amh|bool|isIDFIRST_utf8_safe|U8 * s|U8 * end
d23c7e08 1025=for apidoc Amh|bool|isIDFIRST_utf8|U8 * s|U8 * end
f16858ed
KW
1026=for apidoc Amh|bool|isIDFIRST_LC|int ch
1027=for apidoc Amh|bool|isIDFIRST_LC_uvchr|int ch
1028=for apidoc Amh|bool|isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1029
3c3ecf18
KW
1030=for apidoc Am|bool|isIDCONT|char ch
1031Returns a boolean indicating whether the specified character can be the
1032second or succeeding character of an identifier. This is very close to, but
1033not quite the same as the official Unicode property C<XID_Continue>. The
1034difference is that this returns true only if the input character also matches
dcccc8ff 1035L</isWORDCHAR>. See the L<top of this section|/Character classification> for
059703b0
KW
1036an explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
1037C<isIDCONT_utf8>, C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>,
1038C<isIDCONT_LC_utf8>, and C<isIDCONT_LC_utf8_safe>.
3c3ecf18 1039
f16858ed
KW
1040=for apidoc Amh|bool|isIDCONT_A|int ch
1041=for apidoc Amh|bool|isIDCONT_L1|int ch
1042=for apidoc Amh|bool|isIDCONT_uvchr|int ch
1043=for apidoc Amh|bool|isIDCONT_utf8_safe|U8 * s|U8 * end
d23c7e08 1044=for apidoc Amh|bool|isIDCONT_utf8|U8 * s|U8 * end
f16858ed
KW
1045=for apidoc Amh|bool|isIDCONT_LC|int ch
1046=for apidoc Amh|bool|isIDCONT_LC_uvchr|int ch
1047=for apidoc Amh|bool|isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1048
243effed 1049=head1 Miscellaneous Functions
8eea39dd 1050
95a59cab 1051=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 1052Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
1053Behaviour is only well defined when isXDIGIT(*str) is true.
1054
e7c1e6c1 1055=head1 Character case changing
21da7284
KW
1056Perl uses "full" Unicode case mappings. This means that converting a single
1057character to another case may result in a sequence of more than one character.
1058For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1059character sequence C<SS>. This presents some complications The lowercase of
1060all characters in the range 0..255 is a single character, and thus
1061C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1062return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1063an API that does allow every possible legal result to be returned.) Likewise
1064no other function that is crippled by not being able to give the correct
1065results for the full range of possible inputs has been implemented here.
e7c1e6c1 1066
d713f9d9 1067=for apidoc Am|U8|toUPPER|int ch
1f607577
KW
1068Converts the specified character to uppercase. If the input is anything but an
1069ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 1070C<toUPPER_A> is equivalent.
954c1994 1071
d0da05db
KW
1072=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1073Converts the code point C<cp> to its uppercase version, and
1074stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1075point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1076that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1077bytes since the uppercase version may be longer than the original character.
1078
1079The first code point of the uppercased version is returned
21da7284
KW
1080(but note, as explained at L<the top of this section|/Character case
1081changing>, that there may be more.)
1f607577 1082
059703b0 1083=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1084Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1085extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
1086stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1087that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1088bytes since the uppercase version may be longer than the original character.
1089
1090The first code point of the uppercased version is returned
21da7284
KW
1091(but note, as explained at L<the top of this section|/Character case
1092changing>, that there may be more).
1f607577 1093
059703b0
KW
1094It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1095S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1096the UTF-8 for the input character is malformed in some way, the program may
1097croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1098of the implementation, and subject to change in future releases.
a239b1e2 1099
059703b0
KW
1100=for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1101Same as L</toUPPER_utf8>.
1f607577 1102
25200305
KW
1103=for apidoc Am|U8|toFOLD|U8 ch
1104Converts the specified character to foldcase. If the input is anything but an
1105ASCII uppercase character, that input character itself is returned. Variant
1106C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 1107Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 1108
d0da05db
KW
1109=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1110Converts the code point C<cp> to its foldcase version, and
1111stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1112point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1113that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1114bytes since the foldcase version may be longer than the original character.
1115
1116The first code point of the foldcased version is returned
21da7284
KW
1117(but note, as explained at L<the top of this section|/Character case
1118changing>, that there may be more).
1f607577 1119
059703b0 1120=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1121Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1122extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
1123stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1124that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1125bytes since the foldcase version may be longer than the original character.
1126
1127The first code point of the foldcased version is returned
21da7284
KW
1128(but note, as explained at L<the top of this section|/Character case
1129changing>, that there may be more).
1f607577 1130
059703b0 1131It will not attempt
a239b1e2
KW
1132to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1133true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1134input character is malformed in some way, the program may croak, or the
1135function may return the REPLACEMENT CHARACTER, at the discretion of the
1136implementation, and subject to change in future releases.
1137
059703b0
KW
1138=for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1139Same as L</toFOLD_utf8>.
1f607577
KW
1140
1141=for apidoc Am|U8|toLOWER|U8 ch
1142Converts the specified character to lowercase. If the input is anything but an
1143ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 1144C<toLOWER_A> is equivalent.
954c1994 1145
1f607577 1146=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
1147Converts the specified Latin1 character to lowercase. The results are
1148undefined if the input doesn't fit in a byte.
1f607577
KW
1149
1150=for apidoc Am|U8|toLOWER_LC|U8 ch
1151Converts the specified character to lowercase using the current locale's rules,
1152if possible; otherwise returns the input character itself.
1153
d0da05db
KW
1154=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1155Converts the code point C<cp> to its lowercase version, and
1156stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1157point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1158that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1159bytes since the lowercase version may be longer than the original character.
1160
1161The first code point of the lowercased version is returned
21da7284
KW
1162(but note, as explained at L<the top of this section|/Character case
1163changing>, that there may be more).
1f607577 1164
bd350c85 1165=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1166Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1167extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
1168stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1169that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1170bytes since the lowercase version may be longer than the original character.
1171
1172The first code point of the lowercased version is returned
21da7284
KW
1173(but note, as explained at L<the top of this section|/Character case
1174changing>, that there may be more).
059703b0
KW
1175It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1176S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1177the UTF-8 for the input character is malformed in some way, the program may
1178croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1179of the implementation, and subject to change in future releases.
1f607577 1180
059703b0
KW
1181=for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1182Same as L</toLOWER_utf8>.
1f607577 1183
25200305
KW
1184=for apidoc Am|U8|toTITLE|U8 ch
1185Converts the specified character to titlecase. If the input is anything but an
1186ASCII lowercase character, that input character itself is returned. Variant
b7d90381 1187C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 1188range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 1189not a concept used in locale handling, so there is no functionality for that.)
25200305 1190
d0da05db
KW
1191=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1192Converts the code point C<cp> to its titlecase version, and
1193stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1194point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1195that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1196bytes since the titlecase version may be longer than the original character.
1197
1198The first code point of the titlecased version is returned
21da7284
KW
1199(but note, as explained at L<the top of this section|/Character case
1200changing>, that there may be more).
1f607577 1201
059703b0 1202=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1203Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1204extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1205stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1206that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1207bytes since the titlecase version may be longer than the original character.
1208
1209The first code point of the titlecased version is returned
21da7284
KW
1210(but note, as explained at L<the top of this section|/Character case
1211changing>, that there may be more).
