This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Fix bytes_from_utf8_loc() pod
[perl5.git] / handy.h
CommitLineData
a0d0e21e 1/* handy.h
a687059c 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
da5d8dbb 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
a687059c 5 *
6e21c824
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8d063cd8 8 *
8d063cd8
LW
9 */
10
4650c663
KW
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
6a5bc5ac
KW
14#ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */
15#define PERL_HANDY_H_
9d745869 16
24792b8d
NC
17#ifndef PERL_CORE
18# define Null(type) ((type)NULL)
954c1994
GS
19
20/*
51b56f5c 21=for apidoc_section String Handling
78342678 22=for apidoc AmnU||Nullch
72d33970
FC
23Null character pointer. (No longer available when C<PERL_CORE> is
24defined.)
2307c6d0 25
51b56f5c 26=for apidoc_section SV Handling
78342678 27=for apidoc AmnU||Nullsv
72d33970 28Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
29
30=cut
51b56f5c
KW
31
32Below are signatures of functions from config.h which can't easily be gleaned
33from it, and are very unlikely to change
34
35=for apidoc_section Signals
36=for apidoc Am|int|Sigsetjmp|jmp_buf env|int savesigs
37=for apidoc Am|void|Siglongjmp|jmp_buf env|int val
38
39=for apidoc_section Filesystem configuration values
40=for apidoc Am|void *|FILE_ptr|FILE * f
41=for apidoc Am|Size_t|FILE_cnt|FILE * f
42=for apidoc Am|void *|FILE_base|FILE * f
43=for apidoc Am|Size_t|FILE_bufsiz
44
45=for apidoc_section String Handling
46=for apidoc Am|token|CAT2|token x|token y
47=for apidoc Am|string|STRINGIFY|token x
48
49=for apidoc_section Numeric Functions
50=for apidoc Am|double|Drand01|double x
51=for apidoc Am|void|seedDrand01|Rand_seed_t x
52=for apidoc Am|char *|Gconvert|double x|Size_t n|bool t|char * b
53
54=cut
954c1994
GS
55*/
56
24792b8d
NC
57# define Nullch Null(char*)
58# define Nullfp Null(PerlIO*)
59# define Nullsv Null(SV*)
60#endif
8d063cd8 61
641d3f0b
PP
62#ifdef TRUE
63#undef TRUE
64#endif
65#ifdef FALSE
66#undef FALSE
67#endif
68#define TRUE (1)
69#define FALSE (0)
70
cf3f0ffb
DM
71/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
72 * (compiler permitting) that casting away const-ness will give a warning;
73 * e.g.:
74 *
75 * const SV *sv = ...;
76 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
77 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
78 */
79
b1bc3f34 80#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
6c2255e0 81# define MUTABLE_PTR(p) ({ void *p_ = (p); p_; })
b1bc3f34
NC
82#else
83# define MUTABLE_PTR(p) ((void *) (p))
84#endif
85
a062e10d 86#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 87#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 88#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 89#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 90#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 91#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 92
f789f6a4 93#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
94# include <stdbool.h>
95# ifndef HAS_BOOL
96# define HAS_BOOL 1
97# endif
98#endif
99
8e84507e 100/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
101 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
102 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 103 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
104 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
105 g++ can be identified by __GNUG__.
106 Andy Dougherty February 2000
5d94fbed 107*/
3609ea0d 108#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 109# ifndef PERL_BOOL_AS_CHAR
5d94fbed 110# ifndef HAS_BOOL
c1d22f6b 111# define HAS_BOOL 1
5d94fbed 112# endif
f789f6a4 113# endif
5d94fbed
AD
114#endif
115
116#ifndef HAS_BOOL
f789f6a4
FC
117# ifdef bool
118# undef bool
119# endif
70d5cb32 120# define bool char
c1d22f6b 121# define HAS_BOOL 1
a687059c 122#endif
0d3e774c 123
25ba28ce 124/*
51b56f5c 125=for apidoc_section Casting
25ba28ce
KW
126=for apidoc Am|bool|cBOOL|bool expr
127
128Cast-to-bool. A simple S<C<(bool) I<expr>>> cast may not do the right thing:
129if C<bool> is defined as C<char>, for example, then the cast from C<int> is
130implementation-defined.
131
132C<(bool)!!(cbool)> in a ternary triggers a bug in xlc on AIX
133
134=cut
135*/
18f5643b 136#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 137
46c6c7e2 138/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
139 * XXX Should really be a Configure probe, with HAS__FUNCTION__
140 * and FUNCTION__ as results.
141 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
46c6c7e2
JH
142#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
143# define FUNCTION__ __func__
7adf2470 144#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 145# define FUNCTION__ ""
46c6c7e2 146#else
07798b17 147# define FUNCTION__ __FUNCTION__ /* Common extension. */
46c6c7e2
JH
148#endif
149
27d4fb96
PP
150/* XXX A note on the perl source internal type system. The
151 original intent was that I32 be *exactly* 32 bits.
152
153 Currently, we only guarantee that I32 is *at least* 32 bits.
154 Specifically, if int is 64 bits, then so is I32. (This is the case
155 for the Cray.) This has the advantage of meshing nicely with
156 standard library calls (where we pass an I32 and the library is
157 expecting an int), but the disadvantage that an I32 is not 32 bits.
158 Andy Dougherty August 1996
24fef2a7 159
dc45a647
MB
160 There is no guarantee that there is *any* integral type with
161 exactly 32 bits. It is perfectly legal for a system to have
162 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 163
dc45a647
MB
164 Similarly, there is no guarantee that I16 and U16 have exactly 16
165 bits.
693762b4 166
8e84507e
NIS
167 For dealing with issues that may arise from various 32/64-bit
168 systems, we will ask Configure to check out
8175356b 169
3609ea0d
JH
170 SHORTSIZE == sizeof(short)
171 INTSIZE == sizeof(int)
172 LONGSIZE == sizeof(long)
dc45a647 173 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 174 PTRSIZE == sizeof(void *)
dc45a647
MB
175 DOUBLESIZE == sizeof(double)
176 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 177
27d4fb96
PP
178*/
179
69512466
JH
180#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
181# include <inttypes.h>
dd0eed91
JH
182# ifdef INT32_MIN_BROKEN
183# undef INT32_MIN
184# define INT32_MIN (-2147483647-1)
185# endif
186# ifdef INT64_MIN_BROKEN
187# undef INT64_MIN
188# define INT64_MIN (-9223372036854775807LL-1)
189# endif
69512466
JH
190#endif
191
8175356b
JH
192typedef I8TYPE I8;
193typedef U8TYPE U8;
194typedef I16TYPE I16;
195typedef U16TYPE U16;
196typedef I32TYPE I32;
197typedef U32TYPE U32;
16d89be8 198
74b807c7 199#ifdef QUADKIND
8175356b
JH
200typedef I64TYPE I64;
201typedef U64TYPE U64;
16d89be8 202#endif
8175356b 203
d8668976 204#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 205
5ff3f7a4
GS
206/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
207 Please search CHAR_MAX in perl.h for further details. */
208#define U8_MAX UINT8_MAX
209#define U8_MIN UINT8_MIN
210
5ff3f7a4
GS
211#define I16_MAX INT16_MAX
212#define I16_MIN INT16_MIN
213#define U16_MAX UINT16_MAX
214#define U16_MIN UINT16_MIN
215
5ff3f7a4
GS
216#define I32_MAX INT32_MAX
217#define I32_MIN INT32_MIN
0e983133
GS
218#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
219# define U32_MAX UINT32_MAX
220#else
221# define U32_MAX 4294967295U
222#endif
5ff3f7a4
GS
223#define U32_MIN UINT32_MIN
224
225#else
226
5c9fa16e
KA
227/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
228 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
229#define U8_MAX PERL_UCHAR_MAX
230#define U8_MIN PERL_UCHAR_MIN
79072805 231
27d4fb96
PP
232#define I16_MAX PERL_SHORT_MAX
233#define I16_MIN PERL_SHORT_MIN
234#define U16_MAX PERL_USHORT_MAX
235#define U16_MIN PERL_USHORT_MIN
79072805 236
c4f23d77 237#if LONGSIZE > 4
27d4fb96
PP
238# define I32_MAX PERL_INT_MAX
239# define I32_MIN PERL_INT_MIN
240# define U32_MAX PERL_UINT_MAX
241# define U32_MIN PERL_UINT_MIN
79072805 242#else
27d4fb96
PP
243# define I32_MAX PERL_LONG_MAX
244# define I32_MIN PERL_LONG_MIN
245# define U32_MAX PERL_ULONG_MAX
246# define U32_MIN PERL_ULONG_MIN
79072805
LW
247#endif
248
5ff3f7a4
GS
249#endif
250
247cee9f
KW
251/* These C99 typedefs are useful sometimes for, say, loop variables whose
252 * maximum values are small, but for which speed trumps size. If we have a C99
253 * compiler, use that. Otherwise, a plain 'int' should be good enough.
