This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
perlapi: Add a link at the beginning to the Undocumenteds
[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
b290de04 43=for apidoc Am|Size_t|FILE_bufsiz|FILE *f
51b56f5c
KW
44
45=for apidoc_section String Handling
b290de04
KW
46=for apidoc Amu|token|CAT2|token x|token y
47=for apidoc Amu|string|STRINGIFY|token x
51b56f5c
KW
48
49=for apidoc_section Numeric Functions
b290de04 50=for apidoc Am|double|Drand01
51b56f5c
KW
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
54ba56f3
KW
90C<MUTABLE_PTR> is the base macro used to derive new casts. The other
91already-built-in ones return pointers to what their names indicate.
df87895c
KW
92
93=cut
cf3f0ffb
DM
94 */
95
041c1a23 96#if defined(PERL_USE_GCC_BRACE_GROUPS)
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)
72659597 464#define gv_fetchpvn gv_fetchpvn_flags
9dcc53ea
Z
465#define sv_catxmlpvs(dsv, str, utf8) \
466 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 467
ba3a79e7 468
510966aa
Z
469#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
470
b96d8cd9
NC
471#define get_cvs(str, flags) \
472 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 473
9b6e9510 474/* internal helpers */
4a1bbd3d 475/* Transitional */
d67f622b
N
476#ifndef PERL_VERSION_MAJOR
477# define PERL_VERSION_MAJOR PERL_REVISION
4a1bbd3d
KW
478#else
479# undef PERL_REVISION /* We don't want code to be using these */
480#endif
d67f622b
N
481#ifndef PERL_VERSION_MINOR
482# define PERL_VERSION_MINOR PERL_VERSION
4a1bbd3d
KW
483#else
484# undef PERL_VERSION
485#endif
d67f622b
N
486#ifndef PERL_VERSION_PATCH
487# define PERL_VERSION_PATCH PERL_SUBVERSION
4a1bbd3d
KW
488#else
489# undef PERL_SUBVERSION
490#endif
491
492#define PERL_JNP_TO_DECIMAL_(maJor,miNor,Patch) \
493 /* '10*' leaves room for things like alpha, beta, releases */ \
494 (10 * ((maJor) * 1000000) + ((miNor) * 1000) + (Patch))
9b6e9510 495#define PERL_DECIMAL_VERSION_ \
d67f622b
N
496 PERL_JNP_TO_DECIMAL_(PERL_VERSION_MAJOR, PERL_VERSION_MINOR, \
497 PERL_VERSION_PATCH)
9b6e9510
KW
498
499/*
51b56f5c 500=for apidoc_section Versioning
4a1bbd3d 501=for apidoc AmR|bool|PERL_VERSION_EQ|const U8 major|const U8 minor|const U8 patch
9de44d19
KW
502=for apidoc_item PERL_VERSION_NE
503=for apidoc_item PERL_VERSION_LT
504=for apidoc_item PERL_VERSION_LE
505=for apidoc_item PERL_VERSION_GT
506=for apidoc_item PERL_VERSION_GE
9b6e9510 507
4a1bbd3d 508Returns whether or not the perl currently being compiled has the specified
9b6e9510
KW
509relationship to the perl given by the parameters. For example,
510
511 #if PERL_VERSION_GT(5,24,2)
512 code that will only be compiled on perls after v5.24.2
513 #else
514 fallback code
515 #endif
516
517Note that this is usable in making compile-time decisions
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
9b6e9510
KW
545=cut
546*/
547
4a1bbd3d
KW
548/* N.B. These don't work if the patch version is 42 or 92, as those are what
549 * '*' is in ASCII and EBCDIC respectively */
550# define PERL_VERSION_EQ(j,n,p) \
551 (((p) == '*') \
d67f622b
N
552 ? ( (j) == PERL_VERSION_MAJOR \
553 && (n) == PERL_VERSION_MINOR) \
4a1bbd3d
KW
554 : (PERL_DECIMAL_VERSION_ == PERL_JNP_TO_DECIMAL_(j,n,p)))
555# define PERL_VERSION_NE(j,n,p) (! PERL_VERSION_EQ(j,n,p))
556
557# define PERL_VERSION_LT(j,n,p) /* < '*' effectively means < 0 */ \
558 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
559 (n), \
560 (((p) == '*') ? 0 : p)))
561# define PERL_VERSION_GE(j,n,p) (! PERL_VERSION_LT(j,n,p))
562
563# define PERL_VERSION_LE(j,n,p) /* <= '*' effectively means < n+1 */ \
564 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
565 (((p) == '*') ? ((n)+1) : (n)), \
566 (((p) == '*') ? 0 : p)))
567# define PERL_VERSION_GT(j,n,p) (! PERL_VERSION_LE(j,n,p))
9b6e9510 568
954c1994 569/*
51b56f5c 570=for apidoc_section String Handling
ccfc67b7 571
954c1994 572=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
573Test two C<NUL>-terminated strings to see if they are different. Returns true
574or false.
954c1994
GS
575
576=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
577Test two C<NUL>-terminated strings to see if they are equal. Returns true or
578false.
954c1994
GS
579
580=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
581Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
582second, C<s2>. Returns true or false.
954c1994
GS
583
584=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
585Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
586equal to the second, C<s2>. Returns true or false.
954c1994
GS
587
588=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
589Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
590the second, C<s2>. Returns true or false.
954c1994
GS
591
592=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
593Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
594or equal to the second, C<s2>. Returns true or false.
954c1994
GS
595
596=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
597Test two C<NUL>-terminated strings to see if they are different. The C<len>
598parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
599wrapper for C<strncmp>).
600
601=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
602Test two C<NUL>-terminated strings to see if they are equal. The C<len>
603parameter indicates the number of bytes to compare. Returns true or false. (A
604wrapper for C<strncmp>).
954c1994 605
bd18bd40
KW
606=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
607Test two buffers (which may contain embedded C<NUL> characters, to see if they
608are equal. The C<len> parameter indicates the number of bytes to compare.
609Returns zero if equal, or non-zero if non-equal.
610
3bb9fd01 611=for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
612Like L</memEQ>, but the second string is a literal enclosed in double quotes,
613C<l1> gives the number of bytes in C<s1>.
614Returns zero if equal, or non-zero if non-equal.
615
bd18bd40
KW
616=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
617Test two buffers (which may contain embedded C<NUL> characters, to see if they
618are not equal. The C<len> parameter indicates the number of bytes to compare.
619Returns zero if non-equal, or non-zero if equal.
620
3bb9fd01 621=for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
622Like L</memNE>, but the second string is a literal enclosed in double quotes,
623C<l1> gives the number of bytes in C<s1>.
624Returns zero if non-equal, or zero if non-equal.
625
4aada8b9
KW
626=for apidoc Am|bool|memCHRs|"list"|char c
627Returns the position of the first occurence of the byte C<c> in the literal
628string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
629treated as unsigned char. Thus this macro can be used to determine if C<c> is
630in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
631is C<NUL> (and the set doesn't include C<NUL>).
632
954c1994 633=cut
fc169e00
KW
634
635New macros should use the following conventions for their names (which are
636based on the underlying C library functions):
637
638 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
639
640 Each has two main parameters, string-like operands that are compared
641 against each other, as specified by the macro name. Some macros may
642 additionally have one or potentially even two length parameters. If a length
643 parameter applies to both string parameters, it will be positioned third;
644 otherwise any length parameter immediately follows the string parameter it
645 applies to.
646
647 If the prefix to the name is 'str', the string parameter is a pointer to a C
648 language string. Such a string does not contain embedded NUL bytes; its
649 length may be unknown, but can be calculated by C<strlen()>, since it is
650 terminated by a NUL, which isn't included in its length.
651
a3815e44 652 The optional 'n' following 'str' means that there is a third parameter,
fc169e00
KW
653 giving the maximum number of bytes to look at in each string. Even if both
654 strings are longer than the length parameter, those extra bytes will be
655 unexamined.
656
657 The 's' suffix means that the 2nd byte string parameter is a literal C
658 double-quoted string. Its length will automatically be calculated by the
659 macro, so no length parameter will ever be needed for it.
660
661 If the prefix is 'mem', the string parameters don't have to be C strings;
662 they may contain embedded NUL bytes, do not necessarily have a terminating
663 NUL, and their lengths can be known only through other means, which in
664 practice are additional parameter(s) passed to the function. All 'mem'
665 functions have at least one length parameter. Barring any 'l' or 's' suffix,
666 there is a single length parameter, in position 3, which applies to both
667 string parameters. The 's' suffix means, as described above, that the 2nd
668 string is a literal double-quoted C string (hence its length is calculated by
669 the macro, and the length parameter to the function applies just to the first
670 string parameter, and hence is positioned just after it). An 'l' suffix
671 means that the 2nd string parameter has its own length parameter, and the
672 signature will look like memFOOl(s1, l1, s2, l2).
673
674 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
675 substring of the 1st string. 'P' if present indicates that the substring
676 must be a "proper" one in tha mathematical sense that the first one must be
677 strictly larger than the 2nd.
678
954c1994
GS
679*/
680
62946e08 681
75400963
KW
682#define strNE(s1,s2) (strcmp(s1,s2) != 0)
683#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
684#define strLT(s1,s2) (strcmp(s1,s2) < 0)
685#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
686#define strGT(s1,s2) (strcmp(s1,s2) > 0)
687#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 688
75400963
KW
689#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
690#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 691
9d3980bc
KW
692#define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
693#define memNE(s1,s2,l) (! memEQ(s1,s2,l))
36477c24 694
085b7534 695/* memEQ and memNE where second comparand is a string constant */
568a785a 696#define memEQs(s1, l, s2) \
777fa2cb 697 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 698#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 699
fdbb9a7c
KW
700/* Keep these private until we decide it was a good idea */
701#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
702
703#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
704
bdb7e3f0 705#define memBEGINs(s1, l, s2) \
30a6480c 706 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 707 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158 708#define memBEGINPs(s1, l, s2) \
30a6480c 709 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
de627158 710 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0 711#define memENDs(s1, l, s2) \
30a6480c 712 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 713 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424 714#define memENDPs(s1, l, s2) \
30a6480c 715 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
b80f8424 716 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 717#endif /* End of making macros private */
bdb7e3f0 718
062b6850
KW
719#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
720#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
721#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
722#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
723
4aada8b9
KW
724#define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
725
bbce6d69
PP
726/*
727 * Character classes.
728 *
729 * Unfortunately, the introduction of locales means that we
730 * can't trust isupper(), etc. to tell the truth. And when
731 * it comes to /\w+/ with tainting enabled, we *must* be able
732 * to trust our character classes.
733 *
734 * Therefore, the default tests in the text of Perl will be
735 * independent of locale. Any code that wants to depend on
736 * the current locale will use the tests that begin with "lc".
737 */
738
2304df62
AD
739#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
740# ifndef CTYPE256
741# define CTYPE256
742# endif
743#endif
744
954c1994 745/*
ccfc67b7 746
dcccc8ff 747=head1 Character classification
243effed
KW
748This section is about functions (really macros) that classify characters
749into types, such as punctuation versus alphabetic, etc. Most of these are
750analogous to regular expression character classes. (See
751L<perlrecharclass/POSIX Character Classes>.) There are several variants for
752each class. (Not all macros have all variants; each item below lists the
753ones valid for it.) None are affected by C<use bytes>, and only the ones
754with C<LC> in the name are affected by the current locale.