1f607577 1212
059703b0 1213It will not attempt
a239b1e2
KW
1214to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1215true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1216input character is malformed in some way, the program may croak, or the
1217function may return the REPLACEMENT CHARACTER, at the discretion of the
1218implementation, and subject to change in future releases.
1219
059703b0
KW
1220=for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1221Same as L</toTITLE_utf8>.
1f607577 1222
954c1994 1223=cut
353c9b6f 1224
d0da05db 1225XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1226really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1227and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1228
8a58bdcf 1229Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1230patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1231
954c1994
GS
1232*/
1233
8f5283f4
KW
1234/*
1235 void below because that's the best fit, and works for Devel::PPPort
1236=for apidoc AmnU|void|WIDEST_UTYPE
1237
1238Yields the widest unsigned integer type on the platform, currently either
1239C<U32> or C<64>. This can be used in declarations such as
1240
1241 WIDEST_UTYPE my_uv;
1242
1243or casts
1244
1245 my_uv = (WIDEST_UTYPE) val;
1246
1247=cut
1248
1249*/
de9e2639
KW
1250#ifdef QUADKIND
1251# define WIDEST_UTYPE U64
7c062697
KW
1252#else
1253# define WIDEST_UTYPE U32
1254#endif
1255
3912bc88
KW
1256/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1257 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1258 * bits of information are lost even on a 64-bit machine, but to get the
1259 * compiler to optimize it out if possible. This is because Configure makes
1260 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1261 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1262 *
1263 * For Coverity, be always true, because otherwise Coverity thinks
1264 * it finds several expressions that are always true, independent
1265 * of operands. Well, they are, but that is kind of the point.
220c71bf 1266 */
7e75d1a1 1267#ifndef __COVERITY__
6c5b02ac
KW
1268 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1269 * pointer) */
1270#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1271 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1272#else
1273#define FITS_IN_8_BITS(c) (1)
1274#endif
cf301eb7 1275
45f4bb73 1276/* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
833b0f46 1277 * Written this way so that after optimization, only one conditional test is
76d3ad4c
KW
1278 * needed. (The NV casts stop any warnings about comparison always being true
1279 * if called with an unsigned. The cast preserves the sign, which is all we
1280 * care about.) */
1281#define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1282 __ASSERT_((NV) (n) >= 0) \
1283 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
833b0f46 1284
94250c4f
KW
1285/* Returns true if c is in the range l..u, where 'l' is non-negative
1286 * Written this way so that after optimization, only one conditional test is
4758c20d 1287 * needed. */
1eaefa6e 1288#define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
4758c20d 1289 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
4758c20d
KW
1290 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1291 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
45f4bb73 1292 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
305fe86e 1293
41f43cc2 1294#ifdef EBCDIC
b6340bd0 1295# ifndef _ALL_SOURCE
0852beac
KW
1296 /* The native libc isascii() et.al. functions return the wrong results
1297 * on at least z/OS unless this is defined. */
b6340bd0
KW
1298# error _ALL_SOURCE should probably be defined
1299# endif
41f43cc2 1300#else
0852beac
KW
1301 /* There is a simple definition of ASCII for ASCII platforms. But the
1302 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1303 * other macros below.
3f3c579d
KW
1304 *
1305 * The cast here is used instead of '(c) >= 0', because some compilers emit
1306 * a warning that that test is always true when the parameter is an
1307 * unsigned type. khw supposes that it could be written as
1308 * && ((c) == '\0' || (c) > 0)
1309 * to avoid the message, but the cast will likely avoid extra branches even
1310 * with stupid compilers.
1311 *
1312 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1313 * a pointer) */
9c903d59 1314# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1315#endif
1316
38694112
KW
1317/* Take the eight possible bit patterns of the lower 3 bits and you get the
1318 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1319 * can be ignored. If the rest match '0', we have an octal */
1320#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1321
9fb1bf9d 1322#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1323
a500dc72
KW
1324/* Character class numbers. For internal core Perl use only. The ones less
1325 * than 32 are used in PL_charclass[] and the ones up through the one that
1326 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1327 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1328 * their actual definitions are here. If that file has a name not used here,
1329 * it won't compile.
1709d539
KW
1330 *
1331 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561 1332 * frequency of their use. This gives a slight edge to exiting a loop earlier
58a3ba2c
KW
1333 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1334 * real Posix character class, and some (small) inefficiencies in regular
1335 * expression handling would be introduced by putting it in the middle of those
1336 * that are. Also, cntrl and ascii come after the others as it may be useful
1337 * to group these which have no members that match above Latin1, (or above
1338 * ASCII in the latter case) */
1339
1709d539
KW
1340# define _CC_WORDCHAR 0 /* \w and [:word:] */
1341# define _CC_DIGIT 1 /* \d and [:digit:] */
1342# define _CC_ALPHA 2 /* [:alpha:] */
1343# define _CC_LOWER 3 /* [:lower:] */
1344# define _CC_UPPER 4 /* [:upper:] */
1345# define _CC_PUNCT 5 /* [:punct:] */
1346# define _CC_PRINT 6 /* [:print:] */
15861f94 1347# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1348# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1349# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
779cf272 1350# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1351# define _CC_BLANK 11 /* [:blank:] */
1352# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1353# define _CC_CNTRL 13 /* [:cntrl:] */
1354# define _CC_ASCII 14 /* [:ascii:] */
1355# define _CC_VERTSPACE 15 /* \v */
1709d539 1356
a0947d7b
KW
1357# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1358
1709d539 1359/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1360 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1361# define _CC_IDFIRST 16
1362# define _CC_CHARNAME_CONT 17
1363# define _CC_NONLATIN1_FOLD 18
1364# define _CC_NONLATIN1_SIMPLE_FOLD 19
1365# define _CC_QUOTEMETA 20
1366# define _CC_NON_FINAL_FOLD 21
1367# define _CC_IS_IN_SOME_FOLD 22
2ae9030c
KW
1368# define _CC_BINDIGIT 23
1369# define _CC_OCTDIGIT 24
1370# define _CC_MNEMONIC_CNTRL 25
073c22b3
KW
1371
1372/* This next group is only used on EBCDIC platforms, so theoretically could be
1373 * shared with something entirely different that's only on ASCII platforms */
abb8abf6
KW
1374# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
1375/* Unused: 24-30
f4cdb42c
KW
1376 * If more bits are needed, one could add a second word for non-64bit
1377 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1378 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1379 * is used only for optimization (as of this writing), and differs in the
1380 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1381 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1382 * would just cause /i regexes which match them to run less efficiently.
