This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Document mini_mktime
[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/*
51b56f5c 21=for apidoc_section String Handling
78342678 22=for apidoc AmnU||Nullch
72d33970
FC
23Null character pointer. (No longer available when C<PERL_CORE> is
24defined.)
2307c6d0 25
51b56f5c 26=for apidoc_section SV Handling
78342678 27=for apidoc AmnU||Nullsv
72d33970 28Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
29
30=cut
51b56f5c
KW
31
32Below are signatures of functions from config.h which can't easily be gleaned
33from it, and are very unlikely to change
34
35=for apidoc_section Signals
36=for apidoc Am|int|Sigsetjmp|jmp_buf env|int savesigs
37=for apidoc Am|void|Siglongjmp|jmp_buf env|int val
38
39=for apidoc_section Filesystem configuration values
40=for apidoc Am|void *|FILE_ptr|FILE * f
41=for apidoc Am|Size_t|FILE_cnt|FILE * f
42=for apidoc Am|void *|FILE_base|FILE * f
43=for apidoc Am|Size_t|FILE_bufsiz
44
45=for apidoc_section String Handling
46=for apidoc Am|token|CAT2|token x|token y
47=for apidoc Am|string|STRINGIFY|token x
48
49=for apidoc_section Numeric Functions
50=for apidoc Am|double|Drand01|double x
51=for apidoc Am|void|seedDrand01|Rand_seed_t x
52=for apidoc Am|char *|Gconvert|double x|Size_t n|bool t|char * b
53
54=cut
954c1994
GS
55*/
56
24792b8d
NC
57# define Nullch Null(char*)
58# define Nullfp Null(PerlIO*)
59# define Nullsv Null(SV*)
60#endif
8d063cd8 61
641d3f0b
PP
62#ifdef TRUE
63#undef TRUE
64#endif
65#ifdef FALSE
66#undef FALSE
67#endif
68#define TRUE (1)
69#define FALSE (0)
70
df87895c
KW
71/*
72=for apidoc_section SV Handling
73=for apidoc Am|void *|MUTABLE_PTR|void * p
74=for apidoc_item |AV *|MUTABLE_AV|AV * p
75=for apidoc_item |CV *|MUTABLE_CV|CV * p
76=for apidoc_item |GV *|MUTABLE_GV|GV * p
77=for apidoc_item |HV *|MUTABLE_HV|HV * p
78=for apidoc_item |IO *|MUTABLE_IO|IO * p
79=for apidoc_item |SV *|MUTABLE_SV|SV * p
80
81The C<MUTABLE_I<*>>() macros cast pointers to the types shown, in such a way
82(compiler permitting) that casting away const-ness will give a warning;
83e.g.:
84
85 const SV *sv = ...;
86 AV *av1 = (AV*)sv; <== BAD: the const has been silently
87 cast away
88 AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
89
90C<MUTABLE_PTR> is the base macro used to derive new casts. The other ones already
91built in return pointers to what their names indicate.
92
93=cut
cf3f0ffb
DM
94 */
95
b1bc3f34 96#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
6c2255e0 97# define MUTABLE_PTR(p) ({ void *p_ = (p); p_; })
b1bc3f34
NC
98#else
99# define MUTABLE_PTR(p) ((void *) (p))
100#endif
101
a062e10d 102#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 103#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 104#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 105#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 106#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 107#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 108
f789f6a4 109#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
110# include <stdbool.h>
111# ifndef HAS_BOOL
112# define HAS_BOOL 1
113# endif
114#endif
115
8e84507e 116/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
117 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
118 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 119 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
120 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
121 g++ can be identified by __GNUG__.
122 Andy Dougherty February 2000
5d94fbed 123*/
3609ea0d 124#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 125# ifndef PERL_BOOL_AS_CHAR
5d94fbed 126# ifndef HAS_BOOL
c1d22f6b 127# define HAS_BOOL 1
5d94fbed 128# endif
f789f6a4 129# endif
5d94fbed
AD
130#endif
131
132#ifndef HAS_BOOL
f789f6a4
FC
133# ifdef bool
134# undef bool
135# endif
70d5cb32 136# define bool char
c1d22f6b 137# define HAS_BOOL 1
a687059c 138#endif
0d3e774c 139
25ba28ce 140/*
51b56f5c 141=for apidoc_section Casting
25ba28ce
KW
142=for apidoc Am|bool|cBOOL|bool expr
143
144Cast-to-bool. A simple S<C<(bool) I<expr>>> cast may not do the right thing:
145if C<bool> is defined as C<char>, for example, then the cast from C<int> is
146implementation-defined.
147
148C<(bool)!!(cbool)> in a ternary triggers a bug in xlc on AIX
149
150=cut
151*/
18f5643b 152#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 153
46c6c7e2 154/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
155 * XXX Should really be a Configure probe, with HAS__FUNCTION__
156 * and FUNCTION__ as results.
157 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
46c6c7e2
JH
158#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
159# define FUNCTION__ __func__
7adf2470 160#elif (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
07798b17 161# define FUNCTION__ ""
46c6c7e2 162#else
07798b17 163# define FUNCTION__ __FUNCTION__ /* Common extension. */
46c6c7e2
JH
164#endif
165
27d4fb96
PP
166/* XXX A note on the perl source internal type system. The
167 original intent was that I32 be *exactly* 32 bits.
168
169 Currently, we only guarantee that I32 is *at least* 32 bits.
170 Specifically, if int is 64 bits, then so is I32. (This is the case
171 for the Cray.) This has the advantage of meshing nicely with
172 standard library calls (where we pass an I32 and the library is
173 expecting an int), but the disadvantage that an I32 is not 32 bits.
174 Andy Dougherty August 1996
24fef2a7 175
dc45a647
MB
176 There is no guarantee that there is *any* integral type with
177 exactly 32 bits. It is perfectly legal for a system to have
178 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 179
dc45a647
MB
180 Similarly, there is no guarantee that I16 and U16 have exactly 16
181 bits.
693762b4 182
8e84507e
NIS
183 For dealing with issues that may arise from various 32/64-bit
184 systems, we will ask Configure to check out
8175356b 185
3609ea0d
JH
186 SHORTSIZE == sizeof(short)
187 INTSIZE == sizeof(int)
188 LONGSIZE == sizeof(long)
dc45a647 189 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 190 PTRSIZE == sizeof(void *)
dc45a647
MB
191 DOUBLESIZE == sizeof(double)
192 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 193
27d4fb96
PP
194*/
195
69512466
JH
196#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
197# include <inttypes.h>
dd0eed91
JH
198# ifdef INT32_MIN_BROKEN
199# undef INT32_MIN
200# define INT32_MIN (-2147483647-1)
201# endif
202# ifdef INT64_MIN_BROKEN
203# undef INT64_MIN
204# define INT64_MIN (-9223372036854775807LL-1)
205# endif
69512466
JH
206#endif
207
8175356b
JH
208typedef I8TYPE I8;
209typedef U8TYPE U8;
210typedef I16TYPE I16;
211typedef U16TYPE U16;
212typedef I32TYPE I32;
213typedef U32TYPE U32;
16d89be8 214
74b807c7 215#ifdef QUADKIND
8175356b
JH
216typedef I64TYPE I64;
217typedef U64TYPE U64;
16d89be8 218#endif
8175356b 219
d8668976 220#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 221
5ff3f7a4
GS
222/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
223 Please search CHAR_MAX in perl.h for further details. */
224#define U8_MAX UINT8_MAX
225#define U8_MIN UINT8_MIN
226
5ff3f7a4
GS
227#define I16_MAX INT16_MAX
228#define I16_MIN INT16_MIN
229#define U16_MAX UINT16_MAX
230#define U16_MIN UINT16_MIN
231
5ff3f7a4
GS
232#define I32_MAX INT32_MAX
233#define I32_MIN INT32_MIN
0e983133
GS
234#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
235# define U32_MAX UINT32_MAX
236#else
237# define U32_MAX 4294967295U
238#endif
5ff3f7a4
GS
239#define U32_MIN UINT32_MIN
240
241#else
242
5c9fa16e
KA
243/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
244 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
245#define U8_MAX PERL_UCHAR_MAX
246#define U8_MIN PERL_UCHAR_MIN
79072805 247
27d4fb96
PP
248#define I16_MAX PERL_SHORT_MAX
249#define I16_MIN PERL_SHORT_MIN
250#define U16_MAX PERL_USHORT_MAX
251#define U16_MIN PERL_USHORT_MIN
79072805 252
c4f23d77 253#if LONGSIZE > 4
27d4fb96
PP
254# define I32_MAX PERL_INT_MAX
255# define I32_MIN PERL_INT_MIN
256# define U32_MAX PERL_UINT_MAX
257# define U32_MIN PERL_UINT_MIN
79072805 258#else
27d4fb96
PP
259# define I32_MAX PERL_LONG_MAX
260# define I32_MIN PERL_LONG_MIN
261# define U32_MAX PERL_ULONG_MAX
262# define U32_MIN PERL_ULONG_MIN
79072805
LW
263#endif
264
5ff3f7a4
GS
265#endif
266
247cee9f
KW
267/* These C99 typedefs are useful sometimes for, say, loop variables whose
268 * maximum values are small, but for which speed trumps size. If we have a C99
269 * compiler, use that. Otherwise, a plain 'int' should be good enough.
270 *
271 * Restrict these to core for now until we are more certain this is a good
272 * idea. */
273#if defined(PERL_CORE) || defined(PERL_EXT)
274# ifdef I_STDINT
275 typedef int_fast8_t PERL_INT_FAST8_T;
276 typedef uint_fast8_t PERL_UINT_FAST8_T;
277 typedef int_fast16_t PERL_INT_FAST16_T;
278 typedef uint_fast16_t PERL_UINT_FAST16_T;
279# else
280 typedef int PERL_INT_FAST8_T;
281 typedef unsigned int PERL_UINT_FAST8_T;
282 typedef int PERL_INT_FAST16_T;
283 typedef unsigned int PERL_UINT_FAST16_T;
284# endif
285#endif
286
464decb6 287/* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
ab350cbd
KW
288 * anyone is grepping for it. So BIT_DIGITS gives the number of decimal digits
289 * required to represent any possible unsigned number containing N bits.
290 * TYPE_DIGITS gives the number of decimal digits required to represent any
291 * possible unsigned number of type T. */
464decb6 292#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
fc36a67e
PP
293#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
294#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
295
88794300 296/* Unused by core; should be deprecated */
ff68c719 297#define Ctl(ch) ((ch) & 037)
8d063cd8 298
98fce2a4
KW
299#if defined(PERL_CORE) || defined(PERL_EXT)
300# ifndef MIN
301# define MIN(a,b) ((a) < (b) ? (a) : (b))
302# endif
303# ifndef MAX
304# define MAX(a,b) ((a) > (b) ? (a) : (b))
305# endif
306#endif
307
84ff4fa9
KW
308/* Returns a boolean as to whether the input unsigned number is a power of 2
309 * (2**0, 2**1, etc). In other words if it has just a single bit set.
310 * If not, subtracting 1 would leave the uppermost bit set, so the & would
311 * yield non-zero */
312#if defined(PERL_CORE) || defined(PERL_EXT)
011b1419 313# define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
84ff4fa9
KW
314#endif
315
d223e1ea 316/* Returns a mask with the lowest n bits set */
fae1e72b 317#define nBIT_MASK(n) ((UINTMAX_C(1) << (n)) - 1)
d223e1ea 318
1381ccb1
KW
319/* The largest unsigned number that will fit into n bits */
320#define nBIT_UMAX(n) nBIT_MASK(n)
321
8d9433eb 322/*
51b56f5c 323=for apidoc_section Compiler directives
d23778e6 324=for apidoc Am||__ASSERT_|bool expr
8d9433eb
KW
325
326This is a helper macro to avoid preprocessor issues, replaced by nothing
327unless under DEBUGGING, where it expands to an assert of its argument,
328followed by a comma (hence the comma operator). If we just used a straight
329assert(), we would get a comma with nothing before it when not DEBUGGING.
330
331=cut
332
333We also use empty definition under Coverity since the __ASSERT__
334checks often check for things that Really Cannot Happen, and Coverity
335detects that and gets all excited. */
3e94db23 336
e7ae132e
KW
337#if defined(DEBUGGING) && !defined(__COVERITY__) \
338 && ! defined(PERL_SMALL_MACRO_BUFFER)
0f092d08
KW
339# define __ASSERT_(statement) assert(statement),
340#else
341# define __ASSERT_(statement)
342#endif
343
3fe05580 344/*
51b56f5c 345=for apidoc_section SV Handling
3fe05580 346
3bb9fd01 347=for apidoc Ama|SV*|newSVpvs|"literal string"
1568d13a 348Like C<newSVpvn>, but takes a literal string instead of a
30a15352 349string/length pair.
3fe05580 350
3bb9fd01 351=for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
1568d13a 352Like C<newSVpvn_flags>, but takes a literal string instead of
30a15352 353a string/length pair.
84bafc02 354
3bb9fd01 355=for apidoc Ama|SV*|newSVpvs_share|"literal string"
1568d13a 356Like C<newSVpvn_share>, but takes a literal string instead of
30a15352 357a string/length pair and omits the hash parameter.