254 *
255 * Restrict these to core for now until we are more certain this is a good
256 * idea. */
257#if defined(PERL_CORE) || defined(PERL_EXT)
258# ifdef I_STDINT
259 typedef int_fast8_t PERL_INT_FAST8_T;
260 typedef uint_fast8_t PERL_UINT_FAST8_T;
261 typedef int_fast16_t PERL_INT_FAST16_T;
262 typedef uint_fast16_t PERL_UINT_FAST16_T;
263# else
264 typedef int PERL_INT_FAST8_T;
265 typedef unsigned int PERL_UINT_FAST8_T;
266 typedef int PERL_INT_FAST16_T;
267 typedef unsigned int PERL_UINT_FAST16_T;
268# endif
269#endif
270
464decb6 271/* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
ab350cbd
KW
272 * anyone is grepping for it. So BIT_DIGITS gives the number of decimal digits
273 * required to represent any possible unsigned number containing N bits.
274 * TYPE_DIGITS gives the number of decimal digits required to represent any
275 * possible unsigned number of type T. */
464decb6 276#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
fc36a67e
PP
277#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
278#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
279
88794300 280/* Unused by core; should be deprecated */
ff68c719 281#define Ctl(ch) ((ch) & 037)
8d063cd8 282
98fce2a4
KW
283#if defined(PERL_CORE) || defined(PERL_EXT)
284# ifndef MIN
285# define MIN(a,b) ((a) < (b) ? (a) : (b))
286# endif
287# ifndef MAX
288# define MAX(a,b) ((a) > (b) ? (a) : (b))
289# endif
290#endif
291
84ff4fa9
KW
292/* Returns a boolean as to whether the input unsigned number is a power of 2
293 * (2**0, 2**1, etc). In other words if it has just a single bit set.
294 * If not, subtracting 1 would leave the uppermost bit set, so the & would
295 * yield non-zero */
296#if defined(PERL_CORE) || defined(PERL_EXT)
011b1419 297# define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
84ff4fa9
KW
298#endif
299
d223e1ea 300/* Returns a mask with the lowest n bits set */
fae1e72b 301#define nBIT_MASK(n) ((UINTMAX_C(1) << (n)) - 1)
d223e1ea 302
1381ccb1
KW
303/* The largest unsigned number that will fit into n bits */
304#define nBIT_UMAX(n) nBIT_MASK(n)
305
8d9433eb 306/*
51b56f5c 307=for apidoc_section Compiler directives
8d9433eb
KW
308=for apidoc Am|void|__ASSERT_|bool expr
309
310This is a helper macro to avoid preprocessor issues, replaced by nothing
311unless under DEBUGGING, where it expands to an assert of its argument,
312followed by a comma (hence the comma operator). If we just used a straight
313assert(), we would get a comma with nothing before it when not DEBUGGING.
314
315=cut
316
317We also use empty definition under Coverity since the __ASSERT__
318checks often check for things that Really Cannot Happen, and Coverity
319detects that and gets all excited. */
3e94db23 320
e7ae132e
KW
321#if defined(DEBUGGING) && !defined(__COVERITY__) \
322 && ! defined(PERL_SMALL_MACRO_BUFFER)
0f092d08
KW
323# define __ASSERT_(statement) assert(statement),
324#else
325# define __ASSERT_(statement)
326#endif
327
3fe05580 328/*
51b56f5c 329=for apidoc_section SV Handling
3fe05580 330
3bb9fd01 331=for apidoc Ama|SV*|newSVpvs|"literal string"
1568d13a 332Like C<newSVpvn>, but takes a literal string instead of a
30a15352 333string/length pair.
3fe05580 334
3bb9fd01 335=for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
1568d13a 336Like C<newSVpvn_flags>, but takes a literal string instead of
30a15352 337a string/length pair.
84bafc02 338
3bb9fd01 339=for apidoc Ama|SV*|newSVpvs_share|"literal string"
1568d13a 340Like C<newSVpvn_share>, but takes a literal string instead of
30a15352 341a string/length pair and omits the hash parameter.
3fe05580 342
3bb9fd01 343=for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
1568d13a 344Like C<sv_catpvn_flags>, but takes a literal string instead
30a15352 345of a string/length pair.
9dcc53ea 346
3bb9fd01 347=for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
1568d13a 348Like C<sv_catpvn_nomg>, but takes a literal string instead of
0c395ea5 349a string/length pair.
9dcc53ea 350
3bb9fd01 351=for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
1568d13a 352Like C<sv_catpvn>, but takes a literal string instead of a
0c395ea5 353string/length pair.
3fe05580 354
3bb9fd01 355=for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
1568d13a 356Like C<sv_catpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
357string/length pair.
358
3bb9fd01 359=for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
1568d13a 360Like C<sv_setpvn>, but takes a literal string instead of a
0c395ea5 361string/length pair.
3fe05580 362
3bb9fd01 363=for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
1568d13a 364Like C<sv_setpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
365string/length pair.
366
3bb9fd01 367=for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
1568d13a 368Like C<sv_setref_pvn>, but takes a literal string instead of
0c395ea5 369a string/length pair.
9dcc53ea 370
51b56f5c 371=for apidoc_section String Handling
3fe05580 372
3bb9fd01 373=for apidoc Ama|char*|savepvs|"literal string"
1568d13a 374Like C<savepvn>, but takes a literal string instead of a
30a15352 375string/length pair.
3fe05580 376
3bb9fd01 377=for apidoc Ama|char*|savesharedpvs|"literal string"
9dcc53ea
Z
378A version of C<savepvs()> which allocates the duplicate string in memory
379which is shared between threads.
380
51b56f5c 381=for apidoc_section GV Handling
3fe05580 382
3bb9fd01 383=for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
1568d13a 384Like C<gv_stashpvn>, but takes a literal string instead of a
0c395ea5 385string/length pair.
3fe05580 386
51b56f5c 387=for apidoc_section HV Handling
3fe05580 388
3bb9fd01 389=for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
1568d13a 390Like C<hv_fetch>, but takes a literal string instead of a
0c395ea5 391string/length pair.
3fe05580 392
3bb9fd01 393=for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
1568d13a 394Like C<hv_store>, but takes a literal string instead of a
0c395ea5 395string/length pair
3fe05580
MHM
396and omits the hash parameter.
397
70b05c7c 398=for apidoc_section Lexer interface
510966aa 399
3bb9fd01 400=for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
510966aa 401
1568d13a 402Like L</lex_stuff_pvn>, but takes a literal string instead of
0c395ea5 403a string/length pair.
510966aa 404
3fe05580
MHM
405=cut
406*/
407
a34e53fc 408/*
51b56f5c 409=for apidoc_section String Handling
2efa8cc7 410
a34e53fc
KW
411=for apidoc Amu|pair|STR_WITH_LEN|"literal string"
412
413Returns two comma separated tokens of the input literal string, and its length.
414This is convenience macro which helps out in some API calls.
415Note that it can't be used as an argument to macros or functions that under
416some configurations might be macros, which means that it requires the full
417Perl_xxx(aTHX_ ...) form for any API calls where it's used.