755
d713f9d9
KW
756The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
757treating it as a code point, and returns a boolean as to whether or not the
758character represented by it is (or on non-ASCII platforms, corresponds to) an
6aff1f14
KW
759ASCII character in the named class based on platform, Unicode, and Perl rules.
760If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 761
c98722a4 762Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
763with no suffix C<"_A">. This variant is used to emphasize by its name that
764only ASCII-range characters can return TRUE.
4b9734bf 765
d60679e1 766Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
767onto the platform. That is, the code points that are ASCII are unaffected,
768since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
769as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
770true when called with the code point 0xDF, which is a word character in both
4650c663 771ASCII and EBCDIC (though it represents different characters in each).
d713f9d9
KW
772If the input is a number that doesn't fit in an octet, FALSE is returned.
773(Perl's documentation uses a colloquial definition of Latin-1, to include all
774code points below 256.)
243effed 775
d713f9d9
KW
776Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
777inputs below 256, but if the code point is larger than 255, Unicode rules are
778used to determine if it is in the character class. For example,
d0da05db 779C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 780WITH MACRON in Unicode, and is a word character.
243effed 781
059703b0
KW
782Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
783but are used for UTF-8 encoded strings. The two forms are different names for
784the same thing. Each call to one of these classifies the first character of
785the string starting at C<p>. The second parameter, C<e>, points to anywhere in
786the string beyond the first character, up to one byte past the end of the
787entire string. Although both variants are identical, the suffix C<_safe> in
788one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
789provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
790C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
791some way, the program may croak, or the function may return FALSE, at the
792discretion of the implementation, and subject to change in future releases.
243effed 793
d713f9d9
KW
794Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
795but the result is based on the current locale, which is what C<LC> in the name
796stands for. If Perl can determine that the current locale is a UTF-8 locale,
797it uses the published Unicode rules; otherwise, it uses the C library function
798that gives the named classification. For example, C<isDIGIT_LC()> when not in
799a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
800returned if the input won't fit into an octet. On some platforms where the C
801library function is known to be defective, Perl changes its result to follow
802the POSIX standard's rules.
243effed 803
d713f9d9
KW
804Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
805than 256, but for larger ones it returns the Unicode classification of the code
806point.
243effed 807
059703b0
KW
808Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
809C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
810are different names for the same thing. Each call to one of these classifies
811the first character of the string starting at C<p>. The second parameter,
812C<e>, points to anywhere in the string beyond the first character, up to one
813byte past the end of the entire string. Although both variants are identical,
814the suffix C<_safe> in one name emphasizes that it will not attempt to read
815beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
816is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
817character is malformed in some way, the program may croak, or the function may
818return FALSE, at the discretion of the implementation, and subject to change in
819future releases.
ccfc67b7 820
6fdd32c3
KW
821=for apidoc Am|bool|isALPHA|UV ch
822=for apidoc_item ||isALPHA_A|UV ch
823=for apidoc_item ||isALPHA_L1|UV ch
824=for apidoc_item ||isALPHA_uvchr|UV ch
fa470d81
KW
825=for apidoc_item ||isALPHA_utf8_safe|U8 * s|U8 * end
826=for apidoc_item ||isALPHA_utf8|U8 * s|U8 * end
6fdd32c3
KW
827=for apidoc_item ||isALPHA_LC|UV ch
828=for apidoc_item ||isALPHA_LC_uvchr|UV ch
fa470d81 829=for apidoc_item ||isALPHA_LC_utf8_safe|U8 * s| U8 *end
d713f9d9
KW
830Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
831analogous to C<m/[[:alpha:]]/>.
dcccc8ff 832See the L<top of this section|/Character classification> for an explanation of
fa470d81 833the variants.
8a58bdcf 834
f16858ed
KW
835=cut
836
f1460a66 837Here and below, we add the prototypes of these macros for downstream programs
f16858ed
KW
838that would be interested in them, such as Devel::PPPort
839
6fdd32c3
KW
840=for apidoc Am|bool|isALPHANUMERIC|UV ch
841=for apidoc_item ||isALPHANUMERIC_A|UV ch
842=for apidoc_item ||isALPHANUMERIC_L1|UV ch
843=for apidoc_item ||isALPHANUMERIC_uvchr|UV ch
fa470d81
KW
844=for apidoc_item ||isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
845=for apidoc_item ||isALPHANUMERIC_utf8|U8 * s|U8 * end
6fdd32c3
KW
846=for apidoc_item ||isALPHANUMERIC_LC|UV ch
847=for apidoc_item ||isALPHANUMERIC_LC_uvchr|UV ch
fa470d81 848=for apidoc_item ||isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
6fdd32c3
KW
849=for apidoc_item ||isALNUMC|UV ch
850=for apidoc_item ||isALNUMC_A|UV ch
851=for apidoc_item ||isALNUMC_L1|UV ch
852=for apidoc_item ||isALNUMC_LC|UV ch
853=for apidoc_item ||isALNUMC_LC_uvchr|UV ch
d713f9d9
KW
854Returns a boolean indicating whether the specified character is one of
855C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
dcccc8ff 856See the L<top of this section|/Character classification> for an explanation of
fa470d81 857the variants.
15861f94 858
255b632a
KW
859A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
860this corresponds to the C language alphanumeric definition). Also
861there are the variants
862C<isALNUMC_A>, C<isALNUMC_L1>
863C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
864
6fdd32c3
KW
865=for apidoc Am|bool|isASCII|UV ch
866=for apidoc_item ||isASCII_A|UV ch
867=for apidoc_item ||isASCII_L1|UV ch
868=for apidoc_item ||isASCII_uvchr|UV ch
fa470d81
KW
869=for apidoc_item ||isASCII_utf8_safe|U8 * s|U8 * end
870=for apidoc_item ||isASCII_utf8|U8 * s|U8 * end
6fdd32c3
KW
871=for apidoc_item ||isASCII_LC|UV ch
872=for apidoc_item ||isASCII_LC_uvchr|UV ch
fa470d81 873=for apidoc_item ||isASCII_LC_utf8_safe|U8 * s| U8 *end
8a58bdcf 874Returns a boolean indicating whether the specified character is one of the 128
243effed 875characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 876On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
877character corresponds to an ASCII character. Variants C<isASCII_A()> and
878C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff 879See the L<top of this section|/Character classification> for an explanation of
fa470d81 880the variants.
059703b0
KW
881Note, however, that some platforms do not have the C library routine
882C<isascii()>. In these cases, the variants whose names contain C<LC> are the
883same as the corresponding ones without.
243effed 884
d98532ea
KW
885Also note, that because all ASCII characters are UTF-8 invariant (meaning they
886have the exact same representation (always a single byte) whether encoded in
887UTF-8 or not), C<isASCII> will give the correct results when called with any
059703b0
KW
888byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
889C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
d98532ea 890
6fdd32c3
KW
891=for apidoc Am|bool|isBLANK|UV ch
892=for apidoc_item ||isBLANK_A|UV ch
893=for apidoc_item ||isBLANK_L1|UV ch
894=for apidoc_item ||isBLANK_uvchr|UV ch
fa470d81
KW
895=for apidoc_item ||isBLANK_utf8_safe|U8 * s|U8 * end
896=for apidoc_item ||isBLANK_utf8|U8 * s|U8 * end
6fdd32c3
KW
897=for apidoc_item ||isBLANK_LC|UV ch
898=for apidoc_item ||isBLANK_LC_uvchr|UV ch
fa470d81 899=for apidoc_item ||isBLANK_LC_utf8_safe|U8 * s| U8 *end
243effed 900Returns a boolean indicating whether the specified character is a
6aff1f14 901character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff 902See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
903the variants.
904Note,
da8c1a98
KW
905however, that some platforms do not have the C library routine
906C<isblank()>. In these cases, the variants whose names contain C<LC> are
907the same as the corresponding ones without.
243effed 908
6fdd32c3
KW
909=for apidoc Am|bool|isCNTRL|UV ch
910=for apidoc_item ||isCNTRL_A|UV ch
911=for apidoc_item ||isCNTRL_L1|UV ch
912=for apidoc_item ||isCNTRL_uvchr|UV ch
fa470d81
KW
913=for apidoc_item ||isCNTRL_utf8_safe|U8 * s|U8 * end
914=for apidoc_item ||isCNTRL_utf8|U8 * s|U8 * end
6fdd32c3
KW
915=for apidoc_item ||isCNTRL_LC|UV ch
916=for apidoc_item ||isCNTRL_LC_uvchr|UV ch
fa470d81 917=for apidoc_item ||isCNTRL_LC_utf8_safe|U8 * s| U8 *end
f16858ed 918
243effed 919Returns a boolean indicating whether the specified character is a
6aff1f14 920control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff 921See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
922the variants.
923On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
924
6fdd32c3
KW
925=for apidoc Am|bool|isDIGIT|UV ch
926=for apidoc_item ||isDIGIT_A|UV ch
927=for apidoc_item ||isDIGIT_L1|UV ch
928=for apidoc_item ||isDIGIT_uvchr|UV ch
fa470d81
KW
929=for apidoc_item ||isDIGIT_utf8_safe|U8 * s|U8 * end
930=for apidoc_item ||isDIGIT_utf8|U8 * s|U8 * end
6fdd32c3
KW
931=for apidoc_item ||isDIGIT_LC|UV ch
932=for apidoc_item ||isDIGIT_LC_uvchr|UV ch
fa470d81
KW
933=for apidoc_item ||isDIGIT_LC_utf8_safe|U8 * s| U8 *end
934
2787a470 935Returns a boolean indicating whether the specified character is a
6aff1f14 936digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 937Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff 938See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
939the variants.
940
6fdd32c3
KW
941=for apidoc Am|bool|isGRAPH|UV ch
942=for apidoc_item ||isGRAPH_A|UV ch
943=for apidoc_item ||isGRAPH_L1|UV ch
944=for apidoc_item ||isGRAPH_uvchr|UV ch
fa470d81
KW
945=for apidoc_item ||isGRAPH_utf8_safe|U8 * s|U8 * end
946=for apidoc_item ||isGRAPH_utf8|U8 * s|U8 * end
6fdd32c3
KW
947=for apidoc_item ||isGRAPH_LC|UV ch
948=for apidoc_item ||isGRAPH_LC_uvchr|UV ch
fa470d81 949=for apidoc_item ||isGRAPH_LC_utf8_safe|U8 * s| U8 *end
243effed 950Returns a boolean indicating whether the specified character is a
6aff1f14 951graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 952See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
953the variants.
954
6fdd32c3
KW
955=for apidoc Am|bool|isLOWER|UV ch
956=for apidoc_item ||isLOWER_A|UV ch
957=for apidoc_item ||isLOWER_L1|UV ch
958=for apidoc_item ||isLOWER_uvchr|UV ch
fa470d81
KW
959=for apidoc_item ||isLOWER_utf8_safe|U8 * s|U8 * end
960=for apidoc_item ||isLOWER_utf8|U8 * s|U8 * end
6fdd32c3
KW
961=for apidoc_item ||isLOWER_LC|UV ch
962=for apidoc_item ||isLOWER_LC_uvchr|UV ch
fa470d81 963=for apidoc_item ||isLOWER_LC_utf8_safe|U8 * s| U8 *end
2787a470 964Returns a boolean indicating whether the specified character is a
6aff1f14 965lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff 966See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
967the variants
968
6fdd32c3
KW
969=for apidoc Am|bool|isOCTAL|UV ch
970=for apidoc_item ||isOCTAL_A|UV ch
971=for apidoc_item ||isOCTAL_L1|UV ch
2787a470 972Returns a boolean indicating whether the specified character is an
6aff1f14 973octal digit, [0-7].