1383 * Similarly the EBCDIC-only bits are used just for speed, and could be
1384 * replaced by other means */
96ac0975 1385
3a371f2f
KW
1386#if defined(PERL_CORE) || defined(PERL_EXT)
1387/* An enum version of the character class numbers, to help compilers
1388 * optimize */
1389typedef enum {
3a371f2f 1390 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1391 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1392 _CC_ENUM_ASCII = _CC_ASCII,
1393 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1394 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1395 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1396 _CC_ENUM_DIGIT = _CC_DIGIT,
1397 _CC_ENUM_GRAPH = _CC_GRAPH,
1398 _CC_ENUM_LOWER = _CC_LOWER,
1399 _CC_ENUM_PRINT = _CC_PRINT,
1400 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1401 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1402 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1403 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1404 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1405 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1406} _char_class_number;
1407#endif
1408
86f72d56 1409#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1410
6635f04f 1411START_EXTERN_C
96ac0975
NC
1412# ifdef DOINIT
1413EXTCONST U32 PL_charclass[] = {
1414# include "l1_char_class_tab.h"
1415};
1416
1417# else /* ! DOINIT */
1418EXTCONST U32 PL_charclass[];
1419# endif
6635f04f 1420END_EXTERN_C
96ac0975 1421
265c1f46 1422 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1423# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1424
1425 /* For internal core Perl use only: the base macro for defining macros like
1426 * isALPHA */
ff7ecfc3 1427# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1428 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1429
f4cdb42c
KW
1430 /* The mask for the _A versions of the macros; it just adds in the bit for
1431 * ASCII. */
1432# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1433
4650c663
KW
1434 /* For internal core Perl use only: the base macro for defining macros like
1435 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1436 * the ASCII bit are present */
b7d90381
KW
1437# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1438 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1439 == _CC_mask_A(classnum)))
f4cdb42c 1440
26c1d9d8
KW
1441/* On ASCII platforms certain classes form a single range. It's faster to
1442 * special case these. isDIGIT is a single range on all platforms */
b877c1ff
KW
1443# ifdef EBCDIC
1444# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1445# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1446# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1447# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1448# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1449# else
26c1d9d8 1450 /* By folding the upper and lowercase, we can use a single range */
b877c1ff 1451# define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
26c1d9d8 1452# define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
b877c1ff
KW
1453# define isLOWER_A(c) inRANGE(c, 'a', 'z')
1454# define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1455# define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1456# endif
15861f94 1457# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1458# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1459# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
b877c1ff 1460# define isDIGIT_A(c) inRANGE(c, '0', '9')
f4cdb42c
KW
1461# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1462# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
f4cdb42c 1463# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1464# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1465 */
d95f8b6a 1466# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1467# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1468# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1469# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1470
1471 /* continuation character for legal NAME in \N{NAME} */
1472# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1473
1474# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1475# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1476# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1477# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1478# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1479# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1480# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1481# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1482# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1483# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1484
0852beac
KW
1485# ifdef EBCDIC
1486# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1487# endif
1488
f12c0118
KW
1489 /* Participates in a single-character fold with a character above 255 */
1490# 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)))
1491
1492 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1493# 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 1494
4eeeb416 1495# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1496# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1497 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1498# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1499 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1500# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1501 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1502#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1503
1504 /* If we don't have perl.h, we are compiling a utility program. Below we
1505 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1506 * to it. Most are coded based on first principles. These are written to
74665a89 1507 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
182c4ace 1508# define isDIGIT_A(c) inRANGE(c, '0', '9')
0852beac 1509# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1510# define isSPACE_A(c) (isBLANK_A(c) \
1511 || (c) == '\n' \
1512 || (c) == '\r' \
1513 || (c) == '\v' \
0852beac 1514 || (c) == '\f')
74665a89
KW
1515 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1516 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1517 * performance (if optimization isn't on), and allow the same code to be
1518 * used for both platform types */
182c4ace
KW
1519# define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1520 || inRANGE((c), 'j', 'r') \
1521 || inRANGE((c), 's', 'z')
1522# define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1523 || inRANGE((c), 'J', 'R') \
1524 || inRANGE((c), 'S', 'Z')
a4d7a999
KW
1525# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1526# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1527# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1528# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
182c4ace
KW
1529# define isXDIGIT_A(c) ( isDIGIT_A(c) \
1530 || inRANGE((c), 'a', 'f') \
1531 || inRANGE((c), 'A', 'F')
74665a89
KW
1532# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1533 || (c) == '#' || (c) == '$' || (c) == '%' \
1534 || (c) == '&' || (c) == '\'' || (c) == '(' \
1535 || (c) == ')' || (c) == '*' || (c) == '+' \
1536 || (c) == ',' || (c) == '.' || (c) == '/' \
1537 || (c) == ':' || (c) == ';' || (c) == '<' \
1538 || (c) == '=' || (c) == '>' || (c) == '?' \
1539 || (c) == '@' || (c) == '[' || (c) == '\\' \
1540 || (c) == ']' || (c) == '^' || (c) == '_' \
1541 || (c) == '`' || (c) == '{' || (c) == '|' \
1542 || (c) == '}' || (c) == '~')
1543# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1544# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1545
0852beac 1546# ifdef EBCDIC
74665a89
KW
1547 /* The below is accurate for the 3 EBCDIC code pages traditionally
1548 * supported by perl. The only difference between them in the controls
1549 * is the position of \n, and that is represented symbolically below */
1550# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1551 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1552 || (c) == '\t' || (c) == '\v' \
182c4ace 1553 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
8ec0a736 1554 || (c) == 7F /* U+7F DEL */ \
182c4ace
KW
1555 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1556 DC[1-3] */ \
74665a89
KW
1557 || (c) == 0x18 /* U+18 CAN */ \
1558 || (c) == 0x19 /* U+19 EOM */ \
182c4ace 1559 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
74665a89
KW
1560 || (c) == 0x26 /* U+17 ETB */ \
1561 || (c) == 0x27 /* U+1B ESC */ \
1562 || (c) == 0x2D /* U+05 ENQ */ \
1563 || (c) == 0x2E /* U+06 ACK */ \
1564 || (c) == 0x32 /* U+16 SYN */ \
1565 || (c) == 0x37 /* U+04 EOT */ \
1566 || (c) == 0x3C /* U+14 DC4 */ \
1567 || (c) == 0x3D /* U+15 NAK */ \
1568 || (c) == 0x3F)/* U+1A SUB */
0852beac 1569# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1570# else /* isASCII is already defined for ASCII platforms, so can use that to
1571 define isCNTRL */
1572# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1573# endif
1574
3ffc8c70 1575 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1576 * during debugging, and it seems best to keep them. We may be called
1577 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1578 * do anything anyway, so make it not a problem */
1579# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1580# define NATIVE_TO_LATIN1(ch) (ch)
1581# endif
3ded5eb0
KW
1582# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1583# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1584# define isBLANK_L1(c) (isBLANK_A(c) \
1585 || (FITS_IN_8_BITS(c) \
1586 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1587# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1588# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1589# define isLOWER_L1(c) (isLOWER_A(c) \
1590 || (FITS_IN_8_BITS(c) \
ae683a5f 1591 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1592 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1593 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1594 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1595 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1596# define isPRINT_L1(c) (isPRINT_A(c) \
1597 || (FITS_IN_8_BITS(c) \
1598 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1599# define isPUNCT_L1(c) (isPUNCT_A(c) \
1600 || (FITS_IN_8_BITS(c) \
ae683a5f 1601 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1602 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1603 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1604 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1605 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1606 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1607 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1608# define isSPACE_L1(c) (isSPACE_A(c) \
1609 || (FITS_IN_8_BITS(c) \
ae683a5f 1610 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1611 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1612# define isUPPER_L1(c) (isUPPER_A(c) \
1613 || (FITS_IN_8_BITS(c) \
182c4ace
KW
1614 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1615 0xC0, 0xDE) \
3ded5eb0
KW
1616 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1617# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1618# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1619# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1620 || isBLANK_L1(c) \
1621 || (c) == '-' \
1622 || (c) == '(' \
1623 || (c) == ')')
1624 /* The following are not fully accurate in the above-ASCII range. I (khw)
1625 * don't think it's necessary to be so for the purposes where this gets
1626 * compiled */
1627# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1628# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1629
1630 /* And these aren't accurate at all. They are useful only for above
1631 * Latin1, which utilities and bootstrapping don't deal with */
1632# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1633# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1634# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1635
1636 /* Many of the macros later in this file are defined in terms of these. By
1637 * implementing them with a function, which converts the class number into
1638 * a call to the desired macro, all of the later ones work. However, that
1639 * function won't be actually defined when building a utility program (no
1640 * perl.h), and so a compiler error will be generated if one is attempted
1641 * to be used. And the above-Latin1 code points require Unicode tables to
1642 * be present, something unlikely to be the case when bootstrapping */
1643# define _generic_isCC(c, classnum) \
1644 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1645# define _generic_isCC_A(c, classnum) \
1646 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1647#endif /* End of no perl.h H_PERL */
8a58bdcf 1648
e66b99e9
KW
1649#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1650#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1651#define isASCII_A(c) isASCII(c)
1652#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1653#define isBLANK(c) isBLANK_A(c)
1654#define isCNTRL(c) isCNTRL_A(c)
1655#define isDIGIT(c) isDIGIT_A(c)
1656#define isGRAPH(c) isGRAPH_A(c)
1657#define isIDFIRST(c) isIDFIRST_A(c)
1658#define isLOWER(c) isLOWER_A(c)
1659#define isPRINT(c) isPRINT_A(c)
779cf272 1660#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1661#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1662#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1663#define isPUNCT(c) isPUNCT_A(c)
1664#define isSPACE(c) isSPACE_A(c)
1665#define isUPPER(c) isUPPER_A(c)
1666#define isWORDCHAR(c) isWORDCHAR_A(c)
1667#define isXDIGIT(c) isXDIGIT_A(c)
1668
1669/* ASCII casing. These could also be written as
1670 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1671 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1672 which uses table lookup and mask instead of subtraction. (This would
c5e9991e
KW
1673 work because the _MOD does not apply in the ASCII range).