418
419=cut
420*/
421
a34e53fc 422#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
ba3a79e7
GA
423
424/* STR_WITH_LEN() shortcuts */
425#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
426#define newSVpvs_flags(str,flags) \
427 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 428#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
429#define sv_catpvs_flags(sv, str, flags) \
430 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
431#define sv_catpvs_nomg(sv, str) \
432 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
433#define sv_catpvs(sv, str) \
434 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
435#define sv_catpvs_mg(sv, str) \
436 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 437#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
438#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
439#define sv_setref_pvs(rv, classname, str) \
440 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 441#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
442#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
443#define gv_stashpvs(str, create) \
444 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
445#define gv_fetchpvs(namebeg, add, sv_type) \
446 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
447#define gv_fetchpvn(namebeg, len, add, sv_type) \
448 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
449#define sv_catxmlpvs(dsv, str, utf8) \
450 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 451
ba3a79e7 452
510966aa
Z
453#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
454
b96d8cd9
NC
455#define get_cvs(str, flags) \
456 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 457
9b6e9510 458/* internal helpers */
4a1bbd3d 459/* Transitional */
d67f622b
N
460#ifndef PERL_VERSION_MAJOR
461# define PERL_VERSION_MAJOR PERL_REVISION
4a1bbd3d
KW
462#else
463# undef PERL_REVISION /* We don't want code to be using these */
464#endif
d67f622b
N
465#ifndef PERL_VERSION_MINOR
466# define PERL_VERSION_MINOR PERL_VERSION
4a1bbd3d
KW
467#else
468# undef PERL_VERSION
469#endif
d67f622b
N
470#ifndef PERL_VERSION_PATCH
471# define PERL_VERSION_PATCH PERL_SUBVERSION
4a1bbd3d
KW
472#else
473# undef PERL_SUBVERSION
474#endif
475
476#define PERL_JNP_TO_DECIMAL_(maJor,miNor,Patch) \
477 /* '10*' leaves room for things like alpha, beta, releases */ \
478 (10 * ((maJor) * 1000000) + ((miNor) * 1000) + (Patch))
9b6e9510 479#define PERL_DECIMAL_VERSION_ \
d67f622b
N
480 PERL_JNP_TO_DECIMAL_(PERL_VERSION_MAJOR, PERL_VERSION_MINOR, \
481 PERL_VERSION_PATCH)
9b6e9510
KW
482
483/*
51b56f5c 484=for apidoc_section Versioning
4a1bbd3d 485=for apidoc AmR|bool|PERL_VERSION_EQ|const U8 major|const U8 minor|const U8 patch
9b6e9510 486
4a1bbd3d 487Returns whether or not the perl currently being compiled has the specified
9b6e9510
KW
488relationship to the perl given by the parameters. For example,
489
490 #if PERL_VERSION_GT(5,24,2)
491 code that will only be compiled on perls after v5.24.2
492 #else
493 fallback code
494 #endif
495
496Note that this is usable in making compile-time decisions
497
498The possible comparisons are C<PERL_VERSION_EQ>, C<PERL_VERSION_NE>,
499C<PERL_VERSION_GE>, C<PERL_VERSION_GT>, C<PERL_VERSION_LE>, and
500C<PERL_VERSION_LT>.
501
4a1bbd3d
KW
502You may use the special value '*' for the final number to mean ALL possible
503values for it. Thus,
504
505 #if PERL_VERSION_EQ(5,31,'*')
506
507means all perls in the 5.31 series. And
508
509 #if PERL_VERSION_NE(5,24,'*')
510
511means all perls EXCEPT 5.24 ones. And
512
513 #if PERL_VERSION_LE(5,9,'*')
514
515is effectively
516
517 #if PERL_VERSION_LT(5,10,0)
518
519This means you don't have to think so much when converting from the existing
520deprecated C<PERL_VERSION> to using this macro:
521
522 #if PERL_VERSION <= 9
523
524becomes
525
526 #if PERL_VERSION_LE(5,9,'*')
527
528=for apidoc AmRh|bool|PERL_VERSION_NE|const U8 major|const U8 minor|const U8 patch
529=for apidoc AmRh|bool|PERL_VERSION_GE|const U8 major|const U8 minor|const U8 patch
530=for apidoc AmRh|bool|PERL_VERSION_GT|const U8 major|const U8 minor|const U8 patch
531=for apidoc AmRh|bool|PERL_VERSION_LE|const U8 major|const U8 minor|const U8 patch
532=for apidoc AmRh|bool|PERL_VERSION_LT|const U8 major|const U8 minor|const U8 patch
9b6e9510
KW
533
534=cut
535*/
536
4a1bbd3d
KW
537/* N.B. These don't work if the patch version is 42 or 92, as those are what
538 * '*' is in ASCII and EBCDIC respectively */
539# define PERL_VERSION_EQ(j,n,p) \
540 (((p) == '*') \
d67f622b
N
541 ? ( (j) == PERL_VERSION_MAJOR \
542 && (n) == PERL_VERSION_MINOR) \
4a1bbd3d
KW
543 : (PERL_DECIMAL_VERSION_ == PERL_JNP_TO_DECIMAL_(j,n,p)))
544# define PERL_VERSION_NE(j,n,p) (! PERL_VERSION_EQ(j,n,p))
545
546# define PERL_VERSION_LT(j,n,p) /* < '*' effectively means < 0 */ \
547 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
548 (n), \
549 (((p) == '*') ? 0 : p)))
550# define PERL_VERSION_GE(j,n,p) (! PERL_VERSION_LT(j,n,p))
551
552# define PERL_VERSION_LE(j,n,p) /* <= '*' effectively means < n+1 */ \
553 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
554 (((p) == '*') ? ((n)+1) : (n)), \
555 (((p) == '*') ? 0 : p)))
556# define PERL_VERSION_GT(j,n,p) (! PERL_VERSION_LE(j,n,p))
9b6e9510 557
954c1994 558/*
51b56f5c 559=for apidoc_section String Handling
ccfc67b7 560
954c1994 561=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
562Test two C<NUL>-terminated strings to see if they are different. Returns true
563or false.
954c1994
GS
564
565=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
566Test two C<NUL>-terminated strings to see if they are equal. Returns true or
567false.
954c1994
GS
568
569=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
570Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
571second, C<s2>. Returns true or false.
954c1994
GS
572
573=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
574Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
575equal to the second, C<s2>. Returns true or false.
954c1994
GS
576
577=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
578Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
579the second, C<s2>. Returns true or false.
954c1994
GS
580
581=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
582Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
583or equal to the second, C<s2>. Returns true or false.
954c1994
GS
584
585=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
586Test two C<NUL>-terminated strings to see if they are different. The C<len>
587parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
588wrapper for C<strncmp>).
589
590=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
591Test two C<NUL>-terminated strings to see if they are equal. The C<len>
592parameter indicates the number of bytes to compare. Returns true or false. (A
593wrapper for C<strncmp>).
954c1994 594
bd18bd40
KW
595=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
596Test two buffers (which may contain embedded C<NUL> characters, to see if they
597are equal. The C<len> parameter indicates the number of bytes to compare.
598Returns zero if equal, or non-zero if non-equal.
599
3bb9fd01 600=for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
601Like L</memEQ>, but the second string is a literal enclosed in double quotes,
602C<l1> gives the number of bytes in C<s1>.
603Returns zero if equal, or non-zero if non-equal.
604
bd18bd40
KW
605=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
606Test two buffers (which may contain embedded C<NUL> characters, to see if they
607are not equal. The C<len> parameter indicates the number of bytes to compare.
608Returns zero if non-equal, or non-zero if equal.
609
3bb9fd01 610=for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
611Like L</memNE>, but the second string is a literal enclosed in double quotes,
612C<l1> gives the number of bytes in C<s1>.
613Returns zero if non-equal, or zero if non-equal.
614
4aada8b9
KW
615=for apidoc Am|bool|memCHRs|"list"|char c
616Returns the position of the first occurence of the byte C<c> in the literal
617string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
618treated as unsigned char. Thus this macro can be used to determine if C<c> is
619in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
620is C<NUL> (and the set doesn't include C<NUL>).