3fe05580 358
3bb9fd01 359=for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
1568d13a 360Like C<sv_catpvn_flags>, but takes a literal string instead
30a15352 361of a string/length pair.
9dcc53ea 362
3bb9fd01 363=for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
1568d13a 364Like C<sv_catpvn_nomg>, but takes a literal string instead of
0c395ea5 365a string/length pair.
9dcc53ea 366
3bb9fd01 367=for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
1568d13a 368Like C<sv_catpvn>, but takes a literal string instead of a
0c395ea5 369string/length pair.
3fe05580 370
3bb9fd01 371=for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
1568d13a 372Like C<sv_catpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
373string/length pair.
374
3bb9fd01 375=for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
1568d13a 376Like C<sv_setpvn>, but takes a literal string instead of a
0c395ea5 377string/length pair.
3fe05580 378
3bb9fd01 379=for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
1568d13a 380Like C<sv_setpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
381string/length pair.
382
3bb9fd01 383=for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
1568d13a 384Like C<sv_setref_pvn>, but takes a literal string instead of
0c395ea5 385a string/length pair.
9dcc53ea 386
51b56f5c 387=for apidoc_section String Handling
3fe05580 388
3bb9fd01 389=for apidoc Ama|char*|savepvs|"literal string"
1568d13a 390Like C<savepvn>, but takes a literal string instead of a
30a15352 391string/length pair.
3fe05580 392
3bb9fd01 393=for apidoc Ama|char*|savesharedpvs|"literal string"
9dcc53ea
Z
394A version of C<savepvs()> which allocates the duplicate string in memory
395which is shared between threads.
396
51b56f5c 397=for apidoc_section GV Handling
3fe05580 398
3bb9fd01 399=for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
1568d13a 400Like C<gv_stashpvn>, but takes a literal string instead of a
0c395ea5 401string/length pair.
3fe05580 402
51b56f5c 403=for apidoc_section HV Handling
3fe05580 404
3bb9fd01 405=for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
1568d13a 406Like C<hv_fetch>, but takes a literal string instead of a
0c395ea5 407string/length pair.
3fe05580 408
3bb9fd01 409=for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
1568d13a 410Like C<hv_store>, but takes a literal string instead of a
0c395ea5 411string/length pair
3fe05580
MHM
412and omits the hash parameter.
413
70b05c7c 414=for apidoc_section Lexer interface
510966aa 415
3bb9fd01 416=for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
510966aa 417
1568d13a 418Like L</lex_stuff_pvn>, but takes a literal string instead of
0c395ea5 419a string/length pair.
510966aa 420
3fe05580
MHM
421=cut
422*/
423
a34e53fc 424/*
51b56f5c 425=for apidoc_section String Handling
2efa8cc7 426
a34e53fc
KW
427=for apidoc Amu|pair|STR_WITH_LEN|"literal string"
428
429Returns two comma separated tokens of the input literal string, and its length.
430This is convenience macro which helps out in some API calls.
431Note that it can't be used as an argument to macros or functions that under
432some configurations might be macros, which means that it requires the full
433Perl_xxx(aTHX_ ...) form for any API calls where it's used.
434
435=cut
436*/
437
a34e53fc 438#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
ba3a79e7
GA
439
440/* STR_WITH_LEN() shortcuts */
441#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
442#define newSVpvs_flags(str,flags) \
443 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 444#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
445#define sv_catpvs_flags(sv, str, flags) \
446 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
447#define sv_catpvs_nomg(sv, str) \
448 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
449#define sv_catpvs(sv, str) \
450 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
451#define sv_catpvs_mg(sv, str) \
452 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 453#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
454#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
455#define sv_setref_pvs(rv, classname, str) \
456 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 457#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
458#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
459#define gv_stashpvs(str, create) \
460 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
3752a9fe
KW
461
462#define gv_fetchpvs(namebeg, flags, sv_type) \
463 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), flags, sv_type)
464#define gv_fetchpvn(namebeg, len, flags, sv_type) \
465 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, flags, sv_type)
9dcc53ea
Z
466#define sv_catxmlpvs(dsv, str, utf8) \
467 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 468
ba3a79e7 469
510966aa
Z
470#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
471
b96d8cd9
NC
472#define get_cvs(str, flags) \
473 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 474
9b6e9510 475/* internal helpers */
4a1bbd3d 476/* Transitional */
d67f622b
N
477#ifndef PERL_VERSION_MAJOR
478# define PERL_VERSION_MAJOR PERL_REVISION
4a1bbd3d
KW
479#else
480# undef PERL_REVISION /* We don't want code to be using these */
481#endif
d67f622b
N
482#ifndef PERL_VERSION_MINOR
483# define PERL_VERSION_MINOR PERL_VERSION
4a1bbd3d
KW
484#else
485# undef PERL_VERSION
486#endif
d67f622b
N
487#ifndef PERL_VERSION_PATCH
488# define PERL_VERSION_PATCH PERL_SUBVERSION
4a1bbd3d
KW
489#else
490# undef PERL_SUBVERSION
491#endif
492
493#define PERL_JNP_TO_DECIMAL_(maJor,miNor,Patch) \
494 /* '10*' leaves room for things like alpha, beta, releases */ \
495 (10 * ((maJor) * 1000000) + ((miNor) * 1000) + (Patch))
9b6e9510 496#define PERL_DECIMAL_VERSION_ \
d67f622b
N
497 PERL_JNP_TO_DECIMAL_(PERL_VERSION_MAJOR, PERL_VERSION_MINOR, \
498 PERL_VERSION_PATCH)
9b6e9510
KW
499
500/*
51b56f5c 501=for apidoc_section Versioning
4a1bbd3d 502=for apidoc AmR|bool|PERL_VERSION_EQ|const U8 major|const U8 minor|const U8 patch
9b6e9510 503
4a1bbd3d 504Returns whether or not the perl currently being compiled has the specified
9b6e9510
KW
505relationship to the perl given by the parameters. For example,
506
507 #if PERL_VERSION_GT(5,24,2)
508 code that will only be compiled on perls after v5.24.2
509 #else
510 fallback code
511 #endif
512
513Note that this is usable in making compile-time decisions
514
515The possible comparisons are C<PERL_VERSION_EQ>, C<PERL_VERSION_NE>,
516C<PERL_VERSION_GE>, C<PERL_VERSION_GT>, C<PERL_VERSION_LE>, and
517C<PERL_VERSION_LT>.
518
4a1bbd3d
KW
519You may use the special value '*' for the final number to mean ALL possible
520values for it. Thus,
521
522 #if PERL_VERSION_EQ(5,31,'*')
523
524means all perls in the 5.31 series. And
525
526 #if PERL_VERSION_NE(5,24,'*')
527
528means all perls EXCEPT 5.24 ones. And
529
530 #if PERL_VERSION_LE(5,9,'*')
531
532is effectively
533
534 #if PERL_VERSION_LT(5,10,0)
535
536This means you don't have to think so much when converting from the existing
537deprecated C<PERL_VERSION> to using this macro:
538
539 #if PERL_VERSION <= 9
540
541becomes
542
543 #if PERL_VERSION_LE(5,9,'*')
544
545=for apidoc AmRh|bool|PERL_VERSION_NE|const U8 major|const U8 minor|const U8 patch
546=for apidoc AmRh|bool|PERL_VERSION_GE|const U8 major|const U8 minor|const U8 patch
547=for apidoc AmRh|bool|PERL_VERSION_GT|const U8 major|const U8 minor|const U8 patch
548=for apidoc AmRh|bool|PERL_VERSION_LE|const U8 major|const U8 minor|const U8 patch
549=for apidoc AmRh|bool|PERL_VERSION_LT|const U8 major|const U8 minor|const U8 patch
9b6e9510
KW
550
551=cut
552*/
553
4a1bbd3d
KW
554/* N.B. These don't work if the patch version is 42 or 92, as those are what
555 * '*' is in ASCII and EBCDIC respectively */
556# define PERL_VERSION_EQ(j,n,p) \
557 (((p) == '*') \
d67f622b
N
558 ? ( (j) == PERL_VERSION_MAJOR \
559 && (n) == PERL_VERSION_MINOR) \
4a1bbd3d
KW
560 : (PERL_DECIMAL_VERSION_ == PERL_JNP_TO_DECIMAL_(j,n,p)))
561# define PERL_VERSION_NE(j,n,p) (! PERL_VERSION_EQ(j,n,p))
562
563# define PERL_VERSION_LT(j,n,p) /* < '*' effectively means < 0 */ \
564 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
565 (n), \
566 (((p) == '*') ? 0 : p)))
567# define PERL_VERSION_GE(j,n,p) (! PERL_VERSION_LT(j,n,p))
568
569# define PERL_VERSION_LE(j,n,p) /* <= '*' effectively means < n+1 */ \
570 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
571 (((p) == '*') ? ((n)+1) : (n)), \
572 (((p) == '*') ? 0 : p)))
573# define PERL_VERSION_GT(j,n,p) (! PERL_VERSION_LE(j,n,p))
9b6e9510 574
954c1994 575/*
51b56f5c 576=for apidoc_section String Handling
ccfc67b7 577
954c1994 578=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
579Test two C<NUL>-terminated strings to see if they are different. Returns true
580or false.
954c1994
GS
581
582=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
583Test two C<NUL>-terminated strings to see if they are equal. Returns true or
584false.
954c1994
GS
585
586=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
587Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
588second, C<s2>. Returns true or false.
954c1994
GS
589
590=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
591Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
592equal to the second, C<s2>. Returns true or false.
954c1994
GS
593
594=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
595Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
596the second, C<s2>. Returns true or false.
954c1994
GS
597
598=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
599Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
600or equal to the second, C<s2>. Returns true or false.
954c1994
GS
601
602=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
603Test two C<NUL>-terminated strings to see if they are different. The C<len>
604parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
605wrapper for C<strncmp>).
606
607=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
608Test two C<NUL>-terminated strings to see if they are equal. The C<len>
609parameter indicates the number of bytes to compare. Returns true or false. (A
610wrapper for C<strncmp>).
954c1994 611
bd18bd40
KW
612=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
613Test two buffers (which may contain embedded C<NUL> characters, to see if they
614are equal. The C<len> parameter indicates the number of bytes to compare.
615Returns zero if equal, or non-zero if non-equal.
616
3bb9fd01 617=for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
618Like L</memEQ>, but the second string is a literal enclosed in double quotes,
619C<l1> gives the number of bytes in C<s1>.
620Returns zero if equal, or non-zero if non-equal.
621
bd18bd40
KW
622=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
623Test two buffers (which may contain embedded C<NUL> characters, to see if they
624are not equal. The C<len> parameter indicates the number of bytes to compare.
625Returns zero if non-equal, or non-zero if equal.
626
3bb9fd01 627=for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
628Like L</memNE>, but the second string is a literal enclosed in double quotes,
629C<l1> gives the number of bytes in C<s1>.
630Returns zero if non-equal, or zero if non-equal.
631
4aada8b9
KW
632=for apidoc Am|bool|memCHRs|"list"|char c
633Returns the position of the first occurence of the byte C<c> in the literal
634string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
635treated as unsigned char. Thus this macro can be used to determine if C<c> is
636in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
637is C<NUL> (and the set doesn't include C<NUL>).
638
954c1994 639=cut
fc169e00
KW
640
641New macros should use the following conventions for their names (which are
642based on the underlying C library functions):
643
644 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
645
646 Each has two main parameters, string-like operands that are compared
647 against each other, as specified by the macro name. Some macros may
648 additionally have one or potentially even two length parameters. If a length
649 parameter applies to both string parameters, it will be positioned third;
650 otherwise any length parameter immediately follows the string parameter it
651 applies to.
652
653 If the prefix to the name is 'str', the string parameter is a pointer to a C
654 language string. Such a string does not contain embedded NUL bytes; its
655 length may be unknown, but can be calculated by C<strlen()>, since it is
656 terminated by a NUL, which isn't included in its length.
657
a3815e44 658 The optional 'n' following 'str' means that there is a third parameter,
fc169e00
KW
659 giving the maximum number of bytes to look at in each string. Even if both
660 strings are longer than the length parameter, those extra bytes will be
661 unexamined.
662
663 The 's' suffix means that the 2nd byte string parameter is a literal C
664 double-quoted string. Its length will automatically be calculated by the
665 macro, so no length parameter will ever be needed for it.
666
667 If the prefix is 'mem', the string parameters don't have to be C strings;
668 they may contain embedded NUL bytes, do not necessarily have a terminating
669 NUL, and their lengths can be known only through other means, which in
670 practice are additional parameter(s) passed to the function. All 'mem'
671 functions have at least one length parameter. Barring any 'l' or 's' suffix,
672 there is a single length parameter, in position 3, which applies to both
673 string parameters. The 's' suffix means, as described above, that the 2nd
674 string is a literal double-quoted C string (hence its length is calculated by
675 the macro, and the length parameter to the function applies just to the first
676 string parameter, and hence is positioned just after it). An 'l' suffix
677 means that the 2nd string parameter has its own length parameter, and the
678 signature will look like memFOOl(s1, l1, s2, l2).
679
680 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
681 substring of the 1st string. 'P' if present indicates that the substring
682 must be a "proper" one in tha mathematical sense that the first one must be
683 strictly larger than the 2nd.