1674
1675 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1676 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1677 there non-ASCII invariants, and all of them are controls.) */
68067e4e
DM
1678#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1679#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1680
25200305
KW
1681/* In the ASCII range, these are equivalent to what they're here defined to be.
1682 * But by creating these definitions, other code doesn't have to be aware of
c5e9991e
KW
1683 * this detail. Actually this works for all UTF-8 invariants, not just the
1684 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
25200305 1685#define toFOLD(c) toLOWER(c)
25200305
KW
1686#define toTITLE(c) toUPPER(c)
1687
c753c8d3
KW
1688#define toLOWER_A(c) toLOWER(c)
1689#define toUPPER_A(c) toUPPER(c)
25200305
KW
1690#define toFOLD_A(c) toFOLD(c)
1691#define toTITLE_A(c) toTITLE(c)
1a0901db 1692
4650c663 1693/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1694#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1695 ? (c) \
f4cd282c 1696 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1697#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1698
1a0901db 1699/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1700 * all mapped to one of them, and these need special handling; returns the
1701 * input itself if is out-of-range */
b2bf251f 1702#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1703 ? (c) \
f4cd282c 1704 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1705#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1706
beab9ebe
KW
1707/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1708
1709/* For internal core Perl use only: the base macro for defining macros like
1710 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1711 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1712 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1713 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1714 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1715 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1716 * code points under locale rules, use the macros based on _generic_LC_uvchr
1717 * instead of this. */
beab9ebe
KW
1718#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1719 (! FITS_IN_8_BITS(c) \
1720 ? 0 \
31f05a37
KW
1721 : IN_UTF8_CTYPE_LOCALE \
1722 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1723 : cBOOL(non_utf8))
1724
1725/* For internal core Perl use only: a helper macro for defining macros like
1726 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1727 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1728 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1729 * _generic_LC_base for more info */
1730#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1731 _generic_LC_base(c,utf8_locale_classnum, \
1732 non_utf8_func( (_LC_CAST) (c)))
1733
1734/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1735 * 'c' is the platform's native underscore character */
1736#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1737 _generic_LC_base(c, utf8_locale_classnum, \
1738 (non_utf8_func( (_LC_CAST) (c)) \
1739 || (char)(c) == '_'))
1740
1741/* These next three are also for internal core Perl use only: case-change
247985d4
KW
1742 * helper macros. The reason for using the PL_latin arrays is in case the
1743 * system function is defective; it ensures uniform results that conform to the
b257a28c 1744 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
beab9ebe
KW
1745#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1746 ? (c) \
31f05a37
KW
1747 : (IN_UTF8_CTYPE_LOCALE) \
1748 ? PL_latin1_lc[ (U8) (c) ] \
5a10328c 1749 : (cast)function((cast)(c)))
beab9ebe 1750
31f05a37
KW
1751/* Note that the result can be larger than a byte in a UTF-8 locale. It
1752 * returns a single value, so can't adequately return the upper case of LATIN
1753 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1754 * values "SS"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1755 * otherwise returns its input. It does not handle the anomalies in UTF-8
1756 * Turkic locales. */
beab9ebe
KW
1757#define _generic_toUPPER_LC(c, function, cast) \
1758 (! FITS_IN_8_BITS(c) \
1759 ? (c) \
31f05a37 1760 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1761 ? (cast)function((cast)(c)) \
31f05a37
KW
1762 : ((((U8)(c)) == MICRO_SIGN) \
1763 ? GREEK_CAPITAL_LETTER_MU \
1764 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1765 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1766 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1767 ? (__ASSERT_(0) (c)) \
1768 : PL_mod_latin1_uc[ (U8) (c) ])))))
1769
1770/* Note that the result can be larger than a byte in a UTF-8 locale. It
1771 * returns a single value, so can't adequately return the fold case of LATIN
1772 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1773 * values "ss"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1774 * otherwise returns its input. It does not handle the anomalies in UTF-8
1775 * Turkic locales */
beab9ebe 1776#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1777 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1778 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1779 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1780 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1781 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1782
84061b6a 1783/* Use the libc versions for these if available. */
f05550c0 1784#if defined(HAS_ISASCII)
84061b6a
KW
1785# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1786#else
1787# define isASCII_LC(c) isASCII(c)
1788#endif
1789
f05550c0 1790#if defined(HAS_ISBLANK)
84061b6a 1791# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1792#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1793# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1794#endif
1795
f05550c0 1796#define _LC_CAST U8
bbce6d69 1797
f05550c0 1798#ifdef WIN32
375f5f06
KW
1799 /* The Windows functions don't bother to follow the POSIX standard, which
1800 * for example says that something can't both be a printable and a control.