243effed
KW
974The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
975C<isOCTAL>.
976
6fdd32c3
KW
977=for apidoc Am|bool|isPUNCT|UV ch
978=for apidoc_item ||isPUNCT_A|UV ch
979=for apidoc_item ||isPUNCT_L1|UV ch
980=for apidoc_item ||isPUNCT_uvchr|UV ch
fa470d81
KW
981=for apidoc_item ||isPUNCT_utf8_safe|U8 * s|U8 * end
982=for apidoc_item ||isPUNCT_utf8|U8 * s|U8 * end
6fdd32c3
KW
983=for apidoc_item ||isPUNCT_LC|UV ch
984=for apidoc_item ||isPUNCT_LC_uvchr|UV ch
fa470d81 985=for apidoc_item ||isPUNCT_LC_utf8_safe|U8 * s| U8 *end
243effed 986Returns a boolean indicating whether the specified character is a
6aff1f14
KW
987punctuation character, analogous to C<m/[[:punct:]]/>.
988Note that the definition of what is punctuation isn't as
243effed
KW
989straightforward as one might desire. See L<perlrecharclass/POSIX Character
990Classes> for details.
dcccc8ff 991See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
992the variants.
993
6fdd32c3
KW
994=for apidoc Am|bool|isSPACE|UV ch
995=for apidoc_item ||isSPACE_A|UV ch
996=for apidoc_item ||isSPACE_L1|UV ch
997=for apidoc_item ||isSPACE_uvchr|UV ch
fa470d81
KW
998=for apidoc_item ||isSPACE_utf8_safe|U8 * s|U8 * end
999=for apidoc_item ||isSPACE_utf8|U8 * s|U8 * end
6fdd32c3
KW
1000=for apidoc_item ||isSPACE_LC|UV ch
1001=for apidoc_item ||isSPACE_LC_uvchr|UV ch
fa470d81 1002=for apidoc_item ||isSPACE_LC_utf8_safe|U8 * s| U8 *end
2787a470 1003Returns a boolean indicating whether the specified character is a
6aff1f14 1004whitespace character. This is analogous
398d098a 1005to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 1006this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
1007locale forms of this macro (the ones with C<LC> in their names) matched
1008precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
1009in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
1010(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff 1011See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1012the variants.
1013
6fdd32c3
KW
1014=for apidoc Am|bool|isPSXSPC|UV ch
1015=for apidoc_item ||isPSXSPC_A|UV ch
1016=for apidoc_item ||isPSXSPC_L1|UV ch
1017=for apidoc_item ||isPSXSPC_uvchr|UV ch
fa470d81
KW
1018=for apidoc_item ||isPSXSPC_utf8_safe|U8 * s|U8 * end
1019=for apidoc_item ||isPSXSPC_utf8|U8 * s|U8 * end
6fdd32c3
KW
1020=for apidoc_item ||isPSXSPC_LC|UV ch
1021=for apidoc_item ||isPSXSPC_LC_uvchr|UV ch
fa470d81 1022=for apidoc_item ||isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
398d098a 1023(short for Posix Space)
779cf272
KW
1024Starting in 5.18, this is identical in all its forms to the
1025corresponding C<isSPACE()> macros.
398d098a
KW
1026The locale forms of this macro are identical to their corresponding
1027C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
1028non-locale forms differ from their C<isSPACE()> forms only in that the
1029C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
1030Otherwise they are identical. Thus this macro is analogous to what
1031C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 1032See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1033the variants.
1034
6fdd32c3
KW
1035=for apidoc Am|bool|isUPPER|UV ch
1036=for apidoc_item ||isUPPER_A|UV ch
1037=for apidoc_item ||isUPPER_L1|UV ch
1038=for apidoc_item ||isUPPER_uvchr|UV ch
fa470d81
KW
1039=for apidoc_item ||isUPPER_utf8_safe|U8 * s|U8 * end
1040=for apidoc_item ||isUPPER_utf8|U8 * s|U8 * end
6fdd32c3
KW
1041=for apidoc_item ||isUPPER_LC|UV ch
1042=for apidoc_item ||isUPPER_LC_uvchr|UV ch
fa470d81 1043=for apidoc_item ||isUPPER_LC_utf8_safe|U8 * s| U8 *end
2787a470 1044Returns a boolean indicating whether the specified character is an
6aff1f14 1045uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 1046See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1047the variants.
1048
6fdd32c3
KW
1049=for apidoc Am|bool|isPRINT|UV ch
1050=for apidoc_item ||isPRINT_A|UV ch
1051=for apidoc_item ||isPRINT_L1|UV ch
1052=for apidoc_item ||isPRINT_uvchr|UV ch
fa470d81
KW
1053=for apidoc_item ||isPRINT_utf8_safe|U8 * s|U8 * end
1054=for apidoc_item ||isPRINT_utf8|U8 * s|U8 * end
6fdd32c3
KW
1055=for apidoc_item ||isPRINT_LC|UV ch
1056=for apidoc_item ||isPRINT_LC_uvchr|UV ch
fa470d81 1057=for apidoc_item ||isPRINT_LC_utf8_safe|U8 * s| U8 *end
8eea39dd 1058Returns a boolean indicating whether the specified character is a
6aff1f14 1059printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff 1060See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1061the variants.
1062
6fdd32c3
KW
1063=for apidoc Am|bool|isWORDCHAR|UV ch
1064=for apidoc_item ||isWORDCHAR_A|UV ch
1065=for apidoc_item ||isWORDCHAR_L1|UV ch
1066=for apidoc_item ||isWORDCHAR_uvchr|UV ch
fa470d81
KW
1067=for apidoc_item ||isWORDCHAR_utf8_safe|U8 * s|U8 * end
1068=for apidoc_item ||isWORDCHAR_utf8|U8 * s|U8 * end
6fdd32c3
KW
1069=for apidoc_item ||isWORDCHAR_LC|UV ch
1070=for apidoc_item ||isWORDCHAR_LC_uvchr|UV ch
fa470d81 1071=for apidoc_item ||isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
6fdd32c3
KW
1072=for apidoc_item ||isALNUM|UV ch
1073=for apidoc_item ||isALNUM_A|UV ch
1074=for apidoc_item ||isALNUM_LC|UV ch
1075=for apidoc_item ||isALNUM_LC_uvchr|UV ch
243effed
KW
1076Returns a boolean indicating whether the specified character is a character
1077that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1078in a regular expression. A word character is an alphabetic character, a
1079decimal digit, a connecting punctuation character (such as an underscore), or
1080a "mark" character that attaches to one of those (like some sort of accent).
1081C<isALNUM()> is a synonym provided for backward compatibility, even though a
1082word character includes more than the standard C language meaning of
1083alphanumeric.
dcccc8ff 1084See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1085the variants.
1086C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1087C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and
1088C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
1089include the platform's native underscore.
1090
6fdd32c3
KW
1091=for apidoc Am|bool|isXDIGIT|UV ch
1092=for apidoc_item ||isXDIGIT_A|UV ch
1093=for apidoc_item ||isXDIGIT_L1|UV ch
1094=for apidoc_item ||isXDIGIT_uvchr|UV ch
fa470d81
KW
1095=for apidoc_item ||isXDIGIT_utf8_safe|U8 * s|U8 * end
1096=for apidoc_item ||isXDIGIT_utf8|U8 * s|U8 * end
6fdd32c3
KW
1097=for apidoc_item ||isXDIGIT_LC|UV ch
1098=for apidoc_item ||isXDIGIT_LC_uvchr|UV ch
fa470d81 1099=for apidoc_item ||isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
8a58bdcf 1100Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
1101digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1102and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff 1103See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1104the variants.
1105
6fdd32c3
KW
1106=for apidoc Am|bool|isIDFIRST|UV ch
1107=for apidoc_item ||isIDFIRST_A|UV ch
1108=for apidoc_item ||isIDFIRST_L1|UV ch
1109=for apidoc_item ||isIDFIRST_uvchr|UV ch
fa470d81
KW
1110=for apidoc_item ||isIDFIRST_utf8_safe|U8 * s|U8 * end
1111=for apidoc_item ||isIDFIRST_utf8|U8 * s|U8 * end
6fdd32c3
KW
1112=for apidoc_item ||isIDFIRST_LC|UV ch
1113=for apidoc_item ||isIDFIRST_LC_uvchr|UV ch
fa470d81 1114=for apidoc_item ||isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
3c3ecf18
KW
1115Returns a boolean indicating whether the specified character can be the first
1116character of an identifier. This is very close to, but not quite the same as
1117the official Unicode property C<XID_Start>. The difference is that this
1118returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff 1119See the L<top of this section|/Character classification> for an explanation of
fa470d81
KW
1120the variants.
1121
6fdd32c3
KW
1122=for apidoc Am|bool|isIDCONT|UV ch
1123=for apidoc_item ||isIDCONT_A|UV ch
1124=for apidoc_item ||isIDCONT_L1|UV ch
1125=for apidoc_item ||isIDCONT_uvchr|UV ch
fa470d81
KW
1126=for apidoc_item ||isIDCONT_utf8_safe|U8 * s|U8 * end
1127=for apidoc_item ||isIDCONT_utf8|U8 * s|U8 * end
6fdd32c3
KW
1128=for apidoc_item ||isIDCONT_LC|UV ch
1129=for apidoc_item ||isIDCONT_LC_uvchr|UV ch
fa470d81 1130=for apidoc_item ||isIDCONT_LC_utf8_safe|U8 * s| U8 *end
3c3ecf18
KW
1131Returns a boolean indicating whether the specified character can be the
1132second or succeeding character of an identifier. This is very close to, but
1133not quite the same as the official Unicode property C<XID_Continue>. The
1134difference is that this returns true only if the input character also matches
dcccc8ff 1135L</isWORDCHAR>. See the L<top of this section|/Character classification> for
fa470d81 1136an explanation of the variants.
f16858ed 1137
51b56f5c 1138=for apidoc_section Numeric Functions
8eea39dd 1139
95a59cab 1140=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 1141Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
1142Behaviour is only well defined when isXDIGIT(*str) is true.
1143
e7c1e6c1 1144=head1 Character case changing
21da7284
KW
1145Perl uses "full" Unicode case mappings. This means that converting a single
1146character to another case may result in a sequence of more than one character.
1147For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1148character sequence C<SS>. This presents some complications The lowercase of
1149all characters in the range 0..255 is a single character, and thus
1150C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1151return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1152an API that does allow every possible legal result to be returned.) Likewise
1153no other function that is crippled by not being able to give the correct
1154results for the full range of possible inputs has been implemented here.
e7c1e6c1 1155
d713f9d9 1156=for apidoc Am|U8|toUPPER|int ch
1f607577
KW
1157Converts the specified character to uppercase. If the input is anything but an
1158ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 1159C<toUPPER_A> is equivalent.
954c1994 1160
d0da05db
KW
1161=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1162Converts the code point C<cp> to its uppercase version, and
1163stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1164point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1165that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1166bytes since the uppercase version may be longer than the original character.
1167
1168The first code point of the uppercased version is returned
21da7284
KW
1169(but note, as explained at L<the top of this section|/Character case
1170changing>, that there may be more.)
1f607577 1171
059703b0 1172=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141 1173=for apidoc_item toUPPER_utf8_safe
a239b1e2
KW
1174Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1175extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
1176stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1177that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1178bytes since the uppercase version may be longer than the original character.
1179
1180The first code point of the uppercased version is returned
21da7284
KW
1181(but note, as explained at L<the top of this section|/Character case
1182changing>, that there may be more).