684
954c1994
GS
685*/
686
62946e08 687
75400963
KW
688#define strNE(s1,s2) (strcmp(s1,s2) != 0)
689#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
690#define strLT(s1,s2) (strcmp(s1,s2) < 0)
691#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
692#define strGT(s1,s2) (strcmp(s1,s2) > 0)
693#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 694
75400963
KW
695#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
696#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 697
9d3980bc
KW
698#define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
699#define memNE(s1,s2,l) (! memEQ(s1,s2,l))
36477c24 700
085b7534 701/* memEQ and memNE where second comparand is a string constant */
568a785a 702#define memEQs(s1, l, s2) \
777fa2cb 703 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 704#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 705
fdbb9a7c
KW
706/* Keep these private until we decide it was a good idea */
707#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
708
709#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
710
bdb7e3f0 711#define memBEGINs(s1, l, s2) \
30a6480c 712 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 713 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158 714#define memBEGINPs(s1, l, s2) \
30a6480c 715 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
de627158 716 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0 717#define memENDs(s1, l, s2) \
30a6480c 718 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 719 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424 720#define memENDPs(s1, l, s2) \
30a6480c 721 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
b80f8424 722 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 723#endif /* End of making macros private */
bdb7e3f0 724
062b6850
KW
725#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
726#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
727#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
728#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
729
4aada8b9
KW
730#define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
731
bbce6d69
PP
732/*
733 * Character classes.
734 *
735 * Unfortunately, the introduction of locales means that we
736 * can't trust isupper(), etc. to tell the truth. And when
737 * it comes to /\w+/ with tainting enabled, we *must* be able
738 * to trust our character classes.
739 *
740 * Therefore, the default tests in the text of Perl will be
741 * independent of locale. Any code that wants to depend on
742 * the current locale will use the tests that begin with "lc".
743 */
744
2304df62
AD
745#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
746# ifndef CTYPE256
747# define CTYPE256
748# endif
749#endif
750
954c1994 751/*
ccfc67b7 752
dcccc8ff 753=head1 Character classification
243effed
KW
754This section is about functions (really macros) that classify characters
755into types, such as punctuation versus alphabetic, etc. Most of these are
756analogous to regular expression character classes. (See
757L<perlrecharclass/POSIX Character Classes>.) There are several variants for
758each class. (Not all macros have all variants; each item below lists the
759ones valid for it.) None are affected by C<use bytes>, and only the ones
760with C<LC> in the name are affected by the current locale.
761
d713f9d9
KW
762The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
763treating it as a code point, and returns a boolean as to whether or not the
764character represented by it is (or on non-ASCII platforms, corresponds to) an
6aff1f14
KW
765ASCII character in the named class based on platform, Unicode, and Perl rules.
766If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 767
c98722a4 768Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
769with no suffix C<"_A">. This variant is used to emphasize by its name that
770only ASCII-range characters can return TRUE.
4b9734bf 771
d60679e1 772Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
773onto the platform. That is, the code points that are ASCII are unaffected,
774since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
775as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
776true when called with the code point 0xDF, which is a word character in both
4650c663 777ASCII and EBCDIC (though it represents different characters in each).
d713f9d9
KW
778If the input is a number that doesn't fit in an octet, FALSE is returned.
779(Perl's documentation uses a colloquial definition of Latin-1, to include all
780code points below 256.)
243effed 781
d713f9d9
KW
782Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
783inputs below 256, but if the code point is larger than 255, Unicode rules are
784used to determine if it is in the character class. For example,
d0da05db 785C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 786WITH MACRON in Unicode, and is a word character.
243effed 787
059703b0
KW
788Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
789but are used for UTF-8 encoded strings. The two forms are different names for
790the same thing. Each call to one of these classifies the first character of
791the string starting at C<p>. The second parameter, C<e>, points to anywhere in
792the string beyond the first character, up to one byte past the end of the
793entire string. Although both variants are identical, the suffix C<_safe> in
794one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
795provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
796C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
797some way, the program may croak, or the function may return FALSE, at the
798discretion of the implementation, and subject to change in future releases.
243effed 799
d713f9d9
KW
800Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
801but the result is based on the current locale, which is what C<LC> in the name
802stands for. If Perl can determine that the current locale is a UTF-8 locale,
803it uses the published Unicode rules; otherwise, it uses the C library function
804that gives the named classification. For example, C<isDIGIT_LC()> when not in
805a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
806returned if the input won't fit into an octet. On some platforms where the C
807library function is known to be defective, Perl changes its result to follow
808the POSIX standard's rules.
243effed 809
d713f9d9
KW
810Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
811than 256, but for larger ones it returns the Unicode classification of the code
812point.
243effed 813
059703b0
KW
814Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
815C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
816are different names for the same thing. Each call to one of these classifies
817the first character of the string starting at C<p>. The second parameter,
818C<e>, points to anywhere in the string beyond the first character, up to one
819byte past the end of the entire string. Although both variants are identical,
820the suffix C<_safe> in one name emphasizes that it will not attempt to read
821beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
822is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
823character is malformed in some way, the program may croak, or the function may
824return FALSE, at the discretion of the implementation, and subject to change in
825future releases.
ccfc67b7 826
d713f9d9 827=for apidoc Am|bool|isALPHA|int ch
fa470d81
KW
828=for apidoc_item ||isALPHA_A|int ch
829=for apidoc_item ||isALPHA_L1|int ch
830=for apidoc_item ||isALPHA_uvchr|int ch
831=for apidoc_item ||isALPHA_utf8_safe|U8 * s|U8 * end
832=for apidoc_item ||isALPHA_utf8|U8 * s|U8 * end
833=for apidoc_item ||isALPHA_LC|int ch
834=for apidoc_item ||isALPHA_LC_uvchr|int ch
835=for apidoc_item ||isALPHA_LC_utf8_safe|U8 * s| U8 *end
d713f9d9
KW
836Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
837analogous to C<m/[[:alpha:]]/>.
dcccc8ff 838See the L<top of this section|/Character classification> for an explanation of
fa470d81 839the variants.
8a58bdcf 840
f16858ed
KW
841=cut
842
843Here and below, we add the protoypes of these macros for downstream programs
844that would be interested in them, such as Devel::PPPort
845
d713f9d9 846=for apidoc Am|bool|isALPHANUMERIC|int ch
fa470d81
KW
847=for apidoc_item ||isALPHANUMERIC_A|int ch
848=for apidoc_item ||isALPHANUMERIC_L1|int ch
849=for apidoc_item ||isALPHANUMERIC_uvchr|int ch
850=for apidoc_item ||isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
851=for apidoc_item ||isALPHANUMERIC_utf8|U8 * s|U8 * end
852=for apidoc_item ||isALPHANUMERIC_LC|int ch
853=for apidoc_item ||isALPHANUMERIC_LC_uvchr|int ch
854=for apidoc_item ||isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
855=for apidoc_item ||isALNUMC|int ch
856=for apidoc_item ||isALNUMC_A|int ch
857=for apidoc_item ||isALNUMC_L1|int ch
858=for apidoc_item ||isALNUMC_LC|int ch
859=for apidoc_item ||isALNUMC_LC_uvchr|int ch
d713f9d9
KW
860Returns a boolean indicating whether the specified character is one of
861C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
dcccc8ff 862See the L<top of this section|/Character classification> for an explanation of
fa470d81 863the variants.
15861f94 864
255b632a
KW
865A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
866this corresponds to the C language alphanumeric definition). Also
867there are the variants
868C<isALNUMC_A>, C<isALNUMC_L1>
869C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
870
d713f9d9 871=for apidoc Am|bool|isASCII|int ch
fa470d81
KW
872=for apidoc_item ||isASCII_A|int ch
873=for apidoc_item ||isASCII_L1|int ch
874=for apidoc_item ||isASCII_uvchr|int ch
875=for apidoc_item ||isASCII_utf8_safe|U8 * s|U8 * end
876=for apidoc_item ||isASCII_utf8|U8 * s|U8 * end
877=for apidoc_item ||isASCII_LC|int ch
878=for apidoc_item ||isASCII_LC_uvchr|int ch
879=for apidoc_item ||isASCII_LC_utf8_safe|U8 * s| U8 *end
8a58bdcf 880Returns a boolean indicating whether the specified character is one of the 128
243effed 881characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 882On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
883character corresponds to an ASCII character. Variants C<isASCII_A()> and
884C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff 885See the L<top of this section|/Character classification> for an explanation of
fa470d81 886the variants.
059703b0
KW
887Note, however, that some platforms do not have the C library routine
888C<isascii()>. In these cases, the variants whose names contain C<LC> are the
889same as the corresponding ones without.
243effed 890
d98532ea
KW
891Also note, that because all ASCII characters are UTF-8 invariant (meaning they
892have the exact same representation (always a single byte) whether encoded in
893UTF-8 or not), C<isASCII> will give the correct results when called with any
059703b0
KW
894byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
895C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
d98532ea 896
fa470d81
KW
897=for apidoc Am|bool|isBLANK|int ch
898=for apidoc_item ||isBLANK_A|int ch
899=for apidoc_item ||isBLANK_L1|int ch
900=for apidoc_item ||isBLANK_uvchr|int ch
901=for apidoc_item ||isBLANK_utf8_safe|U8 * s|U8 * end
902=for apidoc_item ||isBLANK_utf8|U8 * s|U8 * end
903=for apidoc_item ||isBLANK_LC|int ch
904=for apidoc_item ||isBLANK_LC_uvchr|int ch
905=for apidoc_item ||isBLANK_LC_utf8_safe|U8 * s| U8 *end
243effed 906Returns a boolean indicating whether the specified character is a
6aff1f14 907character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff 908See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
909the variants.
910Note,
da8c1a98
KW
911however, that some platforms do not have the C library routine
912C<isblank()>. In these cases, the variants whose names contain C<LC> are
913the same as the corresponding ones without.
243effed 914
fa470d81
KW
915=for apidoc Am|bool|isCNTRL|int ch
916=for apidoc_item ||isCNTRL_A|int ch
917=for apidoc_item ||isCNTRL_L1|int ch
918=for apidoc_item ||isCNTRL_uvchr|int ch
919=for apidoc_item ||isCNTRL_utf8_safe|U8 * s|U8 * end
920=for apidoc_item ||isCNTRL_utf8|U8 * s|U8 * end
921=for apidoc_item ||isCNTRL_LC|int ch
922=for apidoc_item ||isCNTRL_LC_uvchr|int ch
923=for apidoc_item ||isCNTRL_LC_utf8_safe|U8 * s| U8 *end
f16858ed 924
243effed 925Returns a boolean indicating whether the specified character is a
6aff1f14 926control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff 927See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
928the variants.
929On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
930
931=for apidoc Am|bool|isDIGIT|int ch
932=for apidoc_item ||isDIGIT_A|int ch
933=for apidoc_item ||isDIGIT_L1|int ch
934=for apidoc_item ||isDIGIT_uvchr|int ch
935=for apidoc_item ||isDIGIT_utf8_safe|U8 * s|U8 * end
936=for apidoc_item ||isDIGIT_utf8|U8 * s|U8 * end
937=for apidoc_item ||isDIGIT_LC|int ch
938=for apidoc_item ||isDIGIT_LC_uvchr|int ch
939=for apidoc_item ||isDIGIT_LC_utf8_safe|U8 * s| U8 *end
940
2787a470 941Returns a boolean indicating whether the specified character is a
6aff1f14 942digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 943Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff 944See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
945the variants.
946
947=for apidoc Am|bool|isGRAPH|int ch
948=for apidoc_item ||isGRAPH_A|int ch
949=for apidoc_item ||isGRAPH_L1|int ch
950=for apidoc_item ||isGRAPH_uvchr|int ch
951=for apidoc_item ||isGRAPH_utf8_safe|U8 * s|U8 * end
952=for apidoc_item ||isGRAPH_utf8|U8 * s|U8 * end
953=for apidoc_item ||isGRAPH_LC|int ch
954=for apidoc_item ||isGRAPH_LC_uvchr|int ch
955=for apidoc_item ||isGRAPH_LC_utf8_safe|U8 * s| U8 *end
243effed 956Returns a boolean indicating whether the specified character is a
6aff1f14 957graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 958See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
959the variants.
960
961=for apidoc Am|bool|isLOWER|int ch
962=for apidoc_item ||isLOWER_A|int ch
963=for apidoc_item ||isLOWER_L1|int ch
964=for apidoc_item ||isLOWER_uvchr|int ch
965=for apidoc_item ||isLOWER_utf8_safe|U8 * s|U8 * end
966=for apidoc_item ||isLOWER_utf8|U8 * s|U8 * end
967=for apidoc_item ||isLOWER_LC|int ch
968=for apidoc_item ||isLOWER_LC_uvchr|int ch
969=for apidoc_item ||isLOWER_LC_utf8_safe|U8 * s| U8 *end
2787a470 970Returns a boolean indicating whether the specified character is a
6aff1f14 971lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff 972See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
973the variants
974
975=for apidoc Am|bool|isOCTAL|int ch
976=for apidoc_item ||isOCTAL_A|int ch
977=for apidoc_item ||isOCTAL_L1|int ch
2787a470 978Returns a boolean indicating whether the specified character is an
6aff1f14 979octal digit, [0-7].
243effed
KW
980The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
981C<isOCTAL>.