621
954c1994 622=cut
fc169e00
KW
623
624New macros should use the following conventions for their names (which are
625based on the underlying C library functions):
626
627 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
628
629 Each has two main parameters, string-like operands that are compared
630 against each other, as specified by the macro name. Some macros may
631 additionally have one or potentially even two length parameters. If a length
632 parameter applies to both string parameters, it will be positioned third;
633 otherwise any length parameter immediately follows the string parameter it
634 applies to.
635
636 If the prefix to the name is 'str', the string parameter is a pointer to a C
637 language string. Such a string does not contain embedded NUL bytes; its
638 length may be unknown, but can be calculated by C<strlen()>, since it is
639 terminated by a NUL, which isn't included in its length.
640
a3815e44 641 The optional 'n' following 'str' means that there is a third parameter,
fc169e00
KW
642 giving the maximum number of bytes to look at in each string. Even if both
643 strings are longer than the length parameter, those extra bytes will be
644 unexamined.
645
646 The 's' suffix means that the 2nd byte string parameter is a literal C
647 double-quoted string. Its length will automatically be calculated by the
648 macro, so no length parameter will ever be needed for it.
649
650 If the prefix is 'mem', the string parameters don't have to be C strings;
651 they may contain embedded NUL bytes, do not necessarily have a terminating
652 NUL, and their lengths can be known only through other means, which in
653 practice are additional parameter(s) passed to the function. All 'mem'
654 functions have at least one length parameter. Barring any 'l' or 's' suffix,
655 there is a single length parameter, in position 3, which applies to both
656 string parameters. The 's' suffix means, as described above, that the 2nd
657 string is a literal double-quoted C string (hence its length is calculated by
658 the macro, and the length parameter to the function applies just to the first
659 string parameter, and hence is positioned just after it). An 'l' suffix
660 means that the 2nd string parameter has its own length parameter, and the
661 signature will look like memFOOl(s1, l1, s2, l2).
662
663 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
664 substring of the 1st string. 'P' if present indicates that the substring
665 must be a "proper" one in tha mathematical sense that the first one must be
666 strictly larger than the 2nd.
667
954c1994
GS
668*/
669
62946e08 670
75400963
KW
671#define strNE(s1,s2) (strcmp(s1,s2) != 0)
672#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
673#define strLT(s1,s2) (strcmp(s1,s2) < 0)
674#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
675#define strGT(s1,s2) (strcmp(s1,s2) > 0)
676#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 677
75400963
KW
678#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
679#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 680
9d3980bc
KW
681#define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
682#define memNE(s1,s2,l) (! memEQ(s1,s2,l))
36477c24 683
085b7534 684/* memEQ and memNE where second comparand is a string constant */
568a785a 685#define memEQs(s1, l, s2) \
777fa2cb 686 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 687#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 688
fdbb9a7c
KW
689/* Keep these private until we decide it was a good idea */
690#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
691
692#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
693
bdb7e3f0 694#define memBEGINs(s1, l, s2) \
30a6480c 695 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 696 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158 697#define memBEGINPs(s1, l, s2) \
30a6480c 698 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
de627158 699 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0 700#define memENDs(s1, l, s2) \
30a6480c 701 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 702 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424 703#define memENDPs(s1, l, s2) \
30a6480c 704 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
b80f8424 705 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 706#endif /* End of making macros private */
bdb7e3f0 707
062b6850
KW
708#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
709#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
710#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
711#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
712
4aada8b9
KW
713#define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
714
bbce6d69
PP
715/*
716 * Character classes.
717 *
718 * Unfortunately, the introduction of locales means that we
719 * can't trust isupper(), etc. to tell the truth. And when
720 * it comes to /\w+/ with tainting enabled, we *must* be able
721 * to trust our character classes.
722 *
723 * Therefore, the default tests in the text of Perl will be
724 * independent of locale. Any code that wants to depend on
725 * the current locale will use the tests that begin with "lc".
726 */
727
2304df62
AD
728#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
729# ifndef CTYPE256
730# define CTYPE256
731# endif
732#endif
733
954c1994 734/*
ccfc67b7 735
dcccc8ff 736=head1 Character classification
243effed
KW
737This section is about functions (really macros) that classify characters
738into types, such as punctuation versus alphabetic, etc. Most of these are
739analogous to regular expression character classes. (See
740L<perlrecharclass/POSIX Character Classes>.) There are several variants for
741each class. (Not all macros have all variants; each item below lists the
742ones valid for it.) None are affected by C<use bytes>, and only the ones
743with C<LC> in the name are affected by the current locale.
744
d713f9d9
KW
745The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
746treating it as a code point, and returns a boolean as to whether or not the
747character represented by it is (or on non-ASCII platforms, corresponds to) an
6aff1f14
KW
748ASCII character in the named class based on platform, Unicode, and Perl rules.
749If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 750
c98722a4 751Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
752with no suffix C<"_A">. This variant is used to emphasize by its name that
753only ASCII-range characters can return TRUE.
4b9734bf 754
d60679e1 755Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
756onto the platform. That is, the code points that are ASCII are unaffected,
757since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
758as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
759true when called with the code point 0xDF, which is a word character in both
4650c663 760ASCII and EBCDIC (though it represents different characters in each).
d713f9d9
KW
761If the input is a number that doesn't fit in an octet, FALSE is returned.
762(Perl's documentation uses a colloquial definition of Latin-1, to include all
763code points below 256.)
243effed 764
d713f9d9
KW
765Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
766inputs below 256, but if the code point is larger than 255, Unicode rules are
767used to determine if it is in the character class. For example,
d0da05db 768C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 769WITH MACRON in Unicode, and is a word character.
243effed 770
059703b0
KW
771Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
772but are used for UTF-8 encoded strings. The two forms are different names for
773the same thing. Each call to one of these classifies the first character of
774the string starting at C<p>. The second parameter, C<e>, points to anywhere in
775the string beyond the first character, up to one byte past the end of the
776entire string. Although both variants are identical, the suffix C<_safe> in
777one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
778provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
779C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
780some way, the program may croak, or the function may return FALSE, at the
781discretion of the implementation, and subject to change in future releases.
243effed 782
d713f9d9
KW
783Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
784but the result is based on the current locale, which is what C<LC> in the name
785stands for. If Perl can determine that the current locale is a UTF-8 locale,
786it uses the published Unicode rules; otherwise, it uses the C library function
787that gives the named classification. For example, C<isDIGIT_LC()> when not in
788a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
789returned if the input won't fit into an octet. On some platforms where the C
790library function is known to be defective, Perl changes its result to follow
791the POSIX standard's rules.
243effed 792
d713f9d9
KW
793Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
794than 256, but for larger ones it returns the Unicode classification of the code
795point.
243effed 796
059703b0
KW
797Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
798C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
799are different names for the same thing. Each call to one of these classifies
800the first character of the string starting at C<p>. The second parameter,
801C<e>, points to anywhere in the string beyond the first character, up to one
802byte past the end of the entire string. Although both variants are identical,
803the suffix C<_safe> in one name emphasizes that it will not attempt to read
804beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
805is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
806character is malformed in some way, the program may croak, or the function may
807return FALSE, at the discretion of the implementation, and subject to change in
808future releases.
ccfc67b7 809
d713f9d9
KW
810=for apidoc Am|bool|isALPHA|int ch
811Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
812analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
813See the L<top of this section|/Character classification> for an explanation of
814variants
059703b0
KW
815C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>,
816C<isALPHA_utf8_safe>, C<isALPHA_LC>, C<isALPHA_LC_uvchr>, C<isALPHA_LC_utf8>,
817and C<isALPHA_LC_utf8_safe>.