1f607577 1183
059703b0
KW
1184It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1185S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1186the UTF-8 for the input character is malformed in some way, the program may
1187croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1188of the implementation, and subject to change in future releases.
a239b1e2 1189
912c8141 1190C<toUPPER_utf8_safe> is now just a different spelling of plain C<toUPPER_utf8>
1f607577 1191
25200305
KW
1192=for apidoc Am|U8|toFOLD|U8 ch
1193Converts the specified character to foldcase. If the input is anything but an
1194ASCII uppercase character, that input character itself is returned. Variant
1195C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 1196Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 1197
d0da05db
KW
1198=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1199Converts the code point C<cp> to its foldcase version, and
1200stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1201point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1202that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1203bytes since the foldcase version may be longer than the original character.
1204
1205The first code point of the foldcased version is returned
21da7284
KW
1206(but note, as explained at L<the top of this section|/Character case
1207changing>, that there may be more).
1f607577 1208
059703b0 1209=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141 1210=for apidoc_item toFOLD_utf8_safe
a239b1e2
KW
1211Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1212extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
1213stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1214that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1215bytes since the foldcase version may be longer than the original character.
1216
1217The first code point of the foldcased version is returned
21da7284
KW
1218(but note, as explained at L<the top of this section|/Character case
1219changing>, that there may be more).
1f607577 1220
059703b0 1221It will not attempt
a239b1e2
KW
1222to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1223true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1224input character is malformed in some way, the program may croak, or the
1225function may return the REPLACEMENT CHARACTER, at the discretion of the
1226implementation, and subject to change in future releases.
1227
912c8141 1228C<toFOLD_utf8_safe> is now just a different spelling of plain C<toFOLD_utf8>
1f607577 1229
3cb048e5
KW
1230=for apidoc Am|UV|toLOWER|UV cp
1231=for apidoc_item |UV|toLOWER_A|UV cp
1232=for apidoc_item |UV|toLOWER_L1|UV cp
1233=for apidoc_item |UV|toLOWER_LATIN1|UV cp
1234=for apidoc_item |UV|toLOWER_LC|UV cp
1235=for apidoc_item |UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1236=for apidoc_item |UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1237=for apidoc_item |UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1238
1239These all return the lowercase of a character. The differences are what domain
1240they operate on, and whether the input is specified as a code point (those
1241forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter
1242case, the code point to use is the first one in the buffer of UTF-8 encoded
1243code points, delineated by the arguments S<C<p .. e - 1>>.
1244
1245C<toLOWER> and C<toLOWER_A> are synonyms of each other. They return the
1246lowercase of any uppercase ASCII-range code point. All other inputs are
1247returned unchanged. Since these are macros, the input type may be any integral
1248one, and the output will occupy the same number of bits as the input.
1249
1250C<toLOWER_L1> and C<toLOWER_LATIN1> are synonyms of each other. They behave
1251identically as C<toLOWER> for ASCII-range input. But additionally will return
1252the lowercase of any uppercase code point in the entire 0..255 range, assuming
1253a Latin-1 encoding (or the EBCDIC equivalent on such platforms).
1254
1255C<toLOWER_LC> returns the lowercase of the input code point according to the
1256rules of the current POSIX locale. Input code points outside the range 0..255
1257are returned unchanged.
1258
1259C<toLOWER_uvchr> returns the lowercase of any Unicode code point. The return
1260value is identical to that of C<toLOWER_L1> for input code points in the 0..255
1261range. The lowercase of the vast majority of Unicode code points is the same
1262as the code point itself. For these, and for code points above the legal
1263Unicode maximum, this returns the input code point unchanged. It additionally
1264stores the UTF-8 of the result into the buffer beginning at C<s>, and its
1265length in bytes into C<*lenp>. The caller must have made C<s> large enough to
1266contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow.
1267
1268NOTE: the lowercase of a code point may be more than one code point. The
1269return value of this function is only the first of these. The entire lowercase
1270is returned in C<s>. To determine if the result is more than a single code
1271point, you can do something like this:
1272
1273 uc = toLOWER_uvchr(cp, s, &len);
1274 if (len > UTF8SKIP(s)) { is multiple code points }
1275 else { is a single code point }
1276
1277C<toLOWER_utf8> and C<toLOWER_utf8_safe> are synonyms of each other. The only
1278difference between these and C<toLOWER_uvchr> is that the source for these is
1279encoded in UTF-8, instead of being a code point. It is passed as a buffer
1280starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p>
1281buffer may certainly contain more than one code point; but only the first one
1282(up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is
1283malformed in some way, the program may croak, or the function may return the
1284REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to
1285change in future releases.
1f607577 1286
25200305
KW
1287=for apidoc Am|U8|toTITLE|U8 ch
1288Converts the specified character to titlecase. If the input is anything but an
1289ASCII lowercase character, that input character itself is returned. Variant
b7d90381 1290C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 1291range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 1292not a concept used in locale handling, so there is no functionality for that.)
25200305 1293
d0da05db
KW
1294=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1295Converts the code point C<cp> to its titlecase version, and
1296stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1297point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1298that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1299bytes since the titlecase version may be longer than the original character.
1300
1301The first code point of the titlecased version is returned
21da7284
KW
1302(but note, as explained at L<the top of this section|/Character case
1303changing>, that there may be more).
1f607577 1304
059703b0 1305=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141
KW
1306=for apidoc_item toTITLE_utf8_safe
1307Convert the first UTF-8 encoded character in the sequence starting at C<p> and
a239b1e2 1308extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1309stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1310that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1311bytes since the titlecase version may be longer than the original character.
1312
1313The first code point of the titlecased version is returned
21da7284
KW
1314(but note, as explained at L<the top of this section|/Character case
1315changing>, that there may be more).
1f607577 1316
059703b0 1317It will not attempt
a239b1e2
KW
1318to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1319true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1320input character is malformed in some way, the program may croak, or the
1321function may return the REPLACEMENT CHARACTER, at the discretion of the
1322implementation, and subject to change in future releases.
1323
912c8141 1324C<toTITLE_utf8_safe> is now just a different spelling of plain C<toTITLE_utf8>
1f607577 1325
954c1994 1326=cut
353c9b6f 1327
d0da05db 1328XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1329really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1330and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1331
8a58bdcf 1332Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1333patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1334
954c1994
GS
1335*/
1336
8f5283f4
KW
1337/*
1338 void below because that's the best fit, and works for Devel::PPPort
51b56f5c 1339=for apidoc_section Integer configuration values
545bca17 1340=for apidoc AyT||WIDEST_UTYPE
8f5283f4
KW
1341
1342Yields the widest unsigned integer type on the platform, currently either
326c768d 1343C<U32> or C<U64>. This can be used in declarations such as
8f5283f4
KW
1344
1345 WIDEST_UTYPE my_uv;
1346
1347or casts
1348
1349 my_uv = (WIDEST_UTYPE) val;
1350
1351=cut
1352
1353*/
de9e2639
KW
1354#ifdef QUADKIND
1355# define WIDEST_UTYPE U64
7c062697
KW
1356#else
1357# define WIDEST_UTYPE U32
1358#endif
1359
3912bc88
KW
1360/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1361 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1362 * bits of information are lost even on a 64-bit machine, but to get the
1363 * compiler to optimize it out if possible. This is because Configure makes
1364 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1365 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1366 *
1367 * For Coverity, be always true, because otherwise Coverity thinks
1368 * it finds several expressions that are always true, independent
1369 * of operands. Well, they are, but that is kind of the point.
220c71bf 1370 */
7e75d1a1 1371#ifndef __COVERITY__
6c5b02ac
KW
1372 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1373 * pointer) */
1374#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1375 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1376#else
1377#define FITS_IN_8_BITS(c) (1)
1378#endif
cf301eb7 1379
45f4bb73 1380/* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
833b0f46 1381 * Written this way so that after optimization, only one conditional test is
76d3ad4c
KW
1382 * needed. (The NV casts stop any warnings about comparison always being true
1383 * if called with an unsigned. The cast preserves the sign, which is all we
1384 * care about.) */
1385#define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1386 __ASSERT_((NV) (n) >= 0) \
1387 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
833b0f46 1388
94250c4f
KW
1389/* Returns true if c is in the range l..u, where 'l' is non-negative
1390 * Written this way so that after optimization, only one conditional test is
4758c20d 1391 * needed. */
1eaefa6e 1392#define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
4758c20d 1393 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
4758c20d
KW
1394 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1395 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
45f4bb73 1396 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
305fe86e 1397
41f43cc2 1398#ifdef EBCDIC
b6340bd0 1399# ifndef _ALL_SOURCE
0852beac
KW
1400 /* The native libc isascii() et.al. functions return the wrong results
1401 * on at least z/OS unless this is defined. */
b6340bd0
KW
1402# error _ALL_SOURCE should probably be defined
1403# endif
41f43cc2 1404#else
0852beac
KW
1405 /* There is a simple definition of ASCII for ASCII platforms. But the
1406 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1407 * other macros below.
3f3c579d
KW
1408 *
1409 * The cast here is used instead of '(c) >= 0', because some compilers emit
1410 * a warning that that test is always true when the parameter is an
1411 * unsigned type. khw supposes that it could be written as
1412 * && ((c) == '\0' || (c) > 0)
1413 * to avoid the message, but the cast will likely avoid extra branches even
1414 * with stupid compilers.
1415 *
1416 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1417 * a pointer) */
9c903d59 1418# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1419#endif
1420
38694112
KW
1421/* Take the eight possible bit patterns of the lower 3 bits and you get the
1422 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1423 * can be ignored. If the rest match '0', we have an octal */
1424#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1425
9fb1bf9d 1426#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1427
a500dc72
KW
1428/* Character class numbers. For internal core Perl use only. The ones less
1429 * than 32 are used in PL_charclass[] and the ones up through the one that
1430 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1431 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1432 * their actual definitions are here. If that file has a name not used here,
1433 * it won't compile.
1709d539
KW
1434 *
1435 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561 1436 * frequency of their use. This gives a slight edge to exiting a loop earlier
58a3ba2c
KW
1437 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1438 * real Posix character class, and some (small) inefficiencies in regular
1439 * expression handling would be introduced by putting it in the middle of those
1440 * that are. Also, cntrl and ascii come after the others as it may be useful
1441 * to group these which have no members that match above Latin1, (or above
1442 * ASCII in the latter case) */
1443
1709d539
KW
1444# define _CC_WORDCHAR 0 /* \w and [:word:] */
1445# define _CC_DIGIT 1 /* \d and [:digit:] */
1446# define _CC_ALPHA 2 /* [:alpha:] */
1447# define _CC_LOWER 3 /* [:lower:] */
1448# define _CC_UPPER 4 /* [:upper:] */
1449# define _CC_PUNCT 5 /* [:punct:] */
1450# define _CC_PRINT 6 /* [:print:] */
15861f94 1451# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1452# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1453# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
779cf272 1454# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1455# define _CC_BLANK 11 /* [:blank:] */
1456# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1457# define _CC_CNTRL 13 /* [:cntrl:] */
1458# define _CC_ASCII 14 /* [:ascii:] */
1459# define _CC_VERTSPACE 15 /* \v */
1709d539 1460
a0947d7b
KW
1461# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1462
1709d539 1463/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1464 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1465# define _CC_IDFIRST 16
1466# define _CC_CHARNAME_CONT 17
1467# define _CC_NONLATIN1_FOLD 18
1468# define _CC_NONLATIN1_SIMPLE_FOLD 19
1469# define _CC_QUOTEMETA 20
1470# define _CC_NON_FINAL_FOLD 21
1471# define _CC_IS_IN_SOME_FOLD 22
2ae9030c
KW
1472# define _CC_BINDIGIT 23
1473# define _CC_OCTDIGIT 24
1474# define _CC_MNEMONIC_CNTRL 25
073c22b3
KW
1475
1476/* This next group is only used on EBCDIC platforms, so theoretically could be
1477 * shared with something entirely different that's only on ASCII platforms */
abb8abf6 1478# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
0654f0ab 1479/* Unused: 26-30
f4cdb42c
KW
1480 * If more bits are needed, one could add a second word for non-64bit
1481 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1482 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1483 * is used only for optimization (as of this writing), and differs in the
1484 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1485 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1486 * would just cause /i regexes which match them to run less efficiently.