982
fa470d81
KW
983=for apidoc Am|bool|isPUNCT|int ch
984=for apidoc_item ||isPUNCT_A|int ch
985=for apidoc_item ||isPUNCT_L1|int ch
986=for apidoc_item ||isPUNCT_uvchr|int ch
987=for apidoc_item ||isPUNCT_utf8_safe|U8 * s|U8 * end
988=for apidoc_item ||isPUNCT_utf8|U8 * s|U8 * end
989=for apidoc_item ||isPUNCT_LC|int ch
990=for apidoc_item ||isPUNCT_LC_uvchr|int ch
991=for apidoc_item ||isPUNCT_LC_utf8_safe|U8 * s| U8 *end
243effed 992Returns a boolean indicating whether the specified character is a
6aff1f14
KW
993punctuation character, analogous to C<m/[[:punct:]]/>.
994Note that the definition of what is punctuation isn't as
243effed
KW
995straightforward as one might desire. See L<perlrecharclass/POSIX Character
996Classes> for details.
dcccc8ff 997See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
998the variants.
999
1000=for apidoc Am|bool|isSPACE|int ch
1001=for apidoc_item ||isSPACE_A|int ch
1002=for apidoc_item ||isSPACE_L1|int ch
1003=for apidoc_item ||isSPACE_uvchr|int ch
1004=for apidoc_item ||isSPACE_utf8_safe|U8 * s|U8 * end
1005=for apidoc_item ||isSPACE_utf8|U8 * s|U8 * end
1006=for apidoc_item ||isSPACE_LC|int ch
1007=for apidoc_item ||isSPACE_LC_uvchr|int ch
1008=for apidoc_item ||isSPACE_LC_utf8_safe|U8 * s| U8 *end
2787a470 1009Returns a boolean indicating whether the specified character is a
6aff1f14 1010whitespace character. This is analogous
398d098a 1011to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 1012this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
1013locale forms of this macro (the ones with C<LC> in their names) matched
1014precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
1015in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
1016(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff 1017See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1018the variants.
1019
1020=for apidoc Am|bool|isPSXSPC|int ch
1021=for apidoc_item ||isPSXSPC_A|int ch
1022=for apidoc_item ||isPSXSPC_L1|int ch
1023=for apidoc_item ||isPSXSPC_uvchr|int ch
1024=for apidoc_item ||isPSXSPC_utf8_safe|U8 * s|U8 * end
1025=for apidoc_item ||isPSXSPC_utf8|U8 * s|U8 * end
1026=for apidoc_item ||isPSXSPC_LC|int ch
1027=for apidoc_item ||isPSXSPC_LC_uvchr|int ch
1028=for apidoc_item ||isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
398d098a 1029(short for Posix Space)
779cf272
KW
1030Starting in 5.18, this is identical in all its forms to the
1031corresponding C<isSPACE()> macros.
398d098a
KW
1032The locale forms of this macro are identical to their corresponding
1033C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
1034non-locale forms differ from their C<isSPACE()> forms only in that the
1035C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
1036Otherwise they are identical. Thus this macro is analogous to what
1037C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 1038See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1039the variants.
1040
1041=for apidoc Am|bool|isUPPER|int ch
1042=for apidoc_item ||isUPPER_A|int ch
1043=for apidoc_item ||isUPPER_L1|int ch
1044=for apidoc_item ||isUPPER_uvchr|int ch
1045=for apidoc_item ||isUPPER_utf8_safe|U8 * s|U8 * end
1046=for apidoc_item ||isUPPER_utf8|U8 * s|U8 * end
1047=for apidoc_item ||isUPPER_LC|int ch
1048=for apidoc_item ||isUPPER_LC_uvchr|int ch
1049=for apidoc_item ||isUPPER_LC_utf8_safe|U8 * s| U8 *end
2787a470 1050Returns a boolean indicating whether the specified character is an
6aff1f14 1051uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 1052See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1053the variants.
1054
1055=for apidoc Am|bool|isPRINT|int ch
1056=for apidoc_item ||isPRINT_A|int ch
1057=for apidoc_item ||isPRINT_L1|int ch
1058=for apidoc_item ||isPRINT_uvchr|int ch
1059=for apidoc_item ||isPRINT_utf8_safe|U8 * s|U8 * end
1060=for apidoc_item ||isPRINT_utf8|U8 * s|U8 * end
1061=for apidoc_item ||isPRINT_LC|int ch
1062=for apidoc_item ||isPRINT_LC_uvchr|int ch
1063=for apidoc_item ||isPRINT_LC_utf8_safe|U8 * s| U8 *end
8eea39dd 1064Returns a boolean indicating whether the specified character is a
6aff1f14 1065printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff 1066See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1067the variants.
1068
1069=for apidoc Am|bool|isWORDCHAR|int ch
1070=for apidoc_item ||isWORDCHAR_A|int ch
1071=for apidoc_item ||isWORDCHAR_L1|int ch
1072=for apidoc_item ||isWORDCHAR_uvchr|int ch
1073=for apidoc_item ||isWORDCHAR_utf8_safe|U8 * s|U8 * end
1074=for apidoc_item ||isWORDCHAR_utf8|U8 * s|U8 * end
1075=for apidoc_item ||isWORDCHAR_LC|int ch
1076=for apidoc_item ||isWORDCHAR_LC_uvchr|int ch
1077=for apidoc_item ||isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
1078=for apidoc_item ||isALNUM|int ch
1079=for apidoc_item ||isALNUM_A|int ch
1080=for apidoc_item ||isALNUM_LC|int ch
1081=for apidoc_item ||isALNUM_LC_uvchr|int ch
243effed
KW
1082Returns a boolean indicating whether the specified character is a character
1083that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1084in a regular expression. A word character is an alphabetic character, a
1085decimal digit, a connecting punctuation character (such as an underscore), or
1086a "mark" character that attaches to one of those (like some sort of accent).
1087C<isALNUM()> is a synonym provided for backward compatibility, even though a
1088word character includes more than the standard C language meaning of
1089alphanumeric.
dcccc8ff 1090See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1091the variants.
1092C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1093C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and
1094C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
1095include the platform's native underscore.
1096
1097=for apidoc Am|bool|isXDIGIT|int ch
1098=for apidoc_item ||isXDIGIT_A|int ch
1099=for apidoc_item ||isXDIGIT_L1|int ch
1100=for apidoc_item ||isXDIGIT_uvchr|int ch
1101=for apidoc_item ||isXDIGIT_utf8_safe|U8 * s|U8 * end
1102=for apidoc_item ||isXDIGIT_utf8|U8 * s|U8 * end
1103=for apidoc_item ||isXDIGIT_LC|int ch
1104=for apidoc_item ||isXDIGIT_LC_uvchr|int ch
1105=for apidoc_item ||isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
8a58bdcf 1106Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
1107digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1108and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff 1109See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1110the variants.
1111
1112=for apidoc Am|bool|isIDFIRST|int ch
1113=for apidoc_item ||isIDFIRST_A|int ch
1114=for apidoc_item ||isIDFIRST_L1|int ch
1115=for apidoc_item ||isIDFIRST_uvchr|int ch
1116=for apidoc_item ||isIDFIRST_utf8_safe|U8 * s|U8 * end
1117=for apidoc_item ||isIDFIRST_utf8|U8 * s|U8 * end
1118=for apidoc_item ||isIDFIRST_LC|int ch
1119=for apidoc_item ||isIDFIRST_LC_uvchr|int ch
1120=for apidoc_item ||isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
3c3ecf18
KW
1121Returns a boolean indicating whether the specified character can be the first
1122character of an identifier. This is very close to, but not quite the same as
1123the official Unicode property C<XID_Start>. The difference is that this
1124returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff 1125See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1126the variants.
1127
1128=for apidoc Am|bool|isIDCONT|int ch
1129=for apidoc_item ||isIDCONT_A|int ch
1130=for apidoc_item ||isIDCONT_L1|int ch
1131=for apidoc_item ||isIDCONT_uvchr|int ch
1132=for apidoc_item ||isIDCONT_utf8_safe|U8 * s|U8 * end
1133=for apidoc_item ||isIDCONT_utf8|U8 * s|U8 * end
1134=for apidoc_item ||isIDCONT_LC|int ch
1135=for apidoc_item ||isIDCONT_LC_uvchr|int ch
1136=for apidoc_item ||isIDCONT_LC_utf8_safe|U8 * s| U8 *end
3c3ecf18
KW
1137Returns a boolean indicating whether the specified character can be the
1138second or succeeding character of an identifier. This is very close to, but
1139not quite the same as the official Unicode property C<XID_Continue>. The
1140difference is that this returns true only if the input character also matches
dcccc8ff 1141L</isWORDCHAR>. See the L<top of this section|/Character classification> for
fa470d81 1142an explanation of the variants.
f16858ed 1143
51b56f5c 1144=for apidoc_section Numeric Functions
8eea39dd 1145
95a59cab 1146=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 1147Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
1148Behaviour is only well defined when isXDIGIT(*str) is true.
1149
e7c1e6c1 1150=head1 Character case changing
21da7284
KW
1151Perl uses "full" Unicode case mappings. This means that converting a single
1152character to another case may result in a sequence of more than one character.
1153For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1154character sequence C<SS>. This presents some complications The lowercase of
1155all characters in the range 0..255 is a single character, and thus
1156C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1157return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1158an API that does allow every possible legal result to be returned.) Likewise
1159no other function that is crippled by not being able to give the correct
1160results for the full range of possible inputs has been implemented here.
e7c1e6c1 1161
d713f9d9 1162=for apidoc Am|U8|toUPPER|int ch
1f607577
KW
1163Converts the specified character to uppercase. If the input is anything but an
1164ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 1165C<toUPPER_A> is equivalent.
954c1994 1166
d0da05db
KW
1167=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1168Converts the code point C<cp> to its uppercase version, and
1169stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1170point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1171that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1172bytes since the uppercase version may be longer than the original character.
1173
1174The first code point of the uppercased version is returned
21da7284
KW
1175(but note, as explained at L<the top of this section|/Character case
1176changing>, that there may be more.)
1f607577 1177
059703b0 1178=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141 1179=for apidoc_item toUPPER_utf8_safe
a239b1e2
KW
1180Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1181extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
1182stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1183that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1184bytes since the uppercase version may be longer than the original character.
1185
1186The first code point of the uppercased version is returned
21da7284
KW
1187(but note, as explained at L<the top of this section|/Character case
1188changing>, that there may be more).
1f607577 1189
059703b0
KW
1190It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1191S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1192the UTF-8 for the input character is malformed in some way, the program may
1193croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1194of the implementation, and subject to change in future releases.
a239b1e2 1195
912c8141 1196C<toUPPER_utf8_safe> is now just a different spelling of plain C<toUPPER_utf8>
1f607577 1197
25200305
KW
1198=for apidoc Am|U8|toFOLD|U8 ch
1199Converts the specified character to foldcase. If the input is anything but an
1200ASCII uppercase character, that input character itself is returned. Variant
1201C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 1202Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 1203
d0da05db
KW
1204=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1205Converts the code point C<cp> to its foldcase version, and
1206stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1207point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1208that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1209bytes since the foldcase version may be longer than the original character.
1210
1211The first code point of the foldcased version is returned
21da7284
KW
1212(but note, as explained at L<the top of this section|/Character case
1213changing>, that there may be more).
1f607577 1214
059703b0 1215=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141 1216=for apidoc_item toFOLD_utf8_safe
a239b1e2
KW
1217Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1218extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
1219stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1220that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1221bytes since the foldcase version may be longer than the original character.
1222
1223The first code point of the foldcased version is returned
21da7284
KW
1224(but note, as explained at L<the top of this section|/Character case
1225changing>, that there may be more).
1f607577 1226
059703b0 1227It will not attempt
a239b1e2
KW
1228to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1229true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1230input character is malformed in some way, the program may croak, or the
1231function may return the REPLACEMENT CHARACTER, at the discretion of the
1232implementation, and subject to change in future releases.
1233
912c8141 1234C<toFOLD_utf8_safe> is now just a different spelling of plain C<toFOLD_utf8>
1f607577
KW
1235
1236=for apidoc Am|U8|toLOWER|U8 ch
1237Converts the specified character to lowercase. If the input is anything but an
1238ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 1239C<toLOWER_A> is equivalent.
954c1994 1240
1f607577 1241=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
1242Converts the specified Latin1 character to lowercase. The results are
1243undefined if the input doesn't fit in a byte.
1f607577
KW
1244
1245=for apidoc Am|U8|toLOWER_LC|U8 ch
1246Converts the specified character to lowercase using the current locale's rules,
1247if possible; otherwise returns the input character itself.
1248
d0da05db
KW
1249=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1250Converts the code point C<cp> to its lowercase version, and
1251stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1252point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1253that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1254bytes since the lowercase version may be longer than the original character.
1255
1256The first code point of the lowercased version is returned
21da7284
KW
1257(but note, as explained at L<the top of this section|/Character case
1258changing>, that there may be more).
1f607577 1259
bd350c85 1260=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141 1261=for apidoc_item toLOWER_utf8_safe
a239b1e2
KW
1262Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1263extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
1264stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1265that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1266bytes since the lowercase version may be longer than the original character.