8a58bdcf 818
f16858ed
KW
819=cut
820
821Here and below, we add the protoypes of these macros for downstream programs
822that would be interested in them, such as Devel::PPPort
823
824=for apidoc Amh|bool|isALPHA_A|int ch
825=for apidoc Amh|bool|isALPHA_L1|int ch
826=for apidoc Amh|bool|isALPHA_uvchr|int ch
827=for apidoc Amh|bool|isALPHA_utf8_safe|U8 * s|U8 * end
d23c7e08 828=for apidoc Amh|bool|isALPHA_utf8|U8 * s|U8 * end
f16858ed
KW
829=for apidoc Amh|bool|isALPHA_LC|int ch
830=for apidoc Amh|bool|isALPHA_LC_uvchr|int ch
831=for apidoc Amh|bool|isALPHA_LC_utf8_safe|U8 * s| U8 *end
832
d713f9d9
KW
833=for apidoc Am|bool|isALPHANUMERIC|int ch
834Returns a boolean indicating whether the specified character is one of
835C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
836See the L<top of this section|/Character classification> for an explanation of
837variants
d0da05db 838C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
059703b0
KW
839C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>,
840C<isALPHANUMERIC_LC_uvchr>, C<isALPHANUMERIC_LC_utf8>, and
841C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 842
255b632a
KW
843A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
844this corresponds to the C language alphanumeric definition). Also
845there are the variants
846C<isALNUMC_A>, C<isALNUMC_L1>
847C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
848
f16858ed
KW
849=for apidoc Amh|bool|isALPHANUMERIC_A|int ch
850=for apidoc Amh|bool|isALPHANUMERIC_L1|int ch
851=for apidoc Amh|bool|isALPHANUMERIC_uvchr|int ch
852=for apidoc Amh|bool|isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
d23c7e08 853=for apidoc Amh|bool|isALPHANUMERIC_utf8|U8 * s|U8 * end
f16858ed
KW
854=for apidoc Amh|bool|isALPHANUMERIC_LC|int ch
855=for apidoc Amh|bool|isALPHANUMERIC_LC_uvchr|int ch
856=for apidoc Amh|bool|isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
857=for apidoc Amh|bool|isALNUMC|int ch
858=for apidoc Amh|bool|isALNUMC_A|int ch
859=for apidoc Amh|bool|isALNUMC_L1|int ch
860=for apidoc Amh|bool|isALNUMC_LC|int ch
861=for apidoc Amh|bool|isALNUMC_LC_uvchr|int ch
862
d713f9d9 863=for apidoc Am|bool|isASCII|int ch
8a58bdcf 864Returns a boolean indicating whether the specified character is one of the 128
243effed 865characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 866On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
867character corresponds to an ASCII character. Variants C<isASCII_A()> and
868C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
869See the L<top of this section|/Character classification> for an explanation of
870variants
059703b0
KW
871C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_utf8_safe>, C<isASCII_LC>,
872C<isASCII_LC_uvchr>, C<isASCII_LC_utf8>, and C<isASCII_LC_utf8_safe>.
873Note, however, that some platforms do not have the C library routine
874C<isascii()>. In these cases, the variants whose names contain C<LC> are the
875same as the corresponding ones without.
243effed 876
f16858ed
KW
877=for apidoc Amh|bool|isASCII_A|int ch
878=for apidoc Amh|bool|isASCII_L1|int ch
879=for apidoc Amh|bool|isASCII_uvchr|int ch
880=for apidoc Amh|bool|isASCII_utf8_safe|U8 * s|U8 * end
d23c7e08 881=for apidoc Amh|bool|isASCII_utf8|U8 * s|U8 * end
f16858ed
KW
882=for apidoc Amh|bool|isASCII_LC|int ch
883=for apidoc Amh|bool|isASCII_LC_uvchr|int ch
884=for apidoc Amh|bool|isASCII_LC_utf8_safe|U8 * s| U8 *end
885
d98532ea
KW
886Also note, that because all ASCII characters are UTF-8 invariant (meaning they
887have the exact same representation (always a single byte) whether encoded in
888UTF-8 or not), C<isASCII> will give the correct results when called with any
059703b0
KW
889byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
890C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
d98532ea 891
243effed
KW
892=for apidoc Am|bool|isBLANK|char ch
893Returns a boolean indicating whether the specified character is a
6aff1f14 894character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
895See the L<top of this section|/Character classification> for an explanation of
896variants
059703b0
KW
897C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>,
898C<isBLANK_utf8_safe>, C<isBLANK_LC>, C<isBLANK_LC_uvchr>, C<isBLANK_LC_utf8>,
899and C<isBLANK_LC_utf8_safe>. Note,
da8c1a98
KW
900however, that some platforms do not have the C library routine
901C<isblank()>. In these cases, the variants whose names contain C<LC> are
902the same as the corresponding ones without.
243effed 903
f16858ed
KW
904=for apidoc Amh|bool|isBLANK_A|int ch
905=for apidoc Amh|bool|isBLANK_L1|int ch
906=for apidoc Amh|bool|isBLANK_uvchr|int ch
907=for apidoc Amh|bool|isBLANK_utf8_safe|U8 * s|U8 * end
d23c7e08 908=for apidoc Amh|bool|isBLANK_utf8|U8 * s|U8 * end
f16858ed
KW
909=for apidoc Amh|bool|isBLANK_LC|int ch
910=for apidoc Amh|bool|isBLANK_LC_uvchr|int ch
911=for apidoc Amh|bool|isBLANK_LC_utf8_safe|U8 * s| U8 *end
912
243effed
KW
913=for apidoc Am|bool|isCNTRL|char ch
914Returns a boolean indicating whether the specified character is a
6aff1f14 915control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
916See the L<top of this section|/Character classification> for an explanation of
917variants
059703b0
KW
918C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>,
919C<isCNTRL_utf8_safe>, C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, C<isCNTRL_LC_utf8>
920and C<isCNTRL_LC_utf8_safe>. On EBCDIC
da8c1a98 921platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 922
f16858ed
KW
923=for apidoc Amh|bool|isCNTRL_A|int ch
924=for apidoc Amh|bool|isCNTRL_L1|int ch
925=for apidoc Amh|bool|isCNTRL_uvchr|int ch
926=for apidoc Amh|bool|isCNTRL_utf8_safe|U8 * s|U8 * end
d23c7e08 927=for apidoc Amh|bool|isCNTRL_utf8|U8 * s|U8 * end
f16858ed
KW
928=for apidoc Amh|bool|isCNTRL_LC|int ch
929=for apidoc Amh|bool|isCNTRL_LC_uvchr|int ch
930=for apidoc Amh|bool|isCNTRL_LC_utf8_safe|U8 * s| U8 *end
931
954c1994 932=for apidoc Am|bool|isDIGIT|char ch
2787a470 933Returns a boolean indicating whether the specified character is a
6aff1f14 934digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 935Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
936See the L<top of this section|/Character classification> for an explanation of
937variants
059703b0
KW
938C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>,
939C<isDIGIT_LC_uvchr>, C<isDIGIT_LC_utf8>, and C<isDIGIT_LC_utf8_safe>.
243effed 940
f16858ed
KW
941=for apidoc Amh|bool|isDIGIT_A|int ch
942=for apidoc Amh|bool|isDIGIT_L1|int ch
943=for apidoc Amh|bool|isDIGIT_uvchr|int ch
944=for apidoc Amh|bool|isDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 945=for apidoc Amh|bool|isDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
946=for apidoc Amh|bool|isDIGIT_LC|int ch
947=for apidoc Amh|bool|isDIGIT_LC_uvchr|int ch
948=for apidoc Amh|bool|isDIGIT_LC_utf8_safe|U8 * s| U8 *end
949
243effed
KW
950=for apidoc Am|bool|isGRAPH|char ch
951Returns a boolean indicating whether the specified character is a
6aff1f14 952graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 953See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
954variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>,
955C<isGRAPH_utf8_safe>, C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>,
956C<isGRAPH_LC_utf8_safe>, and C<isGRAPH_LC_utf8_safe>.