1487 * Similarly the EBCDIC-only bits are used just for speed, and could be
1488 * replaced by other means */
96ac0975 1489
3a371f2f
KW
1490#if defined(PERL_CORE) || defined(PERL_EXT)
1491/* An enum version of the character class numbers, to help compilers
1492 * optimize */
1493typedef enum {
3a371f2f 1494 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1495 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1496 _CC_ENUM_ASCII = _CC_ASCII,
1497 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1498 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1499 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1500 _CC_ENUM_DIGIT = _CC_DIGIT,
1501 _CC_ENUM_GRAPH = _CC_GRAPH,
1502 _CC_ENUM_LOWER = _CC_LOWER,
1503 _CC_ENUM_PRINT = _CC_PRINT,
1504 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1505 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1506 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1507 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1508 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1509 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1510} _char_class_number;
1511#endif
1512
86f72d56 1513#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1514
6635f04f 1515START_EXTERN_C
96ac0975
NC
1516# ifdef DOINIT
1517EXTCONST U32 PL_charclass[] = {
1518# include "l1_char_class_tab.h"
1519};
1520
1521# else /* ! DOINIT */
1522EXTCONST U32 PL_charclass[];
1523# endif
6635f04f 1524END_EXTERN_C
96ac0975 1525
265c1f46 1526 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1527# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1528
1529 /* For internal core Perl use only: the base macro for defining macros like
1530 * isALPHA */
ff7ecfc3 1531# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1532 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1533
f4cdb42c
KW
1534 /* The mask for the _A versions of the macros; it just adds in the bit for
1535 * ASCII. */
1536# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1537
4650c663
KW
1538 /* For internal core Perl use only: the base macro for defining macros like
1539 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1540 * the ASCII bit are present */
b7d90381
KW
1541# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1542 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1543 == _CC_mask_A(classnum)))
f4cdb42c 1544
26c1d9d8
KW
1545/* On ASCII platforms certain classes form a single range. It's faster to
1546 * special case these. isDIGIT is a single range on all platforms */
b877c1ff
KW
1547# ifdef EBCDIC
1548# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1549# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1550# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1551# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1552# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1553# else
26c1d9d8 1554 /* By folding the upper and lowercase, we can use a single range */
b877c1ff 1555# define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
26c1d9d8 1556# define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
b877c1ff
KW
1557# define isLOWER_A(c) inRANGE(c, 'a', 'z')
1558# define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1559# define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1560# endif
15861f94 1561# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1562# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1563# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
b877c1ff 1564# define isDIGIT_A(c) inRANGE(c, '0', '9')
f4cdb42c
KW
1565# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1566# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
f4cdb42c 1567# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1568# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1569 */
d95f8b6a 1570# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1571# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1572# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1573# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1574
1575 /* continuation character for legal NAME in \N{NAME} */
1576# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1577
1578# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1579# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1580# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1581# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1582# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1583# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1584# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1585# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1586# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1587# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1588
0852beac
KW
1589# ifdef EBCDIC
1590# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1591# endif
1592
f12c0118
KW
1593 /* Participates in a single-character fold with a character above 255 */
1594# 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)))
1595
1596 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1597# 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 1598
4eeeb416 1599# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1600# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1601 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1602# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1603 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
5e6ebb12
KW
1604
1605/* is c a control character for which we have a mnemonic? */
1606# if defined(PERL_CORE) || defined(PERL_EXT)
1607# define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1608# endif
687c8d01 1609#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1610
1611 /* If we don't have perl.h, we are compiling a utility program. Below we
1612 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1613 * to it. Most are coded based on first principles. These are written to
74665a89 1614 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
182c4ace 1615# define isDIGIT_A(c) inRANGE(c, '0', '9')
0852beac 1616# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1617# define isSPACE_A(c) (isBLANK_A(c) \
1618 || (c) == '\n' \
1619 || (c) == '\r' \
1620 || (c) == '\v' \
0852beac 1621 || (c) == '\f')
74665a89
KW
1622 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1623 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1624 * performance (if optimization isn't on), and allow the same code to be
1625 * used for both platform types */
182c4ace
KW
1626# define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1627 || inRANGE((c), 'j', 'r') \
1628 || inRANGE((c), 's', 'z')
1629# define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1630 || inRANGE((c), 'J', 'R') \
1631 || inRANGE((c), 'S', 'Z')
a4d7a999
KW
1632# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1633# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1634# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1635# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
182c4ace
KW
1636# define isXDIGIT_A(c) ( isDIGIT_A(c) \
1637 || inRANGE((c), 'a', 'f') \
1638 || inRANGE((c), 'A', 'F')
74665a89
KW
1639# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1640 || (c) == '#' || (c) == '$' || (c) == '%' \
1641 || (c) == '&' || (c) == '\'' || (c) == '(' \
1642 || (c) == ')' || (c) == '*' || (c) == '+' \
1643 || (c) == ',' || (c) == '.' || (c) == '/' \
1644 || (c) == ':' || (c) == ';' || (c) == '<' \
1645 || (c) == '=' || (c) == '>' || (c) == '?' \
1646 || (c) == '@' || (c) == '[' || (c) == '\\' \
1647 || (c) == ']' || (c) == '^' || (c) == '_' \
1648 || (c) == '`' || (c) == '{' || (c) == '|' \
1649 || (c) == '}' || (c) == '~')
1650# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1651# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1652
0852beac 1653# ifdef EBCDIC
74665a89
KW
1654 /* The below is accurate for the 3 EBCDIC code pages traditionally
1655 * supported by perl. The only difference between them in the controls
1656 * is the position of \n, and that is represented symbolically below */
1657# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1658 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1659 || (c) == '\t' || (c) == '\v' \
182c4ace 1660 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
8ec0a736 1661 || (c) == 7F /* U+7F DEL */ \
182c4ace
KW
1662 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1663 DC[1-3] */ \
74665a89
KW
1664 || (c) == 0x18 /* U+18 CAN */ \
1665 || (c) == 0x19 /* U+19 EOM */ \
182c4ace 1666 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
74665a89
KW
1667 || (c) == 0x26 /* U+17 ETB */ \
1668 || (c) == 0x27 /* U+1B ESC */ \
1669 || (c) == 0x2D /* U+05 ENQ */ \
1670 || (c) == 0x2E /* U+06 ACK */ \
1671 || (c) == 0x32 /* U+16 SYN */ \
1672 || (c) == 0x37 /* U+04 EOT */ \
1673 || (c) == 0x3C /* U+14 DC4 */ \
1674 || (c) == 0x3D /* U+15 NAK */ \
1675 || (c) == 0x3F)/* U+1A SUB */
0852beac 1676# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1677# else /* isASCII is already defined for ASCII platforms, so can use that to
1678 define isCNTRL */
1679# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1680# endif
1681
3ffc8c70 1682 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1683 * during debugging, and it seems best to keep them. We may be called
1684 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1685 * do anything anyway, so make it not a problem */
1686# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1687# define NATIVE_TO_LATIN1(ch) (ch)
1688# endif
3ded5eb0
KW
1689# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1690# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1691# define isBLANK_L1(c) (isBLANK_A(c) \
1692 || (FITS_IN_8_BITS(c) \
1693 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1694# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1695# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1696# define isLOWER_L1(c) (isLOWER_A(c) \
1697 || (FITS_IN_8_BITS(c) \
ae683a5f 1698 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1699 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1700 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1701 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1702 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1703# define isPRINT_L1(c) (isPRINT_A(c) \
1704 || (FITS_IN_8_BITS(c) \
1705 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1706# define isPUNCT_L1(c) (isPUNCT_A(c) \
1707 || (FITS_IN_8_BITS(c) \
ae683a5f 1708 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1709 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1710 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1711 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1712 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1713 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1714 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1715# define isSPACE_L1(c) (isSPACE_A(c) \
1716 || (FITS_IN_8_BITS(c) \
ae683a5f 1717 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1718 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1719# define isUPPER_L1(c) (isUPPER_A(c) \
1720 || (FITS_IN_8_BITS(c) \
182c4ace
KW
1721 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1722 0xC0, 0xDE) \
3ded5eb0
KW
1723 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1724# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1725# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1726# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1727 || isBLANK_L1(c) \
1728 || (c) == '-' \
1729 || (c) == '(' \
1730 || (c) == ')')
1731 /* The following are not fully accurate in the above-ASCII range. I (khw)
1732 * don't think it's necessary to be so for the purposes where this gets
1733 * compiled */
1734# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1735# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1736
1737 /* And these aren't accurate at all. They are useful only for above
1738 * Latin1, which utilities and bootstrapping don't deal with */
1739# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1740# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1741# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1742
1743 /* Many of the macros later in this file are defined in terms of these. By
1744 * implementing them with a function, which converts the class number into
1745 * a call to the desired macro, all of the later ones work. However, that
1746 * function won't be actually defined when building a utility program (no
1747 * perl.h), and so a compiler error will be generated if one is attempted
1748 * to be used. And the above-Latin1 code points require Unicode tables to
1749 * be present, something unlikely to be the case when bootstrapping */
1750# define _generic_isCC(c, classnum) \
1751 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1752# define _generic_isCC_A(c, classnum) \
1753 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1754#endif /* End of no perl.h H_PERL */
8a58bdcf 1755
e66b99e9
KW
1756#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1757#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1758#define isASCII_A(c) isASCII(c)
1759#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1760#define isBLANK(c) isBLANK_A(c)
1761#define isCNTRL(c) isCNTRL_A(c)
1762#define isDIGIT(c) isDIGIT_A(c)
1763#define isGRAPH(c) isGRAPH_A(c)
1764#define isIDFIRST(c) isIDFIRST_A(c)
1765#define isLOWER(c) isLOWER_A(c)
1766#define isPRINT(c) isPRINT_A(c)
779cf272 1767#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1768#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1769#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1770#define isPUNCT(c) isPUNCT_A(c)
1771#define isSPACE(c) isSPACE_A(c)
1772#define isUPPER(c) isUPPER_A(c)
1773#define isWORDCHAR(c) isWORDCHAR_A(c)
1774#define isXDIGIT(c) isXDIGIT_A(c)
1775
1776/* ASCII casing. These could also be written as
1777 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1778 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1779 which uses table lookup and mask instead of subtraction. (This would
c5e9991e
KW
1780 work because the _MOD does not apply in the ASCII range).
1781
1782 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1783 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1784 there non-ASCII invariants, and all of them are controls.) */
68067e4e
DM
1785#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1786#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1787
25200305
KW
1788/* In the ASCII range, these are equivalent to what they're here defined to be.