1267
1268The first code point of the lowercased version is returned
21da7284
KW
1269(but note, as explained at L<the top of this section|/Character case
1270changing>, that there may be more).
059703b0
KW
1271It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1272S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1273the UTF-8 for the input character is malformed in some way, the program may
1274croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1275of the implementation, and subject to change in future releases.
1f607577 1276
912c8141 1277C<toLOWER_utf8_safe> is now just a different spelling of plain C<toLOWER_utf8>
1f607577 1278
25200305
KW
1279=for apidoc Am|U8|toTITLE|U8 ch
1280Converts the specified character to titlecase. If the input is anything but an
1281ASCII lowercase character, that input character itself is returned. Variant
b7d90381 1282C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 1283range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 1284not a concept used in locale handling, so there is no functionality for that.)
25200305 1285
d0da05db
KW
1286=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1287Converts the code point C<cp> to its titlecase version, and
1288stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1289point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1290that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1291bytes since the titlecase version may be longer than the original character.
1292
1293The first code point of the titlecased version is returned
21da7284
KW
1294(but note, as explained at L<the top of this section|/Character case
1295changing>, that there may be more).
1f607577 1296
059703b0 1297=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141
KW
1298=for apidoc_item toTITLE_utf8_safe
1299Convert the first UTF-8 encoded character in the sequence starting at C<p> and
a239b1e2 1300extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1301stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1302that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1303bytes since the titlecase version may be longer than the original character.
1304
1305The first code point of the titlecased version is returned
21da7284
KW
1306(but note, as explained at L<the top of this section|/Character case
1307changing>, that there may be more).
1f607577 1308
059703b0 1309It will not attempt
a239b1e2
KW
1310to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1311true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1312input character is malformed in some way, the program may croak, or the
1313function may return the REPLACEMENT CHARACTER, at the discretion of the
1314implementation, and subject to change in future releases.
1315
912c8141 1316C<toTITLE_utf8_safe> is now just a different spelling of plain C<toTITLE_utf8>
1f607577 1317
954c1994 1318=cut
353c9b6f 1319
d0da05db 1320XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1321really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1322and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1323
8a58bdcf 1324Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1325patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1326
954c1994
GS
1327*/
1328
8f5283f4
KW
1329/*
1330 void below because that's the best fit, and works for Devel::PPPort
51b56f5c 1331=for apidoc_section Integer configuration values
8f5283f4
KW
1332=for apidoc AmnU|void|WIDEST_UTYPE
1333
1334Yields the widest unsigned integer type on the platform, currently either
1335C<U32> or C<64>. This can be used in declarations such as
1336
1337 WIDEST_UTYPE my_uv;
1338
1339or casts
1340
1341 my_uv = (WIDEST_UTYPE) val;
1342
1343=cut
1344
1345*/
de9e2639
KW
1346#ifdef QUADKIND
1347# define WIDEST_UTYPE U64
7c062697
KW
1348#else
1349# define WIDEST_UTYPE U32
1350#endif
1351
3912bc88
KW
1352/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1353 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1354 * bits of information are lost even on a 64-bit machine, but to get the
1355 * compiler to optimize it out if possible. This is because Configure makes
1356 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1357 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1358 *
1359 * For Coverity, be always true, because otherwise Coverity thinks
1360 * it finds several expressions that are always true, independent
1361 * of operands. Well, they are, but that is kind of the point.
220c71bf 1362 */
7e75d1a1 1363#ifndef __COVERITY__
6c5b02ac
KW
1364 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1365 * pointer) */
1366#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1367 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1368#else
1369#define FITS_IN_8_BITS(c) (1)
1370#endif
cf301eb7 1371
45f4bb73 1372/* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
833b0f46 1373 * Written this way so that after optimization, only one conditional test is
76d3ad4c
KW
1374 * needed. (The NV casts stop any warnings about comparison always being true
1375 * if called with an unsigned. The cast preserves the sign, which is all we
1376 * care about.) */
1377#define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1378 __ASSERT_((NV) (n) >= 0) \
1379 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
833b0f46 1380
94250c4f
KW
1381/* Returns true if c is in the range l..u, where 'l' is non-negative
1382 * Written this way so that after optimization, only one conditional test is
4758c20d 1383 * needed. */
1eaefa6e 1384#define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
4758c20d 1385 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
4758c20d
KW
1386 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1387 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
45f4bb73 1388 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
305fe86e 1389
41f43cc2 1390#ifdef EBCDIC
b6340bd0 1391# ifndef _ALL_SOURCE
0852beac
KW
1392 /* The native libc isascii() et.al. functions return the wrong results
1393 * on at least z/OS unless this is defined. */
b6340bd0
KW
1394# error _ALL_SOURCE should probably be defined
1395# endif
41f43cc2 1396#else
0852beac
KW
1397 /* There is a simple definition of ASCII for ASCII platforms. But the
1398 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1399 * other macros below.
3f3c579d
KW
1400 *
1401 * The cast here is used instead of '(c) >= 0', because some compilers emit
1402 * a warning that that test is always true when the parameter is an
1403 * unsigned type. khw supposes that it could be written as
1404 * && ((c) == '\0' || (c) > 0)
1405 * to avoid the message, but the cast will likely avoid extra branches even
1406 * with stupid compilers.
1407 *
1408 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1409 * a pointer) */
9c903d59 1410# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1411#endif
1412
38694112
KW
1413/* Take the eight possible bit patterns of the lower 3 bits and you get the
1414 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1415 * can be ignored. If the rest match '0', we have an octal */
1416#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1417
9fb1bf9d 1418#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1419
a500dc72
KW
1420/* Character class numbers. For internal core Perl use only. The ones less
1421 * than 32 are used in PL_charclass[] and the ones up through the one that
1422 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1423 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1424 * their actual definitions are here. If that file has a name not used here,
1425 * it won't compile.
1709d539
KW
1426 *
1427 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561 1428 * frequency of their use. This gives a slight edge to exiting a loop earlier
58a3ba2c
KW
1429 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1430 * real Posix character class, and some (small) inefficiencies in regular
1431 * expression handling would be introduced by putting it in the middle of those
1432 * that are. Also, cntrl and ascii come after the others as it may be useful
1433 * to group these which have no members that match above Latin1, (or above
1434 * ASCII in the latter case) */
1435
1709d539
KW
1436# define _CC_WORDCHAR 0 /* \w and [:word:] */
1437# define _CC_DIGIT 1 /* \d and [:digit:] */
1438# define _CC_ALPHA 2 /* [:alpha:] */
1439# define _CC_LOWER 3 /* [:lower:] */
1440# define _CC_UPPER 4 /* [:upper:] */
1441# define _CC_PUNCT 5 /* [:punct:] */
1442# define _CC_PRINT 6 /* [:print:] */
15861f94 1443# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1444# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1445# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
779cf272 1446# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1447# define _CC_BLANK 11 /* [:blank:] */
1448# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1449# define _CC_CNTRL 13 /* [:cntrl:] */
1450# define _CC_ASCII 14 /* [:ascii:] */
1451# define _CC_VERTSPACE 15 /* \v */
1709d539 1452
a0947d7b
KW
1453# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1454
1709d539 1455/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1456 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1457# define _CC_IDFIRST 16
1458# define _CC_CHARNAME_CONT 17
1459# define _CC_NONLATIN1_FOLD 18
1460# define _CC_NONLATIN1_SIMPLE_FOLD 19
1461# define _CC_QUOTEMETA 20
1462# define _CC_NON_FINAL_FOLD 21
1463# define _CC_IS_IN_SOME_FOLD 22
2ae9030c
KW
1464# define _CC_BINDIGIT 23
1465# define _CC_OCTDIGIT 24
1466# define _CC_MNEMONIC_CNTRL 25
073c22b3
KW
1467
1468/* This next group is only used on EBCDIC platforms, so theoretically could be
1469 * shared with something entirely different that's only on ASCII platforms */
abb8abf6 1470# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
0654f0ab 1471/* Unused: 26-30
f4cdb42c
KW
1472 * If more bits are needed, one could add a second word for non-64bit
1473 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1474 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1475 * is used only for optimization (as of this writing), and differs in the
1476 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1477 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1478 * would just cause /i regexes which match them to run less efficiently.
1479 * Similarly the EBCDIC-only bits are used just for speed, and could be
1480 * replaced by other means */
96ac0975 1481
3a371f2f
KW
1482#if defined(PERL_CORE) || defined(PERL_EXT)
1483/* An enum version of the character class numbers, to help compilers
1484 * optimize */
1485typedef enum {
3a371f2f 1486 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1487 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1488 _CC_ENUM_ASCII = _CC_ASCII,
1489 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1490 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1491 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1492 _CC_ENUM_DIGIT = _CC_DIGIT,
1493 _CC_ENUM_GRAPH = _CC_GRAPH,
1494 _CC_ENUM_LOWER = _CC_LOWER,
1495 _CC_ENUM_PRINT = _CC_PRINT,
1496 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1497 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1498 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1499 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1500 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1501 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1502} _char_class_number;
1503#endif
1504
86f72d56 1505#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1506
6635f04f 1507START_EXTERN_C
96ac0975
NC
1508# ifdef DOINIT
1509EXTCONST U32 PL_charclass[] = {
1510# include "l1_char_class_tab.h"
1511};
1512
1513# else /* ! DOINIT */
1514EXTCONST U32 PL_charclass[];
1515# endif
6635f04f 1516END_EXTERN_C
96ac0975 1517
265c1f46 1518 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1519# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1520
1521 /* For internal core Perl use only: the base macro for defining macros like
1522 * isALPHA */
ff7ecfc3 1523# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1524 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1525
f4cdb42c
KW
1526 /* The mask for the _A versions of the macros; it just adds in the bit for
1527 * ASCII. */
1528# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1529
4650c663
KW
1530 /* For internal core Perl use only: the base macro for defining macros like
1531 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1532 * the ASCII bit are present */
b7d90381
KW
1533# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1534 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1535 == _CC_mask_A(classnum)))
f4cdb42c 1536
26c1d9d8
KW
1537/* On ASCII platforms certain classes form a single range. It's faster to
1538 * special case these. isDIGIT is a single range on all platforms */
b877c1ff
KW
1539# ifdef EBCDIC
1540# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1541# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1542# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1543# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1544# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1545# else
26c1d9d8 1546 /* By folding the upper and lowercase, we can use a single range */
b877c1ff 1547# define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
26c1d9d8 1548# define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
b877c1ff
KW
1549# define isLOWER_A(c) inRANGE(c, 'a', 'z')
1550# define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1551# define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1552# endif
15861f94 1553# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1554# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1555# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
b877c1ff 1556# define isDIGIT_A(c) inRANGE(c, '0', '9')
f4cdb42c
KW
1557# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1558# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
f4cdb42c 1559# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1560# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1561 */
d95f8b6a 1562# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1563# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1564# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1565# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1566
1567 /* continuation character for legal NAME in \N{NAME} */
1568# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1569
1570# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1571# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1572# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1573# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1574# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1575# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1576# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1577# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1578# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1579# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1580
0852beac
KW
1581# ifdef EBCDIC
1582# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1583# endif
1584
f12c0118
KW
1585 /* Participates in a single-character fold with a character above 255 */
1586# 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)))
1587
1588 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1589# 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 1590
4eeeb416 1591# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1592# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1593 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1594# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1595 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
5e6ebb12
KW
1596
1597/* is c a control character for which we have a mnemonic? */
1598# if defined(PERL_CORE) || defined(PERL_EXT)
1599# define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1600# endif
687c8d01 1601#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1602
1603 /* If we don't have perl.h, we are compiling a utility program. Below we
1604 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1605 * to it. Most are coded based on first principles. These are written to
74665a89 1606 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
182c4ace 1607# define isDIGIT_A(c) inRANGE(c, '0', '9')
0852beac 1608# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1609# define isSPACE_A(c) (isBLANK_A(c) \
1610 || (c) == '\n' \
1611 || (c) == '\r' \
1612 || (c) == '\v' \
0852beac 1613 || (c) == '\f')
74665a89
KW
1614 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1615 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1616 * performance (if optimization isn't on), and allow the same code to be
1617 * used for both platform types */
182c4ace
KW
1618# define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1619 || inRANGE((c), 'j', 'r') \
1620 || inRANGE((c), 's', 'z')
1621# define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1622 || inRANGE((c), 'J', 'R') \
1623 || inRANGE((c), 'S', 'Z')
a4d7a999
KW
1624# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1625# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1626# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1627# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
182c4ace
KW
1628# define isXDIGIT_A(c) ( isDIGIT_A(c) \
1629 || inRANGE((c), 'a', 'f') \
1630 || inRANGE((c), 'A', 'F')
74665a89
KW
1631# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1632 || (c) == '#' || (c) == '$' || (c) == '%' \