954c1994 957
f16858ed
KW
958=for apidoc Amh|bool|isGRAPH_A|int ch
959=for apidoc Amh|bool|isGRAPH_L1|int ch
960=for apidoc Amh|bool|isGRAPH_uvchr|int ch
961=for apidoc Amh|bool|isGRAPH_utf8_safe|U8 * s|U8 * end
d23c7e08 962=for apidoc Amh|bool|isGRAPH_utf8|U8 * s|U8 * end
f16858ed
KW
963=for apidoc Amh|bool|isGRAPH_LC|int ch
964=for apidoc Amh|bool|isGRAPH_LC_uvchr|int ch
965=for apidoc Amh|bool|isGRAPH_LC_utf8_safe|U8 * s| U8 *end
966
0c82b6df 967=for apidoc Am|bool|isLOWER|char ch
2787a470 968Returns a boolean indicating whether the specified character is a
6aff1f14 969lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
970See the L<top of this section|/Character classification> for an explanation of
971variants
059703b0
KW
972C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>,
973C<isLOWER_utf8_safe>, C<isLOWER_LC>, C<isLOWER_LC_uvchr>, C<isLOWER_LC_utf8>,
974and C<isLOWER_LC_utf8_safe>.
0c82b6df 975
f16858ed
KW
976=for apidoc Amh|bool|isLOWER_A|int ch
977=for apidoc Amh|bool|isLOWER_L1|int ch
978=for apidoc Amh|bool|isLOWER_uvchr|int ch
979=for apidoc Amh|bool|isLOWER_utf8_safe|U8 * s|U8 * end
d23c7e08 980=for apidoc Amh|bool|isLOWER_utf8|U8 * s|U8 * end
f16858ed
KW
981=for apidoc Amh|bool|isLOWER_LC|int ch
982=for apidoc Amh|bool|isLOWER_LC_uvchr|int ch
983=for apidoc Amh|bool|isLOWER_LC_utf8_safe|U8 * s| U8 *end
984
c99e91e9 985=for apidoc Am|bool|isOCTAL|char ch
2787a470 986Returns a boolean indicating whether the specified character is an
6aff1f14 987octal digit, [0-7].
243effed
KW
988The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
989C<isOCTAL>.
990
f16858ed
KW
991=for apidoc Amh|bool|isOCTAL_A|int ch
992=for apidoc Amh|bool|isOCTAL_L1|int ch
993
243effed
KW
994=for apidoc Am|bool|isPUNCT|char ch
995Returns a boolean indicating whether the specified character is a
6aff1f14
KW
996punctuation character, analogous to C<m/[[:punct:]]/>.
997Note that the definition of what is punctuation isn't as
243effed
KW
998straightforward as one might desire. See L<perlrecharclass/POSIX Character
999Classes> for details.
dcccc8ff 1000See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
1001variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>,
1002C<isPUNCT_utf8_safe>, C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, C<isPUNCT_LC_utf8>,
1003and C<isPUNCT_LC_utf8_safe>.
c99e91e9 1004
f16858ed
KW
1005=for apidoc Amh|bool|isPUNCT_A|int ch
1006=for apidoc Amh|bool|isPUNCT_L1|int ch
1007=for apidoc Amh|bool|isPUNCT_uvchr|int ch
1008=for apidoc Amh|bool|isPUNCT_utf8_safe|U8 * s|U8 * end
d23c7e08 1009=for apidoc Amh|bool|isPUNCT_utf8|U8 * s|U8 * end
f16858ed
KW
1010=for apidoc Amh|bool|isPUNCT_LC|int ch
1011=for apidoc Amh|bool|isPUNCT_LC_uvchr|int ch
1012=for apidoc Amh|bool|isPUNCT_LC_utf8_safe|U8 * s| U8 *end
1013
0c82b6df 1014=for apidoc Am|bool|isSPACE|char ch
2787a470 1015Returns a boolean indicating whether the specified character is a
6aff1f14 1016whitespace character. This is analogous
398d098a 1017to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 1018this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
1019locale forms of this macro (the ones with C<LC> in their names) matched
1020precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
1021in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
1022(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
1023See the L<top of this section|/Character classification> for an explanation of
1024variants
059703b0
KW
1025C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>,
1026C<isSPACE_utf8_safe>, C<isSPACE_LC>, C<isSPACE_LC_uvchr>, C<isSPACE_LC_utf8>,
1027and C<isSPACE_LC_utf8_safe>.
0c82b6df 1028
f16858ed
KW
1029=for apidoc Amh|bool|isSPACE_A|int ch
1030=for apidoc Amh|bool|isSPACE_L1|int ch
1031=for apidoc Amh|bool|isSPACE_uvchr|int ch
1032=for apidoc Amh|bool|isSPACE_utf8_safe|U8 * s|U8 * end
d23c7e08 1033=for apidoc Amh|bool|isSPACE_utf8|U8 * s|U8 * end
f16858ed
KW
1034=for apidoc Amh|bool|isSPACE_LC|int ch
1035=for apidoc Amh|bool|isSPACE_LC_uvchr|int ch
1036=for apidoc Amh|bool|isSPACE_LC_utf8_safe|U8 * s| U8 *end
1037
398d098a
KW
1038=for apidoc Am|bool|isPSXSPC|char ch
1039(short for Posix Space)
779cf272
KW
1040Starting in 5.18, this is identical in all its forms to the
1041corresponding C<isSPACE()> macros.
398d098a
KW
1042The locale forms of this macro are identical to their corresponding
1043C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
1044non-locale forms differ from their C<isSPACE()> forms only in that the
1045C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
1046Otherwise they are identical. Thus this macro is analogous to what
1047C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 1048See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
1049variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
1050C<isPSXSPC_utf8_safe>, C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>,
1051C<isPSXSPC_LC_utf8>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 1052
f16858ed
KW
1053=for apidoc Amh|bool|isPSXSPC_A|int ch
1054=for apidoc Amh|bool|isPSXSPC_L1|int ch
1055=for apidoc Amh|bool|isPSXSPC_uvchr|int ch
1056=for apidoc Amh|bool|isPSXSPC_utf8_safe|U8 * s|U8 * end
d23c7e08 1057=for apidoc Amh|bool|isPSXSPC_utf8|U8 * s|U8 * end
f16858ed
KW
1058=for apidoc Amh|bool|isPSXSPC_LC|int ch
1059=for apidoc Amh|bool|isPSXSPC_LC_uvchr|int ch
1060=for apidoc Amh|bool|isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
1061
954c1994 1062=for apidoc Am|bool|isUPPER|char ch
2787a470 1063Returns a boolean indicating whether the specified character is an
6aff1f14 1064uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 1065See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
1066variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>,
1067C<isUPPER_utf8_safe>, C<isUPPER_LC>, C<isUPPER_LC_uvchr>, C<isUPPER_LC_utf8>,
1068and C<isUPPER_LC_utf8_safe>.
954c1994 1069
f16858ed
KW
1070=for apidoc Amh|bool|isUPPER_A|int ch
1071=for apidoc Amh|bool|isUPPER_L1|int ch
1072=for apidoc Amh|bool|isUPPER_uvchr|int ch
1073=for apidoc Amh|bool|isUPPER_utf8_safe|U8 * s|U8 * end
d23c7e08 1074=for apidoc Amh|bool|isUPPER_utf8|U8 * s|U8 * end
f16858ed
KW
1075=for apidoc Amh|bool|isUPPER_LC|int ch
1076=for apidoc Amh|bool|isUPPER_LC_uvchr|int ch
1077=for apidoc Amh|bool|isUPPER_LC_utf8_safe|U8 * s| U8 *end
1078
243effed 1079=for apidoc Am|bool|isPRINT|char ch
8eea39dd 1080Returns a boolean indicating whether the specified character is a
6aff1f14 1081printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
1082See the L<top of this section|/Character classification> for an explanation of
1083variants
059703b0
KW
1084C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>,
1085C<isPRINT_utf8_safe>, C<isPRINT_LC>, C<isPRINT_LC_uvchr>, C<isPRINT_LC_utf8>,
1086and C<isPRINT_LC_utf8_safe>.