1801 * But Windows treats the \t control as a printable, and does such things
1802 * as making superscripts into both digits and punctuation. This tames
1803 * these flaws by assuming that the definitions of both controls and space
1804 * are correct, and then making sure that other definitions don't have
1805 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1806 * Not all possible weirdnesses are checked for, just the ones that were
1807 * detected on actual Microsoft code pages */
1808
b7d90381
KW
1809# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1810# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1811
1812# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1813 && isALPHANUMERIC_LC(c))
1814# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1815 ! isPUNCT_LC(c))
1816# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1817 isALPHANUMERIC_LC(c))
1818# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1819# define isIDFIRST_LC(c) (((c) == '_') \
1820 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1821# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1822# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1823# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1824# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1825# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1826# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1827 && isALPHANUMERIC_LC(c))
f05550c0
BF
1828
1829# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1830# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1831# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1832
1833#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1834 /* For most other platforms */
beab9ebe 1835
f05550c0
BF
1836# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1837# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1838# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1839# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1840# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1841# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1842# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1843# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1844# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1845# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1846# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1847# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1848# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1849
1850
1851# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1852# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1853# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1854
1855#else /* The final fallback position */
1856
b7d90381
KW
1857# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1858# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1859# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1860# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1861# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1862# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1863# define isLOWER_LC(c) (isascii(c) && islower(c))
1864# define isPRINT_LC(c) (isascii(c) && isprint(c))
1865# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1866# define isSPACE_LC(c) (isascii(c) && isspace(c))
1867# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1868# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1869# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1870
1871# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1872# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1873# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1874
f05550c0 1875#endif
55204971 1876
eba68aa0
KW
1877#define isIDCONT(c) isWORDCHAR(c)
1878#define isIDCONT_A(c) isWORDCHAR_A(c)
1879#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1880#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1881#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1882
4650c663 1883/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1884 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1885 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1886 * classes where there is a macro or function 'above_latin1' that takes the
1887 * single argument 'c' and returns the desired value. These exist for those
2366ba44
KW
1888 * classes which have simple definitions, avoiding the overhead of an inversion
1889 * list binary search. _generic_invlist_uvchr() can be used
4650c663 1890 * for classes where that overhead is faster than a direct lookup.
d0da05db 1891 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1892 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1893 * have duplicate checks here, so could create versions of the macros that
1894 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1895
1896/* Note that all ignore 'use bytes' */
1e222e4f
KW
1897#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1898 ? _generic_isCC(c, classnum) \
cd500f2f 1899 : above_latin1(c))
2366ba44 1900#define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
1e222e4f 1901 ? _generic_isCC(c, classnum) \
922e8cb4 1902 : _is_uni_FOO(classnum, c))
2366ba44
KW
1903#define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
1904#define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
d0da05db
KW
1905#define isASCII_uvchr(c) isASCII(c)
1906#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1907#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2366ba44
KW
1908#define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
1909#define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1910#define isIDCONT_uvchr(c) \
1911 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1912#define isIDFIRST_uvchr(c) \
1913 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2366ba44
KW
1914#define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
1915#define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
d0da05db 1916
2366ba44 1917#define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
d0da05db
KW
1918#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1919#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1920
2366ba44 1921#define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
d0da05db 1922#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2366ba44 1923#define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
d0da05db
KW
1924#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1925
1926#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1927#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1928#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1929#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1930
1931/* For backwards compatibility, even though '_uni' should mean official Unicode
1932 * code points, in Perl it means native for those below 256 */
1933#define isALPHA_uni(c) isALPHA_uvchr(c)
1934#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1935#define isASCII_uni(c) isASCII_uvchr(c)
1936#define isBLANK_uni(c) isBLANK_uvchr(c)
1937#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1938#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1939#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1940#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1941#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1942#define isLOWER_uni(c) isLOWER_uvchr(c)
1943#define isPRINT_uni(c) isPRINT_uvchr(c)
1944#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1945#define isSPACE_uni(c) isSPACE_uvchr(c)
1946#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1947#define isUPPER_uni(c) isUPPER_uvchr(c)
1948#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1949#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1950#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1951#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1952#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1953#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1954#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1955
4650c663
KW
1956/* For internal core Perl use only: the base macros for defining macros like
1957 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1958 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1959 * simple class definitions; the other for more complex. These are like
d0da05db 1960 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1961#define _generic_LC_uvchr(latin1, above_latin1, c) \
1962 (c < 256 ? latin1(c) : above_latin1(c))
2366ba44 1963#define _generic_LC_invlist_uvchr(latin1, classnum, c) \
cd500f2f
KW
1964 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1965
2366ba44
KW
1966#define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
1967#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
cd500f2f 1968 _CC_ALPHANUMERIC, c)
b7d90381
KW
1969#define isASCII_LC_uvchr(c) isASCII_LC(c)
1970#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1971 is_HORIZWS_cp_high, c)
feeab5a9 1972#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2366ba44
KW
1973#define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1974#define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1975#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1976 _is_uni_perl_idcont, c)
b7d90381 1977#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f 1978 _is_uni_perl_idstart, c)
2366ba44
KW
1979#define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
1980#define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1981#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2366ba44 1982#define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1983#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1984 is_XPERLSPACE_cp_high, c)
2366ba44
KW
1985#define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
1986#define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
cd500f2f 1987 _CC_WORDCHAR, c)
b7d90381
KW
1988#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1989 is_XDIGIT_cp_high, c)
e712593e 1990
b7d90381 1991#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1992
da8c1a98
KW
1993/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
1994 * they don't otherwise go out of their way to look for malformed UTF-8. If
1995 * they can return accurate results without knowing if the input is otherwise
1996 * malformed, they do so. For example isASCII is accurate in spite of any
1997 * non-length malformations because it looks only at a single byte. Likewise
1998 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
1999 * variant ones return FALSE. But, if the input has to be well-formed in order
2000 * for the results to be accurate, the macros will test and if malformed will
2001 * call a routine to die
2002 *
2003 * Except for toke.c, the macros do assume that e > p, asserting that on
2004 * DEBUGGING builds. Much code that calls these depends on this being true,
2005 * for other reasons. toke.c is treated specially as using the regular
2006 * assertion breaks it in many ways. All strings that these operate on there
2007 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2008 * bunch of code in toke.c assumes that this is true, so the assertion allows
2009 * for that */
2010#ifdef PERL_IN_TOKE_C
2011# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2012#else
2013# define _utf8_safe_assert(p,e) ((e) > (p))
2014#endif
2015
2016#define _generic_utf8_safe(classnum, p, e, above_latin1) \
c81b3562
KW
2017 ((! _utf8_safe_assert(p, e)) \
2018 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2019 : (UTF8_IS_INVARIANT(*(p))) \
da8c1a98
KW
2020 ? _generic_isCC(*(p), classnum) \
2021 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2022 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2023 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2024 classnum) \
2025 : (_force_out_malformed_utf8_message( \
2026 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2027 : above_latin1))
b7d90381
KW
2028/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2029 * 'above_latin1' can be a macro */
da8c1a98
KW
2030#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2031 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2366ba44 2032#define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2033 _generic_utf8_safe(classnum, p, e, \
2034 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2035 ? (_force_out_malformed_utf8_message( \
2036 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2037 : above_latin1(p)))
2366ba44
KW
2038/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2039 * 'above_latin1' parameter */
2040#define _generic_invlist_utf8_safe(classnum, p, e) \
2041 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
922e8cb4 2042
cc8ab7c0 2043/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
2044 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2045 * class is TRUE for. Hence it can skip the tests for this range.