1633 || (c) == '&' || (c) == '\'' || (c) == '(' \
1634 || (c) == ')' || (c) == '*' || (c) == '+' \
1635 || (c) == ',' || (c) == '.' || (c) == '/' \
1636 || (c) == ':' || (c) == ';' || (c) == '<' \
1637 || (c) == '=' || (c) == '>' || (c) == '?' \
1638 || (c) == '@' || (c) == '[' || (c) == '\\' \
1639 || (c) == ']' || (c) == '^' || (c) == '_' \
1640 || (c) == '`' || (c) == '{' || (c) == '|' \
1641 || (c) == '}' || (c) == '~')
1642# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1643# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1644
0852beac 1645# ifdef EBCDIC
74665a89
KW
1646 /* The below is accurate for the 3 EBCDIC code pages traditionally
1647 * supported by perl. The only difference between them in the controls
1648 * is the position of \n, and that is represented symbolically below */
1649# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1650 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1651 || (c) == '\t' || (c) == '\v' \
182c4ace 1652 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
8ec0a736 1653 || (c) == 7F /* U+7F DEL */ \
182c4ace
KW
1654 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1655 DC[1-3] */ \
74665a89
KW
1656 || (c) == 0x18 /* U+18 CAN */ \
1657 || (c) == 0x19 /* U+19 EOM */ \
182c4ace 1658 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
74665a89
KW
1659 || (c) == 0x26 /* U+17 ETB */ \
1660 || (c) == 0x27 /* U+1B ESC */ \
1661 || (c) == 0x2D /* U+05 ENQ */ \
1662 || (c) == 0x2E /* U+06 ACK */ \
1663 || (c) == 0x32 /* U+16 SYN */ \
1664 || (c) == 0x37 /* U+04 EOT */ \
1665 || (c) == 0x3C /* U+14 DC4 */ \
1666 || (c) == 0x3D /* U+15 NAK */ \
1667 || (c) == 0x3F)/* U+1A SUB */
0852beac 1668# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1669# else /* isASCII is already defined for ASCII platforms, so can use that to
1670 define isCNTRL */
1671# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1672# endif
1673
3ffc8c70 1674 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1675 * during debugging, and it seems best to keep them. We may be called
1676 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1677 * do anything anyway, so make it not a problem */
1678# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1679# define NATIVE_TO_LATIN1(ch) (ch)
1680# endif
3ded5eb0
KW
1681# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1682# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1683# define isBLANK_L1(c) (isBLANK_A(c) \
1684 || (FITS_IN_8_BITS(c) \
1685 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1686# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1687# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1688# define isLOWER_L1(c) (isLOWER_A(c) \
1689 || (FITS_IN_8_BITS(c) \
ae683a5f 1690 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1691 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1692 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1693 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1694 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1695# define isPRINT_L1(c) (isPRINT_A(c) \
1696 || (FITS_IN_8_BITS(c) \
1697 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1698# define isPUNCT_L1(c) (isPUNCT_A(c) \
1699 || (FITS_IN_8_BITS(c) \
ae683a5f 1700 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1701 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1702 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1703 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1704 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1705 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1706 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1707# define isSPACE_L1(c) (isSPACE_A(c) \
1708 || (FITS_IN_8_BITS(c) \
ae683a5f 1709 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1710 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1711# define isUPPER_L1(c) (isUPPER_A(c) \
1712 || (FITS_IN_8_BITS(c) \
182c4ace
KW
1713 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1714 0xC0, 0xDE) \
3ded5eb0
KW
1715 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1716# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1717# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1718# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1719 || isBLANK_L1(c) \
1720 || (c) == '-' \
1721 || (c) == '(' \
1722 || (c) == ')')
1723 /* The following are not fully accurate in the above-ASCII range. I (khw)
1724 * don't think it's necessary to be so for the purposes where this gets
1725 * compiled */
1726# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1727# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1728
1729 /* And these aren't accurate at all. They are useful only for above
1730 * Latin1, which utilities and bootstrapping don't deal with */
1731# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1732# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1733# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1734
1735 /* Many of the macros later in this file are defined in terms of these. By
1736 * implementing them with a function, which converts the class number into
1737 * a call to the desired macro, all of the later ones work. However, that
1738 * function won't be actually defined when building a utility program (no
1739 * perl.h), and so a compiler error will be generated if one is attempted
1740 * to be used. And the above-Latin1 code points require Unicode tables to
1741 * be present, something unlikely to be the case when bootstrapping */
1742# define _generic_isCC(c, classnum) \
1743 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1744# define _generic_isCC_A(c, classnum) \
1745 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1746#endif /* End of no perl.h H_PERL */
8a58bdcf 1747
e66b99e9
KW
1748#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1749#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1750#define isASCII_A(c) isASCII(c)
1751#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1752#define isBLANK(c) isBLANK_A(c)
1753#define isCNTRL(c) isCNTRL_A(c)
1754#define isDIGIT(c) isDIGIT_A(c)
1755#define isGRAPH(c) isGRAPH_A(c)
1756#define isIDFIRST(c) isIDFIRST_A(c)
1757#define isLOWER(c) isLOWER_A(c)
1758#define isPRINT(c) isPRINT_A(c)
779cf272 1759#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1760#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1761#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1762#define isPUNCT(c) isPUNCT_A(c)
1763#define isSPACE(c) isSPACE_A(c)
1764#define isUPPER(c) isUPPER_A(c)
1765#define isWORDCHAR(c) isWORDCHAR_A(c)
1766#define isXDIGIT(c) isXDIGIT_A(c)
1767
1768/* ASCII casing. These could also be written as
1769 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1770 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1771 which uses table lookup and mask instead of subtraction. (This would
c5e9991e
KW
1772 work because the _MOD does not apply in the ASCII range).
1773
1774 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1775 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1776 there non-ASCII invariants, and all of them are controls.) */
68067e4e
DM
1777#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1778#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1779
25200305
KW
1780/* In the ASCII range, these are equivalent to what they're here defined to be.
1781 * But by creating these definitions, other code doesn't have to be aware of
c5e9991e
KW
1782 * this detail. Actually this works for all UTF-8 invariants, not just the
1783 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
25200305 1784#define toFOLD(c) toLOWER(c)
25200305
KW
1785#define toTITLE(c) toUPPER(c)
1786
c753c8d3
KW
1787#define toLOWER_A(c) toLOWER(c)
1788#define toUPPER_A(c) toUPPER(c)
25200305
KW
1789#define toFOLD_A(c) toFOLD(c)
1790#define toTITLE_A(c) toTITLE(c)
1a0901db 1791
4650c663 1792/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1793#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1794 ? (c) \
f4cd282c 1795 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1796#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1797
1a0901db 1798/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1799 * all mapped to one of them, and these need special handling; returns the
1800 * input itself if is out-of-range */
b2bf251f 1801#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1802 ? (c) \
f4cd282c 1803 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1804#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1805
beab9ebe
KW
1806/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1807
1808/* For internal core Perl use only: the base macro for defining macros like
1809 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1810 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1811 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1812 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1813 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1814 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1815 * code points under locale rules, use the macros based on _generic_LC_uvchr
1816 * instead of this. */
beab9ebe
KW
1817#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1818 (! FITS_IN_8_BITS(c) \
1819 ? 0 \
31f05a37
KW
1820 : IN_UTF8_CTYPE_LOCALE \
1821 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1822 : cBOOL(non_utf8))
1823
1824/* For internal core Perl use only: a helper macro for defining macros like
1825 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1826 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1827 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1828 * _generic_LC_base for more info */
1829#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1830 _generic_LC_base(c,utf8_locale_classnum, \
1831 non_utf8_func( (_LC_CAST) (c)))
1832
1833/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1834 * 'c' is the platform's native underscore character */
1835#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1836 _generic_LC_base(c, utf8_locale_classnum, \
1837 (non_utf8_func( (_LC_CAST) (c)) \
1838 || (char)(c) == '_'))
1839
1840/* These next three are also for internal core Perl use only: case-change
247985d4
KW
1841 * helper macros. The reason for using the PL_latin arrays is in case the
1842 * system function is defective; it ensures uniform results that conform to the
b257a28c 1843 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
beab9ebe
KW
1844#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1845 ? (c) \
31f05a37
KW
1846 : (IN_UTF8_CTYPE_LOCALE) \
1847 ? PL_latin1_lc[ (U8) (c) ] \
5a10328c 1848 : (cast)function((cast)(c)))
beab9ebe 1849
31f05a37
KW
1850/* Note that the result can be larger than a byte in a UTF-8 locale. It
1851 * returns a single value, so can't adequately return the upper case of LATIN
1852 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1853 * values "SS"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1854 * otherwise returns its input. It does not handle the anomalies in UTF-8
1855 * Turkic locales. */
beab9ebe
KW
1856#define _generic_toUPPER_LC(c, function, cast) \
1857 (! FITS_IN_8_BITS(c) \
1858 ? (c) \
31f05a37 1859 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1860 ? (cast)function((cast)(c)) \
31f05a37
KW
1861 : ((((U8)(c)) == MICRO_SIGN) \
1862 ? GREEK_CAPITAL_LETTER_MU \
1863 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1864 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1865 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1866 ? (__ASSERT_(0) (c)) \
1867 : PL_mod_latin1_uc[ (U8) (c) ])))))
1868
1869/* Note that the result can be larger than a byte in a UTF-8 locale. It
1870 * returns a single value, so can't adequately return the fold case of LATIN
1871 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1872 * values "ss"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1873 * otherwise returns its input. It does not handle the anomalies in UTF-8
1874 * Turkic locales */
beab9ebe 1875#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1876 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1877 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1878 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1879 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1880 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1881
84061b6a 1882/* Use the libc versions for these if available. */
f05550c0 1883#if defined(HAS_ISASCII)
84061b6a
KW
1884# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1885#else
1886# define isASCII_LC(c) isASCII(c)
1887#endif
1888
f05550c0 1889#if defined(HAS_ISBLANK)
84061b6a 1890# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1891#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1892# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1893#endif
1894
f05550c0 1895#define _LC_CAST U8
bbce6d69 1896
f05550c0 1897#ifdef WIN32
375f5f06
KW
1898 /* The Windows functions don't bother to follow the POSIX standard, which
1899 * for example says that something can't both be a printable and a control.
1900 * But Windows treats the \t control as a printable, and does such things
1901 * as making superscripts into both digits and punctuation. This tames
1902 * these flaws by assuming that the definitions of both controls and space
1903 * are correct, and then making sure that other definitions don't have
1904 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1905 * Not all possible weirdnesses are checked for, just the ones that were
1906 * detected on actual Microsoft code pages */
1907
b7d90381
KW
1908# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1909# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1910
1911# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1912 && isALPHANUMERIC_LC(c))
1913# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1914 ! isPUNCT_LC(c))
1915# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1916 isALPHANUMERIC_LC(c))
1917# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1918# define isIDFIRST_LC(c) (((c) == '_') \
1919 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1920# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1921# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1922# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1923# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1924# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1925# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1926 && isALPHANUMERIC_LC(c))
f05550c0
BF
1927
1928# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1929# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1930# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1931
1932#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1933 /* For most other platforms */
beab9ebe 1934
f05550c0
BF
1935# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1936# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1937# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1938# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1939# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1940# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1941# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1942# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1943# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1944# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1945# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1946# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1947# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1948
1949
1950# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1951# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1952# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1953
1954#else /* The final fallback position */
1955
b7d90381
KW
1956# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1957# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1958# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1959# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1960# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1961# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1962# define isLOWER_LC(c) (isascii(c) && islower(c))
1963# define isPRINT_LC(c) (isascii(c) && isprint(c))
1964# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1965# define isSPACE_LC(c) (isascii(c) && isspace(c))
1966# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1967# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1968# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1969
1970# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1971# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1972# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1973
f05550c0 1974#endif
55204971 1975
eba68aa0
KW
1976#define isIDCONT(c) isWORDCHAR(c)
1977#define isIDCONT_A(c) isWORDCHAR_A(c)
1978#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1979#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1980#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1981
4650c663 1982/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1983 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1984 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1985 * classes where there is a macro or function 'above_latin1' that takes the
1986 * single argument 'c' and returns the desired value. These exist for those
2366ba44
KW
1987 * classes which have simple definitions, avoiding the overhead of an inversion
1988 * list binary search. _generic_invlist_uvchr() can be used
4650c663 1989 * for classes where that overhead is faster than a direct lookup.