243effed 1087
f16858ed
KW
1088=for apidoc Amh|bool|isPRINT_A|int ch
1089=for apidoc Amh|bool|isPRINT_L1|int ch
1090=for apidoc Amh|bool|isPRINT_uvchr|int ch
1091=for apidoc Amh|bool|isPRINT_utf8_safe|U8 * s|U8 * end
d23c7e08 1092=for apidoc Amh|bool|isPRINT_utf8|U8 * s|U8 * end
f16858ed
KW
1093=for apidoc Amh|bool|isPRINT_LC|int ch
1094=for apidoc Amh|bool|isPRINT_LC_uvchr|int ch
1095=for apidoc Amh|bool|isPRINT_LC_utf8_safe|U8 * s| U8 *end
1096
243effed
KW
1097=for apidoc Am|bool|isWORDCHAR|char ch
1098Returns a boolean indicating whether the specified character is a character
1099that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1100in a regular expression. A word character is an alphabetic character, a
1101decimal digit, a connecting punctuation character (such as an underscore), or
1102a "mark" character that attaches to one of those (like some sort of accent).
1103C<isALNUM()> is a synonym provided for backward compatibility, even though a
1104word character includes more than the standard C language meaning of
1105alphanumeric.
dcccc8ff 1106See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
1107variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1108C<isWORDCHAR_utf8>, and C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>,
1109C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and C<isWORDCHAR_LC_utf8_safe>
1110are also as described there, but additionally include the platform's native
1111underscore.
8a58bdcf 1112
f16858ed
KW
1113=for apidoc Amh|bool|isWORDCHAR_A|int ch
1114=for apidoc Amh|bool|isWORDCHAR_L1|int ch
1115=for apidoc Amh|bool|isWORDCHAR_uvchr|int ch
1116=for apidoc Amh|bool|isWORDCHAR_utf8_safe|U8 * s|U8 * end
d23c7e08 1117=for apidoc Amh|bool|isWORDCHAR_utf8|U8 * s|U8 * end
f16858ed
KW
1118=for apidoc Amh|bool|isWORDCHAR_LC|int ch
1119=for apidoc Amh|bool|isWORDCHAR_LC_uvchr|int ch
1120=for apidoc Amh|bool|isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
1121=for apidoc Amh|bool|isALNUM|int ch
1122=for apidoc Amh|bool|isALNUM_A|int ch
1123=for apidoc Amh|bool|isALNUM_LC|int ch
1124=for apidoc Amh|bool|isALNUM_LC_uvchr|int ch
1125
8a58bdcf
KW
1126=for apidoc Am|bool|isXDIGIT|char ch
1127Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
1128digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1129and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
1130See the L<top of this section|/Character classification> for an explanation of
1131variants
059703b0
KW
1132C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>,
1133C<isXDIGIT_LC_uvchr>, C<isXDIGIT_LC_utf8>, and C<isXDIGIT_LC_utf8_safe>.
243effed 1134
f16858ed
KW
1135=for apidoc Amh|bool|isXDIGIT_A|int ch
1136=for apidoc Amh|bool|isXDIGIT_L1|int ch
1137=for apidoc Amh|bool|isXDIGIT_uvchr|int ch
1138=for apidoc Amh|bool|isXDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 1139=for apidoc Amh|bool|isXDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
1140=for apidoc Amh|bool|isXDIGIT_LC|int ch
1141=for apidoc Amh|bool|isXDIGIT_LC_uvchr|int ch
1142=for apidoc Amh|bool|isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1143
3c3ecf18
KW
1144=for apidoc Am|bool|isIDFIRST|char ch
1145Returns a boolean indicating whether the specified character can be the first
1146character of an identifier. This is very close to, but not quite the same as
1147the official Unicode property C<XID_Start>. The difference is that this
1148returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
1149See the L<top of this section|/Character classification> for an explanation of
1150variants
059703b0
KW
1151C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
1152C<isIDFIRST_utf8_safe>, C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>,
1153C<isIDFIRST_LC_utf8>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18 1154
f16858ed
KW
1155=for apidoc Amh|bool|isIDFIRST_A|int ch
1156=for apidoc Amh|bool|isIDFIRST_L1|int ch
1157=for apidoc Amh|bool|isIDFIRST_uvchr|int ch
1158=for apidoc Amh|bool|isIDFIRST_utf8_safe|U8 * s|U8 * end
d23c7e08 1159=for apidoc Amh|bool|isIDFIRST_utf8|U8 * s|U8 * end
f16858ed
KW
1160=for apidoc Amh|bool|isIDFIRST_LC|int ch
1161=for apidoc Amh|bool|isIDFIRST_LC_uvchr|int ch
1162=for apidoc Amh|bool|isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1163
3c3ecf18
KW
1164=for apidoc Am|bool|isIDCONT|char ch
1165Returns a boolean indicating whether the specified character can be the
1166second or succeeding character of an identifier. This is very close to, but
1167not quite the same as the official Unicode property C<XID_Continue>. The
1168difference is that this returns true only if the input character also matches
dcccc8ff 1169L</isWORDCHAR>. See the L<top of this section|/Character classification> for
059703b0
KW
1170an explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
1171C<isIDCONT_utf8>, C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>,
1172C<isIDCONT_LC_utf8>, and C<isIDCONT_LC_utf8_safe>.
3c3ecf18 1173
f16858ed
KW
1174=for apidoc Amh|bool|isIDCONT_A|int ch
1175=for apidoc Amh|bool|isIDCONT_L1|int ch
1176=for apidoc Amh|bool|isIDCONT_uvchr|int ch
1177=for apidoc Amh|bool|isIDCONT_utf8_safe|U8 * s|U8 * end
d23c7e08 1178=for apidoc Amh|bool|isIDCONT_utf8|U8 * s|U8 * end
f16858ed
KW
1179=for apidoc Amh|bool|isIDCONT_LC|int ch
1180=for apidoc Amh|bool|isIDCONT_LC_uvchr|int ch
1181=for apidoc Amh|bool|isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1182
51b56f5c 1183=for apidoc_section Numeric Functions
8eea39dd 1184
95a59cab 1185=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 1186Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
1187Behaviour is only well defined when isXDIGIT(*str) is true.
1188
e7c1e6c1 1189=head1 Character case changing
21da7284
KW
1190Perl uses "full" Unicode case mappings. This means that converting a single
1191character to another case may result in a sequence of more than one character.
1192For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1193character sequence C<SS>. This presents some complications The lowercase of
1194all characters in the range 0..255 is a single character, and thus
1195C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1196return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1197an API that does allow every possible legal result to be returned.) Likewise
1198no other function that is crippled by not being able to give the correct
1199results for the full range of possible inputs has been implemented here.
e7c1e6c1 1200
d713f9d9 1201=for apidoc Am|U8|toUPPER|int ch
1f607577
KW
1202Converts the specified character to uppercase. If the input is anything but an
1203ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 1204C<toUPPER_A> is equivalent.
954c1994 1205
d0da05db
KW
1206=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1207Converts the code point C<cp> to its uppercase version, and
1208stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1209point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1210that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1211bytes since the uppercase version may be longer than the original character.
1212
1213The first code point of the uppercased version is returned
21da7284
KW
1214(but note, as explained at L<the top of this section|/Character case
1215changing>, that there may be more.)
1f607577 1216
059703b0 1217=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141 1218=for apidoc_item toUPPER_utf8_safe
a239b1e2
KW
1219Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1220extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
1221stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1222that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1223bytes since the uppercase version may be longer than the original character.
1224
1225The first code point of the uppercased version is returned
21da7284
KW
1226(but note, as explained at L<the top of this section|/Character case
1227changing>, that there may be more).
1f607577 1228
059703b0
KW
1229It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1230S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1231the UTF-8 for the input character is malformed in some way, the program may
1232croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1233of the implementation, and subject to change in future releases.
a239b1e2 1234
912c8141 1235C<toUPPER_utf8_safe> is now just a different spelling of plain C<toUPPER_utf8>
1f607577 1236
25200305
KW
1237=for apidoc Am|U8|toFOLD|U8 ch
1238Converts the specified character to foldcase. If the input is anything but an
1239ASCII uppercase character, that input character itself is returned. Variant
1240C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 1241Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 1242
d0da05db
KW
1243=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1244Converts the code point C<cp> to its foldcase version, and
1245stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1246point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1247that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1248bytes since the foldcase version may be longer than the original character.
1249
1250The first code point of the foldcased version is returned
21da7284
KW
1251(but note, as explained at L<the top of this section|/Character case
1252changing>, that there may be more).