1789 * But by creating these definitions, other code doesn't have to be aware of
c5e9991e
KW
1790 * this detail. Actually this works for all UTF-8 invariants, not just the
1791 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
25200305 1792#define toFOLD(c) toLOWER(c)
25200305
KW
1793#define toTITLE(c) toUPPER(c)
1794
c753c8d3
KW
1795#define toLOWER_A(c) toLOWER(c)
1796#define toUPPER_A(c) toUPPER(c)
25200305
KW
1797#define toFOLD_A(c) toFOLD(c)
1798#define toTITLE_A(c) toTITLE(c)
1a0901db 1799
4650c663 1800/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1801#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1802 ? (c) \
f4cd282c 1803 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1804#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1805
1a0901db 1806/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1807 * all mapped to one of them, and these need special handling; returns the
1808 * input itself if is out-of-range */
b2bf251f 1809#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1810 ? (c) \
f4cd282c 1811 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1812#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1813
beab9ebe
KW
1814/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1815
1816/* For internal core Perl use only: the base macro for defining macros like
1817 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1818 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1819 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1820 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1821 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1822 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1823 * code points under locale rules, use the macros based on _generic_LC_uvchr
1824 * instead of this. */
beab9ebe
KW
1825#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1826 (! FITS_IN_8_BITS(c) \
1827 ? 0 \
31f05a37
KW
1828 : IN_UTF8_CTYPE_LOCALE \
1829 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1830 : cBOOL(non_utf8))
1831
1832/* For internal core Perl use only: a helper macro for defining macros like
1833 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1834 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1835 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1836 * _generic_LC_base for more info */
1837#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1838 _generic_LC_base(c,utf8_locale_classnum, \
1839 non_utf8_func( (_LC_CAST) (c)))
1840
1841/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1842 * 'c' is the platform's native underscore character */
1843#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1844 _generic_LC_base(c, utf8_locale_classnum, \
1845 (non_utf8_func( (_LC_CAST) (c)) \
1846 || (char)(c) == '_'))
1847
1848/* These next three are also for internal core Perl use only: case-change
247985d4
KW
1849 * helper macros. The reason for using the PL_latin arrays is in case the
1850 * system function is defective; it ensures uniform results that conform to the
b257a28c 1851 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
beab9ebe
KW
1852#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1853 ? (c) \
31f05a37
KW
1854 : (IN_UTF8_CTYPE_LOCALE) \
1855 ? PL_latin1_lc[ (U8) (c) ] \
5a10328c 1856 : (cast)function((cast)(c)))
beab9ebe 1857
31f05a37
KW
1858/* Note that the result can be larger than a byte in a UTF-8 locale. It
1859 * returns a single value, so can't adequately return the upper case of LATIN
1860 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1861 * values "SS"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1862 * otherwise returns its input. It does not handle the anomalies in UTF-8
1863 * Turkic locales. */
beab9ebe
KW
1864#define _generic_toUPPER_LC(c, function, cast) \
1865 (! FITS_IN_8_BITS(c) \
1866 ? (c) \
31f05a37 1867 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1868 ? (cast)function((cast)(c)) \
31f05a37
KW
1869 : ((((U8)(c)) == MICRO_SIGN) \
1870 ? GREEK_CAPITAL_LETTER_MU \
1871 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1872 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1873 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1874 ? (__ASSERT_(0) (c)) \
1875 : PL_mod_latin1_uc[ (U8) (c) ])))))
1876
1877/* Note that the result can be larger than a byte in a UTF-8 locale. It
1878 * returns a single value, so can't adequately return the fold case of LATIN
1879 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1880 * values "ss"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1881 * otherwise returns its input. It does not handle the anomalies in UTF-8
1882 * Turkic locales */
beab9ebe 1883#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1884 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1885 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1886 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1887 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1888 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1889
84061b6a 1890/* Use the libc versions for these if available. */
f05550c0 1891#if defined(HAS_ISASCII)
84061b6a
KW
1892# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1893#else
1894# define isASCII_LC(c) isASCII(c)
1895#endif
1896
f05550c0 1897#if defined(HAS_ISBLANK)
84061b6a 1898# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1899#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1900# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1901#endif
1902
f05550c0 1903#define _LC_CAST U8
bbce6d69 1904
f05550c0 1905#ifdef WIN32
375f5f06
KW
1906 /* The Windows functions don't bother to follow the POSIX standard, which
1907 * for example says that something can't both be a printable and a control.
1908 * But Windows treats the \t control as a printable, and does such things
1909 * as making superscripts into both digits and punctuation. This tames
1910 * these flaws by assuming that the definitions of both controls and space
1911 * are correct, and then making sure that other definitions don't have
1912 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1913 * Not all possible weirdnesses are checked for, just the ones that were
1914 * detected on actual Microsoft code pages */
1915
b7d90381
KW
1916# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1917# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1918
1919# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1920 && isALPHANUMERIC_LC(c))
1921# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1922 ! isPUNCT_LC(c))
1923# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1924 isALPHANUMERIC_LC(c))
1925# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1926# define isIDFIRST_LC(c) (((c) == '_') \
1927 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1928# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1929# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1930# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1931# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1932# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1933# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1934 && isALPHANUMERIC_LC(c))
f05550c0
BF
1935
1936# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1937# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1938# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1939
1940#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1941 /* For most other platforms */
beab9ebe 1942
f05550c0
BF
1943# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1944# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1945# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1946# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1947# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1948# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1949# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1950# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1951# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1952# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1953# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1954# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1955# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1956
1957
1958# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1959# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1960# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1961
1962#else /* The final fallback position */
1963
b7d90381
KW
1964# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1965# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1966# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1967# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1968# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1969# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1970# define isLOWER_LC(c) (isascii(c) && islower(c))
1971# define isPRINT_LC(c) (isascii(c) && isprint(c))
1972# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1973# define isSPACE_LC(c) (isascii(c) && isspace(c))
1974# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1975# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1976# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1977
1978# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1979# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1980# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1981
f05550c0 1982#endif
55204971 1983
eba68aa0
KW
1984#define isIDCONT(c) isWORDCHAR(c)
1985#define isIDCONT_A(c) isWORDCHAR_A(c)
1986#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1987#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1988#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1989
4650c663 1990/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1991 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1992 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1993 * classes where there is a macro or function 'above_latin1' that takes the
1994 * single argument 'c' and returns the desired value. These exist for those
2366ba44
KW
1995 * classes which have simple definitions, avoiding the overhead of an inversion
1996 * list binary search. _generic_invlist_uvchr() can be used
4650c663 1997 * for classes where that overhead is faster than a direct lookup.
d0da05db 1998 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1999 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
2000 * have duplicate checks here, so could create versions of the macros that
2001 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
2002
2003/* Note that all ignore 'use bytes' */
1e222e4f
KW
2004#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
2005 ? _generic_isCC(c, classnum) \
cd500f2f 2006 : above_latin1(c))
2366ba44 2007#define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
1e222e4f 2008 ? _generic_isCC(c, classnum) \
922e8cb4 2009 : _is_uni_FOO(classnum, c))
2366ba44
KW
2010#define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
2011#define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
d0da05db
KW
2012#define isASCII_uvchr(c) isASCII(c)
2013#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
2014#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2366ba44
KW
2015#define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
2016#define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
1e222e4f
KW
2017#define isIDCONT_uvchr(c) \
2018 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
2019#define isIDFIRST_uvchr(c) \
2020 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2366ba44
KW
2021#define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
2022#define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
d0da05db 2023
2366ba44 2024#define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
d0da05db
KW
2025#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
2026#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
2027
2366ba44 2028#define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
d0da05db 2029#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2366ba44 2030#define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
d0da05db
KW
2031#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
2032
2033#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
2034#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
2035#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
2036#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
2037
2038/* For backwards compatibility, even though '_uni' should mean official Unicode
2039 * code points, in Perl it means native for those below 256 */
2040#define isALPHA_uni(c) isALPHA_uvchr(c)
2041#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
2042#define isASCII_uni(c) isASCII_uvchr(c)
2043#define isBLANK_uni(c) isBLANK_uvchr(c)
2044#define isCNTRL_uni(c) isCNTRL_uvchr(c)
2045#define isDIGIT_uni(c) isDIGIT_uvchr(c)
2046#define isGRAPH_uni(c) isGRAPH_uvchr(c)
2047#define isIDCONT_uni(c) isIDCONT_uvchr(c)
2048#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
2049#define isLOWER_uni(c) isLOWER_uvchr(c)
2050#define isPRINT_uni(c) isPRINT_uvchr(c)
2051#define isPUNCT_uni(c) isPUNCT_uvchr(c)
2052#define isSPACE_uni(c) isSPACE_uvchr(c)
2053#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
2054#define isUPPER_uni(c) isUPPER_uvchr(c)
2055#define isVERTWS_uni(c) isVERTWS_uvchr(c)
2056#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2057#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2058#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2059#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2060#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2061#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 2062
4650c663
KW
2063/* For internal core Perl use only: the base macros for defining macros like
2064 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 2065 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 2066 * simple class definitions; the other for more complex. These are like
d0da05db 2067 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
2068#define _generic_LC_uvchr(latin1, above_latin1, c) \
2069 (c < 256 ? latin1(c) : above_latin1(c))
2366ba44 2070#define _generic_LC_invlist_uvchr(latin1, classnum, c) \
cd500f2f
KW
2071 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2072
2366ba44
KW
2073#define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2074#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
cd500f2f 2075 _CC_ALPHANUMERIC, c)
b7d90381
KW
2076#define isASCII_LC_uvchr(c) isASCII_LC(c)
2077#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2078 is_HORIZWS_cp_high, c)
feeab5a9 2079#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2366ba44
KW
2080#define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2081#define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 2082#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 2083 _is_uni_perl_idcont, c)
b7d90381 2084#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f 2085 _is_uni_perl_idstart, c)
2366ba44
KW
2086#define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2087#define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 2088#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2366ba44 2089#define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 2090#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 2091 is_XPERLSPACE_cp_high, c)
2366ba44
KW
2092#define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2093#define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
cd500f2f 2094 _CC_WORDCHAR, c)
b7d90381
KW
2095#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2096 is_XDIGIT_cp_high, c)
e712593e 2097
b7d90381 2098#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 2099
da8c1a98
KW
2100/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2101 * they don't otherwise go out of their way to look for malformed UTF-8. If
2102 * they can return accurate results without knowing if the input is otherwise
2103 * malformed, they do so. For example isASCII is accurate in spite of any
2104 * non-length malformations because it looks only at a single byte. Likewise
2105 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2106 * variant ones return FALSE. But, if the input has to be well-formed in order
2107 * for the results to be accurate, the macros will test and if malformed will
2108 * call a routine to die
2109 *
2110 * Except for toke.c, the macros do assume that e > p, asserting that on
2111 * DEBUGGING builds. Much code that calls these depends on this being true,
2112 * for other reasons. toke.c is treated specially as using the regular
2113 * assertion breaks it in many ways. All strings that these operate on there
2114 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2115 * bunch of code in toke.c assumes that this is true, so the assertion allows
2116 * for that */
2117#ifdef PERL_IN_TOKE_C
2118# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2119#else
2120# define _utf8_safe_assert(p,e) ((e) > (p))
2121#endif
2122
2123#define _generic_utf8_safe(classnum, p, e, above_latin1) \
c81b3562
KW
2124 ((! _utf8_safe_assert(p, e)) \
2125 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2126 : (UTF8_IS_INVARIANT(*(p))) \
da8c1a98
KW
2127 ? _generic_isCC(*(p), classnum) \
2128 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2129 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2130 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2131 classnum) \
2132 : (_force_out_malformed_utf8_message( \
2133 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2134 : above_latin1))
b7d90381
KW
2135/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2136 * 'above_latin1' can be a macro */
da8c1a98
KW
2137#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2138 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2366ba44 2139#define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2140 _generic_utf8_safe(classnum, p, e, \
2141 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2142 ? (_force_out_malformed_utf8_message( \
2143 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2144 : above_latin1(p)))
2366ba44
KW
2145/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2146 * 'above_latin1' parameter */
2147#define _generic_invlist_utf8_safe(classnum, p, e) \
2148 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
922e8cb4 2149
cc8ab7c0 2150/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
2151 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2152 * class is TRUE for. Hence it can skip the tests for this range.