d0da05db 1990 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1991 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1992 * have duplicate checks here, so could create versions of the macros that
1993 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1994
1995/* Note that all ignore 'use bytes' */
1e222e4f
KW
1996#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1997 ? _generic_isCC(c, classnum) \
cd500f2f 1998 : above_latin1(c))
2366ba44 1999#define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
1e222e4f 2000 ? _generic_isCC(c, classnum) \
922e8cb4 2001 : _is_uni_FOO(classnum, c))
2366ba44
KW
2002#define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
2003#define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
d0da05db
KW
2004#define isASCII_uvchr(c) isASCII(c)
2005#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
2006#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2366ba44
KW
2007#define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
2008#define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
1e222e4f
KW
2009#define isIDCONT_uvchr(c) \
2010 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
2011#define isIDFIRST_uvchr(c) \
2012 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2366ba44
KW
2013#define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
2014#define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
d0da05db 2015
2366ba44 2016#define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
d0da05db
KW
2017#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
2018#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
2019
2366ba44 2020#define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
d0da05db 2021#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2366ba44 2022#define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
d0da05db
KW
2023#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
2024
2025#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
2026#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
2027#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
2028#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
2029
2030/* For backwards compatibility, even though '_uni' should mean official Unicode
2031 * code points, in Perl it means native for those below 256 */
2032#define isALPHA_uni(c) isALPHA_uvchr(c)
2033#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
2034#define isASCII_uni(c) isASCII_uvchr(c)
2035#define isBLANK_uni(c) isBLANK_uvchr(c)
2036#define isCNTRL_uni(c) isCNTRL_uvchr(c)
2037#define isDIGIT_uni(c) isDIGIT_uvchr(c)
2038#define isGRAPH_uni(c) isGRAPH_uvchr(c)
2039#define isIDCONT_uni(c) isIDCONT_uvchr(c)
2040#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
2041#define isLOWER_uni(c) isLOWER_uvchr(c)
2042#define isPRINT_uni(c) isPRINT_uvchr(c)
2043#define isPUNCT_uni(c) isPUNCT_uvchr(c)
2044#define isSPACE_uni(c) isSPACE_uvchr(c)
2045#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
2046#define isUPPER_uni(c) isUPPER_uvchr(c)
2047#define isVERTWS_uni(c) isVERTWS_uvchr(c)
2048#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2049#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2050#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2051#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2052#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2053#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 2054
4650c663
KW
2055/* For internal core Perl use only: the base macros for defining macros like
2056 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 2057 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 2058 * simple class definitions; the other for more complex. These are like
d0da05db 2059 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
2060#define _generic_LC_uvchr(latin1, above_latin1, c) \
2061 (c < 256 ? latin1(c) : above_latin1(c))
2366ba44 2062#define _generic_LC_invlist_uvchr(latin1, classnum, c) \
cd500f2f
KW
2063 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2064
2366ba44
KW
2065#define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2066#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
cd500f2f 2067 _CC_ALPHANUMERIC, c)
b7d90381
KW
2068#define isASCII_LC_uvchr(c) isASCII_LC(c)
2069#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2070 is_HORIZWS_cp_high, c)
feeab5a9 2071#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2366ba44
KW
2072#define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2073#define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 2074#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 2075 _is_uni_perl_idcont, c)
b7d90381 2076#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f 2077 _is_uni_perl_idstart, c)
2366ba44
KW
2078#define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2079#define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 2080#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2366ba44 2081#define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 2082#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 2083 is_XPERLSPACE_cp_high, c)
2366ba44
KW
2084#define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2085#define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
cd500f2f 2086 _CC_WORDCHAR, c)
b7d90381
KW
2087#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2088 is_XDIGIT_cp_high, c)
e712593e 2089
b7d90381 2090#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 2091
da8c1a98
KW
2092/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2093 * they don't otherwise go out of their way to look for malformed UTF-8. If
2094 * they can return accurate results without knowing if the input is otherwise
2095 * malformed, they do so. For example isASCII is accurate in spite of any
2096 * non-length malformations because it looks only at a single byte. Likewise
2097 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2098 * variant ones return FALSE. But, if the input has to be well-formed in order
2099 * for the results to be accurate, the macros will test and if malformed will
2100 * call a routine to die
2101 *
2102 * Except for toke.c, the macros do assume that e > p, asserting that on
2103 * DEBUGGING builds. Much code that calls these depends on this being true,
2104 * for other reasons. toke.c is treated specially as using the regular
2105 * assertion breaks it in many ways. All strings that these operate on there
2106 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2107 * bunch of code in toke.c assumes that this is true, so the assertion allows
2108 * for that */
2109#ifdef PERL_IN_TOKE_C
2110# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2111#else
2112# define _utf8_safe_assert(p,e) ((e) > (p))
2113#endif
2114
2115#define _generic_utf8_safe(classnum, p, e, above_latin1) \
c81b3562
KW
2116 ((! _utf8_safe_assert(p, e)) \
2117 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2118 : (UTF8_IS_INVARIANT(*(p))) \
da8c1a98
KW
2119 ? _generic_isCC(*(p), classnum) \
2120 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2121 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2122 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2123 classnum) \
2124 : (_force_out_malformed_utf8_message( \
2125 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2126 : above_latin1))
b7d90381
KW
2127/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2128 * 'above_latin1' can be a macro */
da8c1a98
KW
2129#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2130 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2366ba44 2131#define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2132 _generic_utf8_safe(classnum, p, e, \
2133 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2134 ? (_force_out_malformed_utf8_message( \
2135 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2136 : above_latin1(p)))
2366ba44
KW
2137/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2138 * 'above_latin1' parameter */
2139#define _generic_invlist_utf8_safe(classnum, p, e) \
2140 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
922e8cb4 2141
cc8ab7c0 2142/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
2143 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2144 * class is TRUE for. Hence it can skip the tests for this range.
2145 * 'above_latin1' should include its arguments */
da8c1a98
KW
2146#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2147 (__ASSERT_(_utf8_safe_assert(p, e)) \
2148 (UTF8_IS_INVARIANT(*(p))) \
2149 ? _generic_isCC(*(p), classnum) \
2150 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2151 ? 0 /* Note that doesn't check validity for latin1 */ \
2152 : above_latin1)
2153
84238efa 2154
059703b0
KW
2155#define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2156#define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2157#define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2158#define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2159#define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2160#define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2161#define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2162#define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2163#define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2164#define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2165#define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2166#define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2167#define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2168#define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2169#define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2170#define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2171#define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2172#define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
e8fa43e2 2173
2366ba44 2174#define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
da8c1a98 2175#define isALPHANUMERIC_utf8_safe(p, e) \
2366ba44 2176 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
da8c1a98
KW
2177#define isASCII_utf8_safe(p, e) \
2178 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2179 * works */ \
2180 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2181#define isBLANK_utf8_safe(p, e) \
2366ba44 2182 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
da8c1a98 2183
e8fa43e2
KW
2184#ifdef EBCDIC
2185 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2186 * more efficient macro instead of the more general one */
da8c1a98 2187# define isCNTRL_utf8_safe(p, e) \
56d02b8c 2188 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 2189#else
da8c1a98 2190# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
2191#endif
2192
da8c1a98
KW
2193#define isDIGIT_utf8_safe(p, e) \
2194 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2366ba44
KW
2195 _is_utf8_FOO(_CC_DIGIT, p, e))
2196#define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
da8c1a98 2197#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
dd1a3ba7 2198 _is_utf8_perl_idcont, p, e)
e5dcd934 2199
c11ff943
KW
2200/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2201 * IDFIRST is an alnum. See
8034715d 2202 * https://github.com/Perl/perl5/issues/10275 for more detail than you
f91dcd13
KW
2203 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2204 * This used to be not the XID version, but we decided to go with the more
2205 * modern Unicode definition */
da8c1a98
KW
2206#define isIDFIRST_utf8_safe(p, e) \
2207 _generic_func_utf8_safe(_CC_IDFIRST, \
dd1a3ba7 2208 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
da8c1a98 2209
2366ba44
KW
2210#define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2211#define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
da8c1a98 2212#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2366ba44 2213#define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
da8c1a98 2214#define isSPACE_utf8_safe(p, e) \
2366ba44
KW
2215 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2216#define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
da8c1a98 2217#define isVERTWS_utf8_safe(p, e) \
2366ba44 2218 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
da8c1a98 2219#define isWORDCHAR_utf8_safe(p, e) \
2366ba44 2220 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
da8c1a98
KW
2221#define isXDIGIT_utf8_safe(p, e) \
2222 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2223 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2224 ? (_force_out_malformed_utf8_message( \
2225 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2226 : is_XDIGIT_high(p)))
a0ed51b3 2227
059703b0
KW
2228#define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2229#define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2230#define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2231#define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2e8adce6 2232
567b353c 2233/* For internal core use only, subject to change */
059703b0
KW
2234#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2235#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2236#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2237#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
a1a5ec35
KW
2238
2239#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2240#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2241#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2242#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 2243
059703b0
KW
2244#define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2245#define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2246#define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2247#define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2248#define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2249#define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2250#define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2251#define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2252#define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2253#define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2254#define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2255#define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2256#define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2257#define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2258#define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2259#define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2260#define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
34aeb2e9 2261
da8c1a98
KW
2262/* For internal core Perl use only: the base macros for defining macros like
2263 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2264 * point in 'p' is within the 0-255 range, it uses locale rules from the
2265 * passed-in 'macro' parameter */
2266#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2267 (__ASSERT_(_utf8_safe_assert(p, e)) \
2268 (UTF8_IS_INVARIANT(*(p))) \
2269 ? macro(*(p)) \
2270 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2271 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2272 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2273 : (_force_out_malformed_utf8_message( \
2274 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2275 : above_latin1))
2276
2366ba44
KW
2277#define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2278 _generic_LC_utf8_safe(macro, p, e, \
2279 _is_utf8_FOO(classnum, p, e))
da8c1a98
KW
2280
2281#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2282 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2283
2366ba44 2284#define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2285 _generic_LC_utf8_safe(classnum, p, e, \
2286 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2287 ? (_force_out_malformed_utf8_message( \
2288 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2289 : above_latin1(p)))
2290
2291#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2366ba44 2292 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
da8c1a98
KW
2293 _CC_ALPHANUMERIC, p, e)
2294#define isALPHA_LC_utf8_safe(p, e) \
2366ba44 2295 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
da8c1a98
KW
2296#define isASCII_LC_utf8_safe(p, e) \
2297 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2298#define isBLANK_LC_utf8_safe(p, e) \
2366ba44 2299 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
da8c1a98
KW
2300#define isCNTRL_LC_utf8_safe(p, e) \
2301 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2302#define isDIGIT_LC_utf8_safe(p, e) \
2366ba44 2303 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
da8c1a98 2304#define isGRAPH_LC_utf8_safe(p, e) \
2366ba44 2305 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
da8c1a98
KW
2306#define isIDCONT_LC_utf8_safe(p, e) \
2307 _generic_LC_func_utf8_safe(isIDCONT_LC, \
dd1a3ba7 2308 _is_utf8_perl_idcont, p, e)
da8c1a98
KW
2309#define isIDFIRST_LC_utf8_safe(p, e) \
2310 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
dd1a3ba7 2311 _is_utf8_perl_idstart, p, e)
da8c1a98 2312#define isLOWER_LC_utf8_safe(p, e) \
2366ba44 2313 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
da8c1a98 2314#define isPRINT_LC_utf8_safe(p, e) \
2366ba44 2315 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
da8c1a98
KW
2316#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2317#define isPUNCT_LC_utf8_safe(p, e) \
2366ba44 2318 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
da8c1a98 2319#define isSPACE_LC_utf8_safe(p, e) \
2366ba44 2320 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
da8c1a98 2321#define isUPPER_LC_utf8_safe(p, e) \
2366ba44 2322 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
da8c1a98 2323#define isWORDCHAR_LC_utf8_safe(p, e) \
2366ba44 2324 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
da8c1a98 2325#define isXDIGIT_LC_utf8_safe(p, e) \
2366ba44 2326 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2327
fbc19f27
KW
2328/* Macros for backwards compatibility and for completeness when the ASCII and
2329 * Latin1 values are identical */
b7d90381
KW
2330#define isALPHAU(c) isALPHA_L1(c)
2331#define isDIGIT_L1(c) isDIGIT_A(c)
2332#define isOCTAL(c) isOCTAL_A(c)
2333#define isOCTAL_L1(c) isOCTAL_A(c)
2334#define isXDIGIT_L1(c) isXDIGIT_A(c)
2335#define isALNUM(c) isWORDCHAR(c)
a377c856 2336#define isALNUM_A(c) isALNUM(c)
b7d90381
KW
2337#define isALNUMU(c) isWORDCHAR_L1(c)
2338#define isALNUM_LC(c) isWORDCHAR_LC(c)
2339#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2340#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
059703b0 2341#define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
4c1d9526 2342#define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
059703b0 2343#define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
4c1d9526 2344#define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
b7d90381
KW
2345#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2346#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2347#define isALNUMC(c) isALPHANUMERIC(c)
2348#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2349#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2350#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
059703b0 2351#define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
4c1d9526
KW
2352#define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2353#define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
fbc19f27 2354
2bd1cbf6
KW
2355/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2356 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2357 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2358 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2359 * the outlier from the block that contains the other controls, just like
2360 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2361 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2362 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2363 * etc. */
2bd1cbf6 2364#ifndef EBCDIC
75763b3a 2365# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2366#else
75763b3a
KW
2367# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2368 ((isPRINT_A(c)) \
2369 ? (UNLIKELY((c) == '?') \
2370 ? QUESTION_MARK_CTRL \
2371 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2372 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2373 ? '?' \
2374 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2375#endif
bbce6d69 2376
dea28490
JJ
2377/* Line numbers are unsigned, 32 bits. */
2378typedef U32 line_t;
e5dcd934 2379#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2380
91152fc1
DG
2381/* Helpful alias for version prescan */
2382#define is_LAX_VERSION(a,b) \
2383 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2384
2385#define is_STRICT_VERSION(a,b) \
2386 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2387
2388#define BADVERSION(a,b,c) \
2389 if (b) { \
2390 *b = c; \
2391 } \
2392 return a;
8c52afec 2393
1ce77b7d
KW
2394/* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2395 * a-f) to its numeric value without using any branches. The input is
2396 * validated only by an assert() in DEBUGGING builds.