2046 * 'above_latin1' should include its arguments */
da8c1a98
KW
2047#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2048 (__ASSERT_(_utf8_safe_assert(p, e)) \
2049 (UTF8_IS_INVARIANT(*(p))) \
2050 ? _generic_isCC(*(p), classnum) \
2051 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2052 ? 0 /* Note that doesn't check validity for latin1 */ \
2053 : above_latin1)
2054
84238efa 2055
059703b0
KW
2056#define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2057#define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2058#define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2059#define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2060#define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2061#define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2062#define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2063#define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2064#define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2065#define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2066#define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2067#define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2068#define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2069#define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2070#define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2071#define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2072#define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2073#define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
e8fa43e2 2074
2366ba44 2075#define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
da8c1a98 2076#define isALPHANUMERIC_utf8_safe(p, e) \
2366ba44 2077 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
da8c1a98
KW
2078#define isASCII_utf8_safe(p, e) \
2079 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2080 * works */ \
2081 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2082#define isBLANK_utf8_safe(p, e) \
2366ba44 2083 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
da8c1a98 2084
e8fa43e2
KW
2085#ifdef EBCDIC
2086 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2087 * more efficient macro instead of the more general one */
da8c1a98 2088# define isCNTRL_utf8_safe(p, e) \
56d02b8c 2089 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 2090#else
da8c1a98 2091# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
2092#endif
2093
da8c1a98
KW
2094#define isDIGIT_utf8_safe(p, e) \
2095 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2366ba44
KW
2096 _is_utf8_FOO(_CC_DIGIT, p, e))
2097#define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
da8c1a98 2098#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
dd1a3ba7 2099 _is_utf8_perl_idcont, p, e)
e5dcd934 2100
c11ff943
KW
2101/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2102 * IDFIRST is an alnum. See
8034715d 2103 * https://github.com/Perl/perl5/issues/10275 for more detail than you
f91dcd13
KW
2104 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2105 * This used to be not the XID version, but we decided to go with the more
2106 * modern Unicode definition */
da8c1a98
KW
2107#define isIDFIRST_utf8_safe(p, e) \
2108 _generic_func_utf8_safe(_CC_IDFIRST, \
dd1a3ba7 2109 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
da8c1a98 2110
2366ba44
KW
2111#define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2112#define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
da8c1a98 2113#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2366ba44 2114#define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
da8c1a98 2115#define isSPACE_utf8_safe(p, e) \
2366ba44
KW
2116 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2117#define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
da8c1a98 2118#define isVERTWS_utf8_safe(p, e) \
2366ba44 2119 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
da8c1a98 2120#define isWORDCHAR_utf8_safe(p, e) \
2366ba44 2121 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
da8c1a98
KW
2122#define isXDIGIT_utf8_safe(p, e) \
2123 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2124 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2125 ? (_force_out_malformed_utf8_message( \
2126 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2127 : is_XDIGIT_high(p)))
a0ed51b3 2128
059703b0
KW
2129#define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2130#define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2131#define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2132#define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2e8adce6 2133
567b353c 2134/* For internal core use only, subject to change */
059703b0
KW
2135#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2136#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2137#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2138#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
a1a5ec35
KW
2139
2140#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2141#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2142#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2143#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 2144
059703b0
KW
2145#define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2146#define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2147#define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2148#define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2149#define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2150#define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2151#define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2152#define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2153#define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2154#define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2155#define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2156#define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2157#define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2158#define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2159#define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2160#define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2161#define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
34aeb2e9 2162
da8c1a98
KW
2163/* For internal core Perl use only: the base macros for defining macros like
2164 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2165 * point in 'p' is within the 0-255 range, it uses locale rules from the
2166 * passed-in 'macro' parameter */
2167#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2168 (__ASSERT_(_utf8_safe_assert(p, e)) \
2169 (UTF8_IS_INVARIANT(*(p))) \
2170 ? macro(*(p)) \
2171 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2172 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2173 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2174 : (_force_out_malformed_utf8_message( \
2175 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2176 : above_latin1))
2177
2366ba44
KW
2178#define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2179 _generic_LC_utf8_safe(macro, p, e, \
2180 _is_utf8_FOO(classnum, p, e))
da8c1a98
KW
2181
2182#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2183 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2184
2366ba44 2185#define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2186 _generic_LC_utf8_safe(classnum, p, e, \
2187 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2188 ? (_force_out_malformed_utf8_message( \
2189 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2190 : above_latin1(p)))
2191
2192#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2366ba44 2193 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
da8c1a98
KW
2194 _CC_ALPHANUMERIC, p, e)
2195#define isALPHA_LC_utf8_safe(p, e) \
2366ba44 2196 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
da8c1a98
KW
2197#define isASCII_LC_utf8_safe(p, e) \
2198 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2199#define isBLANK_LC_utf8_safe(p, e) \
2366ba44 2200 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
da8c1a98
KW
2201#define isCNTRL_LC_utf8_safe(p, e) \
2202 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2203#define isDIGIT_LC_utf8_safe(p, e) \
2366ba44 2204 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
da8c1a98 2205#define isGRAPH_LC_utf8_safe(p, e) \
2366ba44 2206 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
da8c1a98
KW
2207#define isIDCONT_LC_utf8_safe(p, e) \
2208 _generic_LC_func_utf8_safe(isIDCONT_LC, \
dd1a3ba7 2209 _is_utf8_perl_idcont, p, e)
da8c1a98
KW
2210#define isIDFIRST_LC_utf8_safe(p, e) \
2211 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
dd1a3ba7 2212 _is_utf8_perl_idstart, p, e)
da8c1a98 2213#define isLOWER_LC_utf8_safe(p, e) \
2366ba44 2214 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
da8c1a98 2215#define isPRINT_LC_utf8_safe(p, e) \
2366ba44 2216 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
da8c1a98
KW
2217#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2218#define isPUNCT_LC_utf8_safe(p, e) \
2366ba44 2219 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
da8c1a98 2220#define isSPACE_LC_utf8_safe(p, e) \
2366ba44 2221 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
da8c1a98 2222#define isUPPER_LC_utf8_safe(p, e) \
2366ba44 2223 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
da8c1a98 2224#define isWORDCHAR_LC_utf8_safe(p, e) \
2366ba44 2225 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
da8c1a98 2226#define isXDIGIT_LC_utf8_safe(p, e) \
2366ba44 2227 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2228
fbc19f27
KW
2229/* Macros for backwards compatibility and for completeness when the ASCII and
2230 * Latin1 values are identical */
b7d90381
KW
2231#define isALPHAU(c) isALPHA_L1(c)
2232#define isDIGIT_L1(c) isDIGIT_A(c)
2233#define isOCTAL(c) isOCTAL_A(c)
2234#define isOCTAL_L1(c) isOCTAL_A(c)
2235#define isXDIGIT_L1(c) isXDIGIT_A(c)
2236#define isALNUM(c) isWORDCHAR(c)
a377c856 2237#define isALNUM_A(c) isALNUM(c)
b7d90381
KW
2238#define isALNUMU(c) isWORDCHAR_L1(c)
2239#define isALNUM_LC(c) isWORDCHAR_LC(c)
2240#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2241#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
059703b0 2242#define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
4c1d9526 2243#define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
059703b0 2244#define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
4c1d9526 2245#define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
b7d90381
KW
2246#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2247#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2248#define isALNUMC(c) isALPHANUMERIC(c)
2249#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2250#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2251#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
059703b0 2252#define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
4c1d9526
KW
2253#define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2254#define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
fbc19f27 2255
2bd1cbf6
KW
2256/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2257 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2258 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2259 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2260 * the outlier from the block that contains the other controls, just like
2261 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2262 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2263 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2264 * etc. */
2bd1cbf6 2265#ifndef EBCDIC
75763b3a 2266# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2267#else
75763b3a
KW
2268# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2269 ((isPRINT_A(c)) \
2270 ? (UNLIKELY((c) == '?') \
2271 ? QUESTION_MARK_CTRL \
2272 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2273 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2274 ? '?' \
2275 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2276#endif
bbce6d69 2277
dea28490
JJ
2278/* Line numbers are unsigned, 32 bits. */
2279typedef U32 line_t;
e5dcd934 2280#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2281
91152fc1
DG
2282/* Helpful alias for version prescan */
2283#define is_LAX_VERSION(a,b) \
2284 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2285
2286#define is_STRICT_VERSION(a,b) \
2287 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2288
2289#define BADVERSION(a,b,c) \
2290 if (b) { \
2291 *b = c; \
2292 } \
2293 return a;
8c52afec 2294
1ce77b7d
KW
2295/* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2296 * a-f) to its numeric value without using any branches. The input is
2297 * validated only by an assert() in DEBUGGING builds.