2153 * 'above_latin1' should include its arguments */
da8c1a98
KW
2154#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2155 (__ASSERT_(_utf8_safe_assert(p, e)) \
2156 (UTF8_IS_INVARIANT(*(p))) \
2157 ? _generic_isCC(*(p), classnum) \
2158 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2159 ? 0 /* Note that doesn't check validity for latin1 */ \
2160 : above_latin1)
2161
84238efa 2162
059703b0
KW
2163#define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2164#define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2165#define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2166#define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2167#define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2168#define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2169#define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2170#define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2171#define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2172#define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2173#define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2174#define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2175#define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2176#define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2177#define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2178#define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2179#define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2180#define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
e8fa43e2 2181
2366ba44 2182#define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
da8c1a98 2183#define isALPHANUMERIC_utf8_safe(p, e) \
2366ba44 2184 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
da8c1a98
KW
2185#define isASCII_utf8_safe(p, e) \
2186 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2187 * works */ \
2188 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2189#define isBLANK_utf8_safe(p, e) \
2366ba44 2190 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
da8c1a98 2191
e8fa43e2
KW
2192#ifdef EBCDIC
2193 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2194 * more efficient macro instead of the more general one */
da8c1a98 2195# define isCNTRL_utf8_safe(p, e) \
56d02b8c 2196 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 2197#else
da8c1a98 2198# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
2199#endif
2200
da8c1a98
KW
2201#define isDIGIT_utf8_safe(p, e) \
2202 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2366ba44
KW
2203 _is_utf8_FOO(_CC_DIGIT, p, e))
2204#define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
da8c1a98 2205#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
dd1a3ba7 2206 _is_utf8_perl_idcont, p, e)
e5dcd934 2207
c11ff943
KW
2208/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2209 * IDFIRST is an alnum. See
8034715d 2210 * https://github.com/Perl/perl5/issues/10275 for more detail than you
f91dcd13
KW
2211 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2212 * This used to be not the XID version, but we decided to go with the more
2213 * modern Unicode definition */
da8c1a98
KW
2214#define isIDFIRST_utf8_safe(p, e) \
2215 _generic_func_utf8_safe(_CC_IDFIRST, \
dd1a3ba7 2216 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
da8c1a98 2217
2366ba44
KW
2218#define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2219#define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
da8c1a98 2220#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2366ba44 2221#define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
da8c1a98 2222#define isSPACE_utf8_safe(p, e) \
2366ba44
KW
2223 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2224#define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
da8c1a98 2225#define isVERTWS_utf8_safe(p, e) \
2366ba44 2226 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
da8c1a98 2227#define isWORDCHAR_utf8_safe(p, e) \
2366ba44 2228 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
da8c1a98
KW
2229#define isXDIGIT_utf8_safe(p, e) \
2230 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2231 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2232 ? (_force_out_malformed_utf8_message( \
2233 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2234 : is_XDIGIT_high(p)))
a0ed51b3 2235
059703b0
KW
2236#define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2237#define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2238#define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2239#define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2e8adce6 2240
567b353c 2241/* For internal core use only, subject to change */
059703b0
KW
2242#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2243#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2244#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2245#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
a1a5ec35
KW
2246
2247#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2248#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2249#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2250#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 2251
059703b0
KW
2252#define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2253#define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2254#define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2255#define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2256#define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2257#define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2258#define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2259#define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2260#define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2261#define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2262#define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2263#define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2264#define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2265#define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2266#define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2267#define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2268#define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
34aeb2e9 2269
da8c1a98
KW
2270/* For internal core Perl use only: the base macros for defining macros like
2271 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2272 * point in 'p' is within the 0-255 range, it uses locale rules from the
2273 * passed-in 'macro' parameter */
2274#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2275 (__ASSERT_(_utf8_safe_assert(p, e)) \
2276 (UTF8_IS_INVARIANT(*(p))) \
2277 ? macro(*(p)) \
2278 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2279 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2280 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2281 : (_force_out_malformed_utf8_message( \
2282 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2283 : above_latin1))
2284
2366ba44
KW
2285#define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2286 _generic_LC_utf8_safe(macro, p, e, \
2287 _is_utf8_FOO(classnum, p, e))
da8c1a98
KW
2288
2289#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2290 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2291
2366ba44 2292#define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2293 _generic_LC_utf8_safe(classnum, p, e, \
2294 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2295 ? (_force_out_malformed_utf8_message( \
2296 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2297 : above_latin1(p)))
2298
2299#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2366ba44 2300 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
da8c1a98
KW
2301 _CC_ALPHANUMERIC, p, e)
2302#define isALPHA_LC_utf8_safe(p, e) \
2366ba44 2303 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
da8c1a98
KW
2304#define isASCII_LC_utf8_safe(p, e) \
2305 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2306#define isBLANK_LC_utf8_safe(p, e) \
2366ba44 2307 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
da8c1a98
KW
2308#define isCNTRL_LC_utf8_safe(p, e) \
2309 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2310#define isDIGIT_LC_utf8_safe(p, e) \
2366ba44 2311 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
da8c1a98 2312#define isGRAPH_LC_utf8_safe(p, e) \
2366ba44 2313 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
da8c1a98
KW
2314#define isIDCONT_LC_utf8_safe(p, e) \
2315 _generic_LC_func_utf8_safe(isIDCONT_LC, \
dd1a3ba7 2316 _is_utf8_perl_idcont, p, e)
da8c1a98
KW
2317#define isIDFIRST_LC_utf8_safe(p, e) \
2318 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
dd1a3ba7 2319 _is_utf8_perl_idstart, p, e)
da8c1a98 2320#define isLOWER_LC_utf8_safe(p, e) \
2366ba44 2321 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
da8c1a98 2322#define isPRINT_LC_utf8_safe(p, e) \
2366ba44 2323 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
da8c1a98
KW
2324#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2325#define isPUNCT_LC_utf8_safe(p, e) \
2366ba44 2326 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
da8c1a98 2327#define isSPACE_LC_utf8_safe(p, e) \
2366ba44 2328 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
da8c1a98 2329#define isUPPER_LC_utf8_safe(p, e) \
2366ba44 2330 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
da8c1a98 2331#define isWORDCHAR_LC_utf8_safe(p, e) \
2366ba44 2332 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
da8c1a98 2333#define isXDIGIT_LC_utf8_safe(p, e) \
2366ba44 2334 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2335
fbc19f27
KW
2336/* Macros for backwards compatibility and for completeness when the ASCII and
2337 * Latin1 values are identical */
b7d90381
KW
2338#define isALPHAU(c) isALPHA_L1(c)
2339#define isDIGIT_L1(c) isDIGIT_A(c)
2340#define isOCTAL(c) isOCTAL_A(c)
2341#define isOCTAL_L1(c) isOCTAL_A(c)
2342#define isXDIGIT_L1(c) isXDIGIT_A(c)
2343#define isALNUM(c) isWORDCHAR(c)
a377c856 2344#define isALNUM_A(c) isALNUM(c)
b7d90381
KW
2345#define isALNUMU(c) isWORDCHAR_L1(c)
2346#define isALNUM_LC(c) isWORDCHAR_LC(c)
2347#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2348#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
059703b0 2349#define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
4c1d9526 2350#define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
059703b0 2351#define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
4c1d9526 2352#define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
b7d90381
KW
2353#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2354#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2355#define isALNUMC(c) isALPHANUMERIC(c)
2356#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2357#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2358#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
059703b0 2359#define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
4c1d9526
KW
2360#define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2361#define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
fbc19f27 2362
2bd1cbf6
KW
2363/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2364 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2365 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2366 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2367 * the outlier from the block that contains the other controls, just like
2368 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2369 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2370 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2371 * etc. */
2bd1cbf6 2372#ifndef EBCDIC
75763b3a 2373# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2374#else
75763b3a
KW
2375# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2376 ((isPRINT_A(c)) \
2377 ? (UNLIKELY((c) == '?') \
2378 ? QUESTION_MARK_CTRL \
2379 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2380 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2381 ? '?' \
2382 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2383#endif
bbce6d69 2384
dea28490
JJ
2385/* Line numbers are unsigned, 32 bits. */
2386typedef U32 line_t;
e5dcd934 2387#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2388
91152fc1
DG
2389/* Helpful alias for version prescan */
2390#define is_LAX_VERSION(a,b) \
2391 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2392
2393#define is_STRICT_VERSION(a,b) \
2394 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2395
2396#define BADVERSION(a,b,c) \
2397 if (b) { \
2398 *b = c; \
2399 } \
2400 return a;
8c52afec 2401
1ce77b7d
KW
2402/* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2403 * a-f) to its numeric value without using any branches. The input is
2404 * validated only by an assert() in DEBUGGING builds.
2405 *
2406 * It works by right shifting and isolating the bit that is 0 for the digits,
2407 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2408 * position, and then to the eights position. Both are added together to form
2409 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2410 * final four bits of the input to form the correct value. */
2411#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2412 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2413 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2414 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2415
2416/* The argument is a string pointer, which is advanced. */
2417#define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
95a59cab 2418
cb27eebd
KW
2419/* Converts a character known to represent an octal digit (0-7) to its numeric
2420 * value. The input is validated only by an assert() in DEBUGGING builds. In
2421 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2422#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2423
305b8651 2424/* Efficiently returns a boolean as to if two native characters are equivalent
f1460a66 2425 * case-insensitively. At least one of the characters must be one of [A-Za-z];
305b8651
KW
2426 * the ALPHA in the name is to remind you of that. This is asserted() in
2427 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2428 * works (on valid input) for both non- and UTF-8-encoded bytes.
2429 *
2430 * When one of the inputs is a compile-time constant and gets folded by the
2431 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2432 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2433 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2434 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2435 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2436 * just a single 0, in the bit position where the upper- and lowercase differ.
2437 * */
305b8651
KW
2438#define isALPHA_FOLD_EQ(c1, c2) \
2439 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2440 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2441#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2442
8e84507e 2443/*
70b05c7c 2444=for apidoc_section Memory Management
ccfc67b7 2445
a02a5408 2446=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2447The XSUB-writer's interface to the C C<malloc> function.
2448
596f7718 2449Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2450
c5008215
JC
2451In 5.9.3, Newx() and friends replace the older New() API, and drops
2452the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2453themselves. This aid has been superseded by a new build option,
d10b4965 2454PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2455there for use in XS modules supporting older perls.
2456
a02a5408 2457=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2458The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2459cast. See also C<L</Newx>>.
954c1994 2460
596f7718 2461Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2462
a02a5408 2463=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2464The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2465memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2466
596f7718 2467Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2468
954c1994
GS
2469=for apidoc Am|void|Renew|void* ptr|int nitems|type
2470The XSUB-writer's interface to the C C<realloc> function.
2471
596f7718 2472Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2473
954c1994
GS
2474=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2475The XSUB-writer's interface to the C C<realloc> function, with
2476cast.