2397 *
2398 * It works by right shifting and isolating the bit that is 0 for the digits,
2399 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2400 * position, and then to the eights position. Both are added together to form
2401 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2402 * final four bits of the input to form the correct value. */
2403#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2404 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2405 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2406 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2407
2408/* The argument is a string pointer, which is advanced. */
2409#define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
95a59cab 2410
cb27eebd
KW
2411/* Converts a character known to represent an octal digit (0-7) to its numeric
2412 * value. The input is validated only by an assert() in DEBUGGING builds. In
2413 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2414#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2415
305b8651
KW
2416/* Efficiently returns a boolean as to if two native characters are equivalent
2417 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2418 * the ALPHA in the name is to remind you of that. This is asserted() in
2419 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2420 * works (on valid input) for both non- and UTF-8-encoded bytes.
2421 *
2422 * When one of the inputs is a compile-time constant and gets folded by the
2423 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2424 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2425 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2426 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2427 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2428 * just a single 0, in the bit position where the upper- and lowercase differ.
2429 * */
305b8651
KW
2430#define isALPHA_FOLD_EQ(c1, c2) \
2431 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2432 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2433#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2434
8e84507e 2435/*
70b05c7c 2436=for apidoc_section Memory Management
ccfc67b7 2437
a02a5408 2438=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2439The XSUB-writer's interface to the C C<malloc> function.
2440
596f7718 2441Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2442
c5008215
JC
2443In 5.9.3, Newx() and friends replace the older New() API, and drops
2444the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2445themselves. This aid has been superseded by a new build option,
d10b4965 2446PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2447there for use in XS modules supporting older perls.
2448
a02a5408 2449=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2450The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2451cast. See also C<L</Newx>>.
954c1994 2452
596f7718 2453Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2454
a02a5408 2455=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2456The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2457memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2458
596f7718 2459Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2460
954c1994
GS
2461=for apidoc Am|void|Renew|void* ptr|int nitems|type
2462The XSUB-writer's interface to the C C<realloc> function.
2463
596f7718 2464Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2465
954c1994
GS
2466=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2467The XSUB-writer's interface to the C C<realloc> function, with
2468cast.
2469
596f7718 2470Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2471
49b8b560 2472=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2473The XSUB-writer's interface to the C C<free> function.
2474
596f7718 2475This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2476
51b56f5c 2477=for apidoc_section String Handling
954c1994
GS
2478=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2479The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2480source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2481C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2482
e90e2364 2483=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2484Like C<Move> but returns C<dest>. Useful
72d33970 2485for encouraging compilers to tail-call
e90e2364
NC
2486optimise.
2487
954c1994
GS
2488=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2489The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2490source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2491C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2492
e90e2364
NC
2493=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2494
796b6530 2495Like C<Copy> but returns C<dest>. Useful
72d33970 2496for encouraging compilers to tail-call
e90e2364
NC
2497optimise.
2498
954c1994
GS
2499=for apidoc Am|void|Zero|void* dest|int nitems|type
2500
2501The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2502destination, C<nitems> is the number of items, and C<type> is the type.
2503
e90e2364
NC
2504=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2505
72d33970
FC
2506Like C<Zero> but returns dest. Useful
2507for encouraging compilers to tail-call
e90e2364
NC
2508optimise.
2509
51b56f5c 2510=for apidoc_section Utility Functions
da5d8dbb 2511=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2512This is an architecture-independent macro to copy one structure to another.
954c1994 2513
7e337ee0
JH
2514=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2515
2516Fill up memory with a byte pattern (a byte repeated over and over
2517again) that hopefully catches attempts to access uninitialized memory.
2518
2519=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2520
2521PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2522
1c12ffb4 2523=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2524
2525PoisonWith(0xEF) for catching access to freed memory.
2526
9965345d
JH
2527=for apidoc Am|void|Poison|void* dest|int nitems|type
2528
7e337ee0 2529PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2530
2531=cut */
954c1994 2532
561b68a9
SH
2533/* Maintained for backwards-compatibility only. Use newSV() instead. */
2534#ifndef PERL_CORE
ff06c60c 2535#define NEWSV(x,len) newSV(len)
561b68a9 2536#endif
ff06c60c 2537
b7112dce 2538#define MEM_SIZE_MAX ((MEM_SIZE)-1)
19a94d75 2539
a500027b 2540#define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
e6bdf523 2541
27d5b266 2542#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2543
2544/* This expression will be constant-folded at compile time. It checks
2545 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2546 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2547 * As well as avoiding the need for a run-time check in some cases, it's
2548 * designed to avoid compiler warnings like:
2549 * comparison is always false due to limited range of data type
73e8ff00
DM
2550 * It's mathematically equivalent to
2551 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2552 */
2553
2554# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
445198b9
LM
2555 ( sizeof(MEM_SIZE) < sizeof(n) \
2556 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 2557
88f9f128 2558/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2559 * compiler warnings. We *want* to write the expression as
2560 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2561 * (for some compile-time constant C), but even when the LHS
2562 * constant-folds to false at compile-time, g++ insists on emitting
2563 * warnings about the RHS (e.g. "comparison is always false"), so instead
2564 * we write it as
e6bdf523 2565 *
d98e5cde 2566 * (cond ? n : X) > C
88f9f128 2567 *
d98e5cde
DM
2568 * where X is a constant with X > C always false. Choosing a value for X
2569 * is tricky. If 0, some compilers will complain about 0 > C always being
2570 * false; if 1, Coverity complains when n happens to be the constant value
2571 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2572 * for X and hope that nothing else whines.
e6bdf523
DM
2573 */
2574
2575# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2576 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2577 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2578
2579# define MEM_WRAP_CHECK(n,t) \
2580 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2581 && (croak_memory_wrap(),0))
2582
2583# define MEM_WRAP_CHECK_1(n,t,a) \
2584 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2585 && (Perl_croak_nocontext("%s",(a)),0))
2586
814eedc8
DD
2587/* "a" arg must be a string literal */
2588# define MEM_WRAP_CHECK_s(n,t,a) \
2589 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2590 && (Perl_croak_nocontext("" a ""),0))
2591
8b44ba4c 2592#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2593
a500027b 2594#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
2595#else
2596
410319be
NC
2597#define MEM_WRAP_CHECK(n,t)
2598#define MEM_WRAP_CHECK_1(n,t,a)
814eedc8 2599#define MEM_WRAP_CHECK_s(n,t,a)
8b44ba4c
NC
2600#define MEM_WRAP_CHECK_(n,t)
2601
a500027b 2602#define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
27d5b266 2603
1936d2a7 2604#endif
8b44ba4c 2605
fe4f188c 2606#ifdef PERL_MEM_LOG
46c6c7e2 2607/*
9f653bb5 2608 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2609 * go through functions, which are handy for debugging breakpoints, but
2610 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2611 * line number, and C function name if available) passed in. This info can
2612 * then be used for logging the calls, for which one gets a sample
73d1d973 2613 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2614 *
46c6c7e2 2615 * Known problems:
94e892a6 2616 * - not all memory allocs get logged, only those
46c6c7e2 2617 * that go through Newx() and derivatives (while all
94e892a6 2618 * Safefrees do get logged)
46c6c7e2
JH
2619 * - __FILE__ and __LINE__ do not work everywhere
2620 * - __func__ or __FUNCTION__ even less so
2621 * - I think more goes on after the perlio frees but
2622 * the thing is that STDERR gets closed (as do all
2623 * the file descriptors)
2624 * - no deeper calling stack than the caller of the Newx()
2625 * or the kind, but do I look like a C reflection/introspection
2626 * utility to you?
2627 * - the function prototypes for the logging functions
2628 * probably should maybe be somewhere else than handy.h
2629 * - one could consider inlining (macrofying) the logging
2630 * for speed, but I am too lazy
2631 * - one could imagine recording the allocations in a hash,
2632 * (keyed by the allocation address?), and maintain that
2633 * through reallocs and frees, but how to do that without
2634 * any News() happening...?
73d1d973 2635 * - lots of -Ddefines to get useful/controllable output
b953482e 2636 * - lots of ENV reads
46c6c7e2
JH
2637 */
2638
0b0ab801 2639# ifdef PERL_CORE
73d1d973 2640# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2641enum mem_log_type {
2642 MLT_ALLOC,
2643 MLT_REALLOC,
d7a2c63c
MHM
2644 MLT_FREE,
2645 MLT_NEW_SV,
2646 MLT_DEL_SV
0b0ab801
MHM
2647};
2648# endif
12754f92 2649# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2650void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2651void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2652# endif
0b0ab801
MHM
2653# endif
2654
fe4f188c
JH
2655#endif
2656
2657#ifdef PERL_MEM_LOG
d1401ee9
MHM
2658#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2659#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2660#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2661#endif
2662
2663#ifndef MEM_LOG_ALLOC
2664#define MEM_LOG_ALLOC(n,t,a) (a)
2665#endif
2666#ifndef MEM_LOG_REALLOC
2667#define MEM_LOG_REALLOC(n,t,v,a) (a)
2668#endif
2669#ifndef MEM_LOG_FREE
2670#define MEM_LOG_FREE(a) (a)
2671#endif
2672
d1401ee9
MHM
2673#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2674#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2675#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2676
2677#ifndef PERL_CORE
a02a5408
JC
2678/* pre 5.9.x compatibility */
2679#define New(x,v,n,t) Newx(v,n,t)
2680#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2681#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2682#endif
a02a5408 2683
ff68c719 2684#define Renew(v,n,t) \
d1401ee9 2685 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2686#define Renewc(v,n,t,c) \
d1401ee9 2687 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2688
2689#ifdef PERL_POISON
2690#define Safefree(d) \
06c0cc96 2691 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2692#else
fe4f188c 2693#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2694#endif
55497cff 2695
dbb57106
YO
2696/* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2697 * as it handles cases like constant string arguments without throwing warnings *
2698 * the cast is required, as is the inequality check, to avoid warnings */
45908e4d 2699#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2700
45908e4d
YO
2701
2702#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)))
2703#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)))
2704#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2705
bdd1531d 2706/* Like above, but returns a pointer to 'd' */
45908e4d
YO
2707#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)))
2708#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 2709#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2710
7e337ee0
JH
2711#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2712#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2713#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2714#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2715
caa674f3
DD
2716#ifdef PERL_POISON
2717# define PERL_POISON_EXPR(x) x
2718#else
2719# define PERL_POISON_EXPR(x)
2720#endif
2721
ff68c719 2722#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2723
1b7e2294 2724/*
51b56f5c 2725=for apidoc_section Utility Functions
1b7e2294
KW
2726
2727=for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2728
2729Returns the number of elements in the input C array (so you want your
2730zero-based indices to be less than but not equal to).
2731
2732=for apidoc Am|void *|C_ARRAY_END|void *a
2733
2734Returns a pointer to one element past the final element of the input C array.
2735
2736=cut
2737
2738C_ARRAY_END is one past the last: half-open/half-closed range, not
2739last-inclusive range.
2740*/
622913ab 2741#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2742#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2743
2cc61e15
DD
2744#ifdef NEED_VA_COPY
2745# ifdef va_copy
2746# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2747# elif defined(__va_copy)
2748# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2749# else
07798b17 2750# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2751# endif
2752#endif
2753
472d47bc
SB
2754/* convenience debug macros */
2755#ifdef USE_ITHREADS
2756#define pTHX_FORMAT "Perl interpreter: 0x%p"
2757#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2758#define pTHX_VALUE_ (void *)my_perl,
2759#define pTHX_VALUE (void *)my_perl
2760#define pTHX__VALUE_ ,(void *)my_perl,
2761#define pTHX__VALUE ,(void *)my_perl
472d47bc 2762#else
3609ea0d 2763#define pTHX_FORMAT
472d47bc 2764#define pTHX__FORMAT
3609ea0d 2765#define pTHX_VALUE_
472d47bc 2766#define pTHX_VALUE
3609ea0d 2767#define pTHX__VALUE_
472d47bc
SB
2768#define pTHX__VALUE
2769#endif /* USE_ITHREADS */
3609ea0d 2770
2acdbac1
NC
2771/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2772 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2773 CPAN::Unpack show any users outside the core. */
2774#ifdef PERL_CORE
dc6e8de0
A
2775# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2776 "Use of " s " is deprecated")
c9680906
A
2777# define deprecate_disappears_in(when,message) \
2778 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2779 message ", and will disappear in Perl " when)
ac641426
A
2780# define deprecate_fatal_in(when,message) \
2781 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2782 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2783#endif
2784
dfff4baf
BF
2785/* Internal macros to deal with gids and uids */
2786#ifdef PERL_CORE
2787
2788# if Uid_t_size > IVSIZE
2789# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2790# define SvUID(sv) SvNV(sv)
07798b17
AC
2791# elif Uid_t_sign <= 0
2792# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2793# define SvUID(sv) SvIV(sv)
dfff4baf 2794# else
07798b17
AC
2795# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2796# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2797# endif /* Uid_t_size */
2798
2799# if Gid_t_size > IVSIZE
2800# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2801# define SvGID(sv) SvNV(sv)
07798b17
AC
2802# elif Gid_t_sign <= 0
2803# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2804# define SvGID(sv) SvIV(sv)
dfff4baf 2805# else
07798b17
AC
2806# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2807# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2808# endif /* Gid_t_size */
2809
2810#endif
2811
6a5bc5ac 2812#endif /* PERL_HANDY_H_ */
9d745869 2813
e9a8c099 2814/*
14d04a33 2815 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2816 */