2298 *
2299 * It works by right shifting and isolating the bit that is 0 for the digits,
2300 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2301 * position, and then to the eights position. Both are added together to form
2302 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2303 * final four bits of the input to form the correct value. */
2304#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2305 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2306 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2307 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2308
2309/* The argument is a string pointer, which is advanced. */
2310#define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
95a59cab 2311
cb27eebd
KW
2312/* Converts a character known to represent an octal digit (0-7) to its numeric
2313 * value. The input is validated only by an assert() in DEBUGGING builds. In
2314 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2315#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2316
305b8651
KW
2317/* Efficiently returns a boolean as to if two native characters are equivalent
2318 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2319 * the ALPHA in the name is to remind you of that. This is asserted() in
2320 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2321 * works (on valid input) for both non- and UTF-8-encoded bytes.
2322 *
2323 * When one of the inputs is a compile-time constant and gets folded by the
2324 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2325 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2326 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2327 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2328 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2329 * just a single 0, in the bit position where the upper- and lowercase differ.
2330 * */
305b8651
KW
2331#define isALPHA_FOLD_EQ(c1, c2) \
2332 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2333 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2334#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2335
8e84507e 2336/*
ccfc67b7
JH
2337=head1 Memory Management
2338
a02a5408 2339=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2340The XSUB-writer's interface to the C C<malloc> function.
2341
596f7718 2342Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2343
c5008215
JC
2344In 5.9.3, Newx() and friends replace the older New() API, and drops
2345the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2346themselves. This aid has been superseded by a new build option,
d10b4965 2347PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2348there for use in XS modules supporting older perls.
2349
a02a5408 2350=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2351The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2352cast. See also C<L</Newx>>.
954c1994 2353
596f7718 2354Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2355
a02a5408 2356=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2357The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2358memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2359
596f7718 2360Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2361
954c1994
GS
2362=for apidoc Am|void|Renew|void* ptr|int nitems|type
2363The XSUB-writer's interface to the C C<realloc> function.
2364
596f7718 2365Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2366
954c1994
GS
2367=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2368The XSUB-writer's interface to the C C<realloc> function, with
2369cast.
2370
596f7718 2371Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2372
49b8b560 2373=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2374The XSUB-writer's interface to the C C<free> function.
2375
596f7718 2376This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2377
954c1994
GS
2378=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2379The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2380source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2381C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2382
e90e2364 2383=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2384Like C<Move> but returns C<dest>. Useful
72d33970 2385for encouraging compilers to tail-call
e90e2364
NC
2386optimise.
2387
954c1994
GS
2388=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2389The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2390source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2391C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2392
e90e2364
NC
2393=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2394
796b6530 2395Like C<Copy> but returns C<dest>. Useful
72d33970 2396for encouraging compilers to tail-call
e90e2364
NC
2397optimise.
2398
954c1994
GS
2399=for apidoc Am|void|Zero|void* dest|int nitems|type
2400
2401The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2402destination, C<nitems> is the number of items, and C<type> is the type.
2403
e90e2364
NC
2404=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2405
72d33970
FC
2406Like C<Zero> but returns dest. Useful
2407for encouraging compilers to tail-call
e90e2364
NC
2408optimise.
2409
da5d8dbb 2410=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2411This is an architecture-independent macro to copy one structure to another.
954c1994 2412
7e337ee0
JH
2413=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2414
2415Fill up memory with a byte pattern (a byte repeated over and over
2416again) that hopefully catches attempts to access uninitialized memory.
2417
2418=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2419
2420PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2421
1c12ffb4 2422=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2423
2424PoisonWith(0xEF) for catching access to freed memory.
2425
9965345d
JH
2426=for apidoc Am|void|Poison|void* dest|int nitems|type
2427
7e337ee0 2428PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2429
2430=cut */
954c1994 2431
561b68a9
SH
2432/* Maintained for backwards-compatibility only. Use newSV() instead. */
2433#ifndef PERL_CORE
ff06c60c 2434#define NEWSV(x,len) newSV(len)
561b68a9 2435#endif
ff06c60c 2436
b7112dce 2437#define MEM_SIZE_MAX ((MEM_SIZE)-1)
19a94d75 2438
a500027b 2439#define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
e6bdf523 2440
27d5b266 2441#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2442
2443/* This expression will be constant-folded at compile time. It checks
2444 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2445 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2446 * As well as avoiding the need for a run-time check in some cases, it's
2447 * designed to avoid compiler warnings like:
2448 * comparison is always false due to limited range of data type
73e8ff00
DM
2449 * It's mathematically equivalent to
2450 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2451 */
2452
2453# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
445198b9
LM
2454 ( sizeof(MEM_SIZE) < sizeof(n) \
2455 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 2456
88f9f128 2457/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2458 * compiler warnings. We *want* to write the expression as
2459 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2460 * (for some compile-time constant C), but even when the LHS
2461 * constant-folds to false at compile-time, g++ insists on emitting
2462 * warnings about the RHS (e.g. "comparison is always false"), so instead
2463 * we write it as
e6bdf523 2464 *
d98e5cde 2465 * (cond ? n : X) > C
88f9f128 2466 *
d98e5cde
DM
2467 * where X is a constant with X > C always false. Choosing a value for X
2468 * is tricky. If 0, some compilers will complain about 0 > C always being
2469 * false; if 1, Coverity complains when n happens to be the constant value
2470 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2471 * for X and hope that nothing else whines.