2477
596f7718 2478Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2479
49b8b560 2480=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2481The XSUB-writer's interface to the C C<free> function.
2482
596f7718 2483This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2484
51b56f5c 2485=for apidoc_section String Handling
954c1994
GS
2486=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2487The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2488source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2489C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2490
e90e2364 2491=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2492Like C<Move> but returns C<dest>. Useful
72d33970 2493for encouraging compilers to tail-call
e90e2364
NC
2494optimise.
2495
954c1994
GS
2496=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2497The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2498source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2499C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2500
e90e2364
NC
2501=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2502
796b6530 2503Like C<Copy> but returns C<dest>. Useful
72d33970 2504for encouraging compilers to tail-call
e90e2364
NC
2505optimise.
2506
954c1994
GS
2507=for apidoc Am|void|Zero|void* dest|int nitems|type
2508
2509The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2510destination, C<nitems> is the number of items, and C<type> is the type.
2511
e90e2364
NC
2512=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2513
72d33970
FC
2514Like C<Zero> but returns dest. Useful
2515for encouraging compilers to tail-call
e90e2364
NC
2516optimise.
2517
51b56f5c 2518=for apidoc_section Utility Functions
e538104b 2519=for apidoc Amu|void|StructCopy|type *src|type *dest|type
4375e838 2520This is an architecture-independent macro to copy one structure to another.
954c1994 2521
7e337ee0
JH
2522=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2523
2524Fill up memory with a byte pattern (a byte repeated over and over
2525again) that hopefully catches attempts to access uninitialized memory.
2526
2527=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2528
2529PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2530
1c12ffb4 2531=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2532
2533PoisonWith(0xEF) for catching access to freed memory.
2534
9965345d
JH
2535=for apidoc Am|void|Poison|void* dest|int nitems|type
2536
7e337ee0 2537PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2538
2539=cut */
954c1994 2540
561b68a9
SH
2541/* Maintained for backwards-compatibility only. Use newSV() instead. */
2542#ifndef PERL_CORE
ff06c60c 2543#define NEWSV(x,len) newSV(len)
561b68a9 2544#endif
ff06c60c 2545
b7112dce 2546#define MEM_SIZE_MAX ((MEM_SIZE)-1)
19a94d75 2547
a500027b 2548#define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
e6bdf523 2549
27d5b266 2550#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2551
2552/* This expression will be constant-folded at compile time. It checks
2553 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2554 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2555 * As well as avoiding the need for a run-time check in some cases, it's
2556 * designed to avoid compiler warnings like:
2557 * comparison is always false due to limited range of data type
73e8ff00
DM
2558 * It's mathematically equivalent to
2559 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2560 */
2561
2562# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
445198b9
LM
2563 ( sizeof(MEM_SIZE) < sizeof(n) \
2564 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 2565
88f9f128 2566/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2567 * compiler warnings. We *want* to write the expression as
2568 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2569 * (for some compile-time constant C), but even when the LHS
2570 * constant-folds to false at compile-time, g++ insists on emitting
2571 * warnings about the RHS (e.g. "comparison is always false"), so instead
2572 * we write it as
e6bdf523 2573 *
d98e5cde 2574 * (cond ? n : X) > C
88f9f128 2575 *
d98e5cde
DM
2576 * where X is a constant with X > C always false. Choosing a value for X
2577 * is tricky. If 0, some compilers will complain about 0 > C always being
2578 * false; if 1, Coverity complains when n happens to be the constant value
2579 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2580 * for X and hope that nothing else whines.
e6bdf523
DM
2581 */
2582
2583# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2584 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2585 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2586
2587# define MEM_WRAP_CHECK(n,t) \
2588 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2589 && (croak_memory_wrap(),0))
2590
2591# define MEM_WRAP_CHECK_1(n,t,a) \
2592 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2593 && (Perl_croak_nocontext("%s",(a)),0))
2594
814eedc8
DD
2595/* "a" arg must be a string literal */
2596# define MEM_WRAP_CHECK_s(n,t,a) \
2597 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2598 && (Perl_croak_nocontext("" a ""),0))
2599
8b44ba4c 2600#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2601
a500027b 2602#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
2603#else
2604
410319be
NC
2605#define MEM_WRAP_CHECK(n,t)
2606#define MEM_WRAP_CHECK_1(n,t,a)
814eedc8 2607#define MEM_WRAP_CHECK_s(n,t,a)
8b44ba4c
NC
2608#define MEM_WRAP_CHECK_(n,t)
2609
a500027b 2610#define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
27d5b266 2611
1936d2a7 2612#endif
8b44ba4c 2613
fe4f188c 2614#ifdef PERL_MEM_LOG
46c6c7e2 2615/*
9f653bb5 2616 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2617 * go through functions, which are handy for debugging breakpoints, but
2618 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2619 * line number, and C function name if available) passed in. This info can
2620 * then be used for logging the calls, for which one gets a sample
73d1d973 2621 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2622 *
46c6c7e2 2623 * Known problems:
94e892a6 2624 * - not all memory allocs get logged, only those
46c6c7e2 2625 * that go through Newx() and derivatives (while all
94e892a6 2626 * Safefrees do get logged)
46c6c7e2
JH
2627 * - __FILE__ and __LINE__ do not work everywhere
2628 * - __func__ or __FUNCTION__ even less so
2629 * - I think more goes on after the perlio frees but
2630 * the thing is that STDERR gets closed (as do all
2631 * the file descriptors)
2632 * - no deeper calling stack than the caller of the Newx()
2633 * or the kind, but do I look like a C reflection/introspection
2634 * utility to you?
2635 * - the function prototypes for the logging functions
2636 * probably should maybe be somewhere else than handy.h
2637 * - one could consider inlining (macrofying) the logging
2638 * for speed, but I am too lazy
2639 * - one could imagine recording the allocations in a hash,
2640 * (keyed by the allocation address?), and maintain that
2641 * through reallocs and frees, but how to do that without
2642 * any News() happening...?
73d1d973 2643 * - lots of -Ddefines to get useful/controllable output
b953482e 2644 * - lots of ENV reads
46c6c7e2
JH
2645 */
2646
0b0ab801 2647# ifdef PERL_CORE
73d1d973 2648# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2649enum mem_log_type {
2650 MLT_ALLOC,
2651 MLT_REALLOC,
d7a2c63c
MHM
2652 MLT_FREE,
2653 MLT_NEW_SV,
2654 MLT_DEL_SV
0b0ab801
MHM
2655};
2656# endif
12754f92 2657# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2658void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2659void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2660# endif
0b0ab801
MHM
2661# endif
2662
fe4f188c
JH
2663#endif
2664
2665#ifdef PERL_MEM_LOG
d1401ee9
MHM
2666#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2667#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2668#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2669#endif
2670
2671#ifndef MEM_LOG_ALLOC
2672#define MEM_LOG_ALLOC(n,t,a) (a)
2673#endif
2674#ifndef MEM_LOG_REALLOC
2675#define MEM_LOG_REALLOC(n,t,v,a) (a)
2676#endif
2677#ifndef MEM_LOG_FREE
2678#define MEM_LOG_FREE(a) (a)
2679#endif
2680
d1401ee9
MHM
2681#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2682#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2683#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2684
2685#ifndef PERL_CORE
a02a5408
JC
2686/* pre 5.9.x compatibility */
2687#define New(x,v,n,t) Newx(v,n,t)
2688#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2689#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2690#endif
a02a5408 2691
ff68c719 2692#define Renew(v,n,t) \
d1401ee9 2693 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2694#define Renewc(v,n,t,c) \
d1401ee9 2695 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2696
2697#ifdef PERL_POISON
2698#define Safefree(d) \
06c0cc96 2699 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2700#else
fe4f188c 2701#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2702#endif
55497cff 2703
dbb57106
YO
2704/* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2705 * as it handles cases like constant string arguments without throwing warnings *
2706 * the cast is required, as is the inequality check, to avoid warnings */
45908e4d 2707#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2708
45908e4d
YO
2709
2710#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)))
2711#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)))
2712#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2713
bdd1531d 2714/* Like above, but returns a pointer to 'd' */
45908e4d
YO
2715#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)))
2716#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 2717#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2718
7e337ee0
JH
2719#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2720#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2721#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2722#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2723
caa674f3
DD
2724#ifdef PERL_POISON
2725# define PERL_POISON_EXPR(x) x
2726#else
2727# define PERL_POISON_EXPR(x)
2728#endif
2729
ff68c719 2730#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2731
1b7e2294 2732/*
51b56f5c 2733=for apidoc_section Utility Functions
1b7e2294
KW
2734
2735=for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2736
2737Returns the number of elements in the input C array (so you want your
2738zero-based indices to be less than but not equal to).
2739
2740=for apidoc Am|void *|C_ARRAY_END|void *a
2741
2742Returns a pointer to one element past the final element of the input C array.
2743
2744=cut
2745
2746C_ARRAY_END is one past the last: half-open/half-closed range, not
2747last-inclusive range.
2748*/
622913ab 2749#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2750#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2751
2cc61e15
DD
2752#ifdef NEED_VA_COPY
2753# ifdef va_copy
2754# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2755# elif defined(__va_copy)
2756# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2757# else
07798b17 2758# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2759# endif
2760#endif
2761
472d47bc
SB
2762/* convenience debug macros */
2763#ifdef USE_ITHREADS
2764#define pTHX_FORMAT "Perl interpreter: 0x%p"
2765#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2766#define pTHX_VALUE_ (void *)my_perl,
2767#define pTHX_VALUE (void *)my_perl
2768#define pTHX__VALUE_ ,(void *)my_perl,
2769#define pTHX__VALUE ,(void *)my_perl
472d47bc 2770#else
3609ea0d 2771#define pTHX_FORMAT
472d47bc 2772#define pTHX__FORMAT
3609ea0d 2773#define pTHX_VALUE_
472d47bc 2774#define pTHX_VALUE
3609ea0d 2775#define pTHX__VALUE_
472d47bc
SB
2776#define pTHX__VALUE
2777#endif /* USE_ITHREADS */
3609ea0d 2778
2acdbac1
NC
2779/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2780 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2781 CPAN::Unpack show any users outside the core. */
2782#ifdef PERL_CORE
dc6e8de0
A
2783# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2784 "Use of " s " is deprecated")
c9680906
A
2785# define deprecate_disappears_in(when,message) \
2786 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2787 message ", and will disappear in Perl " when)
ac641426
A
2788# define deprecate_fatal_in(when,message) \
2789 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2790 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2791#endif
2792
dfff4baf
BF
2793/* Internal macros to deal with gids and uids */
2794#ifdef PERL_CORE
2795
2796# if Uid_t_size > IVSIZE
2797# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2798# define SvUID(sv) SvNV(sv)
07798b17
AC
2799# elif Uid_t_sign <= 0
2800# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2801# define SvUID(sv) SvIV(sv)
dfff4baf 2802# else
07798b17
AC
2803# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2804# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2805# endif /* Uid_t_size */
2806
2807# if Gid_t_size > IVSIZE
2808# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2809# define SvGID(sv) SvNV(sv)
07798b17
AC
2810# elif Gid_t_sign <= 0
2811# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2812# define SvGID(sv) SvIV(sv)
dfff4baf 2813# else
07798b17
AC
2814# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2815# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2816# endif /* Gid_t_size */
2817
2818#endif
2819
6a5bc5ac 2820#endif /* PERL_HANDY_H_ */
9d745869 2821
e9a8c099 2822/*
14d04a33 2823 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2824 */