e6bdf523
DM
2472 */
2473
2474# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2475 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2476 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2477
2478# define MEM_WRAP_CHECK(n,t) \
2479 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2480 && (croak_memory_wrap(),0))
2481
2482# define MEM_WRAP_CHECK_1(n,t,a) \
2483 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2484 && (Perl_croak_nocontext("%s",(a)),0))
2485
814eedc8
DD
2486/* "a" arg must be a string literal */
2487# define MEM_WRAP_CHECK_s(n,t,a) \
2488 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2489 && (Perl_croak_nocontext("" a ""),0))
2490
8b44ba4c 2491#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2492
a500027b 2493#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
2494#else
2495
410319be
NC
2496#define MEM_WRAP_CHECK(n,t)
2497#define MEM_WRAP_CHECK_1(n,t,a)
814eedc8 2498#define MEM_WRAP_CHECK_s(n,t,a)
8b44ba4c
NC
2499#define MEM_WRAP_CHECK_(n,t)
2500
a500027b 2501#define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
27d5b266 2502
1936d2a7 2503#endif
8b44ba4c 2504
fe4f188c 2505#ifdef PERL_MEM_LOG
46c6c7e2 2506/*
9f653bb5 2507 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2508 * go through functions, which are handy for debugging breakpoints, but
2509 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2510 * line number, and C function name if available) passed in. This info can
2511 * then be used for logging the calls, for which one gets a sample
73d1d973 2512 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2513 *
46c6c7e2 2514 * Known problems:
94e892a6 2515 * - not all memory allocs get logged, only those
46c6c7e2 2516 * that go through Newx() and derivatives (while all
94e892a6 2517 * Safefrees do get logged)
46c6c7e2
JH
2518 * - __FILE__ and __LINE__ do not work everywhere
2519 * - __func__ or __FUNCTION__ even less so
2520 * - I think more goes on after the perlio frees but
2521 * the thing is that STDERR gets closed (as do all
2522 * the file descriptors)
2523 * - no deeper calling stack than the caller of the Newx()
2524 * or the kind, but do I look like a C reflection/introspection
2525 * utility to you?
2526 * - the function prototypes for the logging functions
2527 * probably should maybe be somewhere else than handy.h
2528 * - one could consider inlining (macrofying) the logging
2529 * for speed, but I am too lazy
2530 * - one could imagine recording the allocations in a hash,
2531 * (keyed by the allocation address?), and maintain that
2532 * through reallocs and frees, but how to do that without
2533 * any News() happening...?
73d1d973 2534 * - lots of -Ddefines to get useful/controllable output
b953482e 2535 * - lots of ENV reads
46c6c7e2
JH
2536 */
2537
0b0ab801 2538# ifdef PERL_CORE
73d1d973 2539# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2540enum mem_log_type {
2541 MLT_ALLOC,
2542 MLT_REALLOC,
d7a2c63c
MHM
2543 MLT_FREE,
2544 MLT_NEW_SV,
2545 MLT_DEL_SV
0b0ab801
MHM
2546};
2547# endif
12754f92 2548# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2549void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2550void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2551# endif
0b0ab801
MHM
2552# endif
2553
fe4f188c
JH
2554#endif
2555
2556#ifdef PERL_MEM_LOG
d1401ee9
MHM
2557#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2558#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2559#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2560#endif
2561
2562#ifndef MEM_LOG_ALLOC
2563#define MEM_LOG_ALLOC(n,t,a) (a)
2564#endif
2565#ifndef MEM_LOG_REALLOC
2566#define MEM_LOG_REALLOC(n,t,v,a) (a)
2567#endif
2568#ifndef MEM_LOG_FREE
2569#define MEM_LOG_FREE(a) (a)
2570#endif
2571
d1401ee9
MHM
2572#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2573#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2574#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2575
2576#ifndef PERL_CORE
a02a5408
JC
2577/* pre 5.9.x compatibility */
2578#define New(x,v,n,t) Newx(v,n,t)
2579#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2580#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2581#endif
a02a5408 2582
ff68c719 2583#define Renew(v,n,t) \
d1401ee9 2584 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2585#define Renewc(v,n,t,c) \
d1401ee9 2586 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2587
2588#ifdef PERL_POISON
2589#define Safefree(d) \
06c0cc96 2590 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2591#else
fe4f188c 2592#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2593#endif
55497cff 2594
dbb57106
YO
2595/* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2596 * as it handles cases like constant string arguments without throwing warnings *
2597 * the cast is required, as is the inequality check, to avoid warnings */
45908e4d 2598#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2599
45908e4d
YO
2600
2601#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)))
2602#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)))
2603#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2604
bdd1531d 2605/* Like above, but returns a pointer to 'd' */
45908e4d
YO
2606#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)))
2607#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 2608#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2609
7e337ee0
JH
2610#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2611#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2612#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2613#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2614
caa674f3
DD
2615#ifdef PERL_POISON
2616# define PERL_POISON_EXPR(x) x
2617#else
2618# define PERL_POISON_EXPR(x)
2619#endif
2620
ff68c719 2621#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2622
1b7e2294
KW
2623/*
2624=head1 Handy Values
2625
2626=for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2627
2628Returns the number of elements in the input C array (so you want your
2629zero-based indices to be less than but not equal to).
2630
2631=for apidoc Am|void *|C_ARRAY_END|void *a
2632
2633Returns a pointer to one element past the final element of the input C array.
2634
2635=cut
2636
2637C_ARRAY_END is one past the last: half-open/half-closed range, not
2638last-inclusive range.
2639*/
622913ab 2640#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2641#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2642
2cc61e15
DD
2643#ifdef NEED_VA_COPY
2644# ifdef va_copy
2645# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2646# elif defined(__va_copy)
2647# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2648# else
07798b17 2649# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2650# endif
2651#endif
2652
472d47bc
SB
2653/* convenience debug macros */
2654#ifdef USE_ITHREADS
2655#define pTHX_FORMAT "Perl interpreter: 0x%p"
2656#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2657#define pTHX_VALUE_ (void *)my_perl,
2658#define pTHX_VALUE (void *)my_perl
2659#define pTHX__VALUE_ ,(void *)my_perl,
2660#define pTHX__VALUE ,(void *)my_perl
472d47bc 2661#else
3609ea0d 2662#define pTHX_FORMAT
472d47bc 2663#define pTHX__FORMAT
3609ea0d 2664#define pTHX_VALUE_
472d47bc 2665#define pTHX_VALUE
3609ea0d 2666#define pTHX__VALUE_
472d47bc
SB
2667#define pTHX__VALUE
2668#endif /* USE_ITHREADS */
3609ea0d 2669
2acdbac1
NC
2670/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2671 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2672 CPAN::Unpack show any users outside the core. */
2673#ifdef PERL_CORE
dc6e8de0
A
2674# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2675 "Use of " s " is deprecated")
c9680906
A
2676# define deprecate_disappears_in(when,message) \
2677 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2678 message ", and will disappear in Perl " when)
ac641426
A
2679# define deprecate_fatal_in(when,message) \
2680 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2681 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2682#endif
2683
dfff4baf
BF
2684/* Internal macros to deal with gids and uids */
2685#ifdef PERL_CORE
2686
2687# if Uid_t_size > IVSIZE
2688# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2689# define SvUID(sv) SvNV(sv)
07798b17
AC
2690# elif Uid_t_sign <= 0
2691# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2692# define SvUID(sv) SvIV(sv)
dfff4baf 2693# else
07798b17
AC
2694# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2695# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2696# endif /* Uid_t_size */
2697
2698# if Gid_t_size > IVSIZE
2699# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2700# define SvGID(sv) SvNV(sv)
07798b17
AC
2701# elif Gid_t_sign <= 0
2702# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2703# define SvGID(sv) SvIV(sv)
dfff4baf 2704# else
07798b17
AC
2705# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2706# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2707# endif /* Gid_t_size */
2708
2709#endif
2710
6a5bc5ac 2711#endif /* PERL_HANDY_H_ */
9d745869 2712
e9a8c099 2713/*
14d04a33 2714 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2715 */