1f607577 1253
059703b0 1254=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141 1255=for apidoc_item toFOLD_utf8_safe
a239b1e2
KW
1256Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1257extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
1258stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1259that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1260bytes since the foldcase version may be longer than the original character.
1261
1262The first code point of the foldcased version is returned
21da7284
KW
1263(but note, as explained at L<the top of this section|/Character case
1264changing>, that there may be more).
1f607577 1265
059703b0 1266It will not attempt
a239b1e2
KW
1267to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1268true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1269input character is malformed in some way, the program may croak, or the
1270function may return the REPLACEMENT CHARACTER, at the discretion of the
1271implementation, and subject to change in future releases.
1272
912c8141 1273C<toFOLD_utf8_safe> is now just a different spelling of plain C<toFOLD_utf8>
1f607577
KW
1274
1275=for apidoc Am|U8|toLOWER|U8 ch
1276Converts the specified character to lowercase. If the input is anything but an
1277ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 1278C<toLOWER_A> is equivalent.
954c1994 1279
1f607577 1280=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
1281Converts the specified Latin1 character to lowercase. The results are
1282undefined if the input doesn't fit in a byte.
1f607577
KW
1283
1284=for apidoc Am|U8|toLOWER_LC|U8 ch
1285Converts the specified character to lowercase using the current locale's rules,
1286if possible; otherwise returns the input character itself.
1287
d0da05db
KW
1288=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1289Converts the code point C<cp> to its lowercase version, and
1290stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1291point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1292that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1293bytes since the lowercase version may be longer than the original character.
1294
1295The first code point of the lowercased version is returned
21da7284
KW
1296(but note, as explained at L<the top of this section|/Character case
1297changing>, that there may be more).
1f607577 1298
bd350c85 1299=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141 1300=for apidoc_item toLOWER_utf8_safe
a239b1e2
KW
1301Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1302extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
1303stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1304that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1305bytes since the lowercase version may be longer than the original character.
1306
1307The first code point of the lowercased version is returned
21da7284
KW
1308(but note, as explained at L<the top of this section|/Character case
1309changing>, that there may be more).
059703b0
KW
1310It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1311S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1312the UTF-8 for the input character is malformed in some way, the program may
1313croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1314of the implementation, and subject to change in future releases.
1f607577 1315
912c8141 1316C<toLOWER_utf8_safe> is now just a different spelling of plain C<toLOWER_utf8>
1f607577 1317
25200305
KW
1318=for apidoc Am|U8|toTITLE|U8 ch
1319Converts the specified character to titlecase. If the input is anything but an
1320ASCII lowercase character, that input character itself is returned. Variant
b7d90381 1321C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 1322range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 1323not a concept used in locale handling, so there is no functionality for that.)
25200305 1324
d0da05db
KW
1325=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1326Converts the code point C<cp> to its titlecase version, and
1327stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1328point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1329that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1330bytes since the titlecase version may be longer than the original character.
1331
1332The first code point of the titlecased version is returned
21da7284
KW
1333(but note, as explained at L<the top of this section|/Character case
1334changing>, that there may be more).
1f607577 1335
059703b0 1336=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
912c8141
KW
1337=for apidoc_item toTITLE_utf8_safe
1338Convert the first UTF-8 encoded character in the sequence starting at C<p> and
a239b1e2 1339extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1340stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1341that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1342bytes since the titlecase version may be longer than the original character.
1343
1344The first code point of the titlecased version is returned
21da7284
KW
1345(but note, as explained at L<the top of this section|/Character case
1346changing>, that there may be more).
1f607577 1347
059703b0 1348It will not attempt
a239b1e2
KW
1349to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1350true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1351input character is malformed in some way, the program may croak, or the
1352function may return the REPLACEMENT CHARACTER, at the discretion of the
1353implementation, and subject to change in future releases.
1354
912c8141 1355C<toTITLE_utf8_safe> is now just a different spelling of plain C<toTITLE_utf8>
1f607577 1356
954c1994 1357=cut
353c9b6f 1358
d0da05db 1359XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1360really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1361and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1362
8a58bdcf 1363Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1364patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1365
954c1994
GS
1366*/
1367
8f5283f4
KW
1368/*
1369 void below because that's the best fit, and works for Devel::PPPort
51b56f5c 1370=for apidoc_section Integer configuration values
8f5283f4
KW
1371=for apidoc AmnU|void|WIDEST_UTYPE
1372
1373Yields the widest unsigned integer type on the platform, currently either
1374C<U32> or C<64>. This can be used in declarations such as
1375
1376 WIDEST_UTYPE my_uv;
1377
1378or casts
1379
1380 my_uv = (WIDEST_UTYPE) val;
1381
1382=cut
1383
1384*/
de9e2639
KW
1385#ifdef QUADKIND
1386# define WIDEST_UTYPE U64
7c062697
KW
1387#else
1388# define WIDEST_UTYPE U32
1389#endif
1390
3912bc88
KW
1391/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1392 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1393 * bits of information are lost even on a 64-bit machine, but to get the
1394 * compiler to optimize it out if possible. This is because Configure makes
1395 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1396 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1397 *
1398 * For Coverity, be always true, because otherwise Coverity thinks
1399 * it finds several expressions that are always true, independent
1400 * of operands. Well, they are, but that is kind of the point.
220c71bf 1401 */
7e75d1a1 1402#ifndef __COVERITY__
6c5b02ac
KW
1403 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1404 * pointer) */
1405#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1406 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1407#else
1408#define FITS_IN_8_BITS(c) (1)
1409#endif
cf301eb7 1410
45f4bb73 1411/* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
833b0f46 1412 * Written this way so that after optimization, only one conditional test is
76d3ad4c
KW
1413 * needed. (The NV casts stop any warnings about comparison always being true
1414 * if called with an unsigned. The cast preserves the sign, which is all we
1415 * care about.) */
1416#define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1417 __ASSERT_((NV) (n) >= 0) \
1418 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
833b0f46 1419
94250c4f
KW
1420/* Returns true if c is in the range l..u, where 'l' is non-negative
1421 * Written this way so that after optimization, only one conditional test is
4758c20d 1422 * needed. */
1eaefa6e 1423#define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
4758c20d 1424 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
4758c20d
KW
1425 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1426 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
45f4bb73 1427 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
305fe86e 1428
41f43cc2 1429#ifdef EBCDIC
b6340bd0 1430# ifndef _ALL_SOURCE
0852beac
KW
1431 /* The native libc isascii() et.al. functions return the wrong results
1432 * on at least z/OS unless this is defined. */
b6340bd0
KW
1433# error _ALL_SOURCE should probably be defined
1434# endif
41f43cc2 1435#else
0852beac
KW
1436 /* There is a simple definition of ASCII for ASCII platforms. But the
1437 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1438 * other macros below.
3f3c579d
KW
1439 *
1440 * The cast here is used instead of '(c) >= 0', because some compilers emit
1441 * a warning that that test is always true when the parameter is an
1442 * unsigned type. khw supposes that it could be written as
1443 * && ((c) == '\0' || (c) > 0)
1444 * to avoid the message, but the cast will likely avoid extra branches even
1445 * with stupid compilers.
1446 *
1447 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1448 * a pointer) */
9c903d59 1449# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1450#endif
1451
38694112
KW
1452/* Take the eight possible bit patterns of the lower 3 bits and you get the
1453 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1454 * can be ignored. If the rest match '0', we have an octal */
1455#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1456
9fb1bf9d 1457#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1458
a500dc72
KW
1459/* Character class numbers. For internal core Perl use only. The ones less
1460 * than 32 are used in PL_charclass[] and the ones up through the one that
1461 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1462 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1463 * their actual definitions are here. If that file has a name not used here,
1464 * it won't compile.
1709d539
KW
1465 *
1466 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561 1467 * frequency of their use. This gives a slight edge to exiting a loop earlier
58a3ba2c
KW
1468 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1469 * real Posix character class, and some (small) inefficiencies in regular
1470 * expression handling would be introduced by putting it in the middle of those
1471 * that are. Also, cntrl and ascii come after the others as it may be useful
1472 * to group these which have no members that match above Latin1, (or above
1473 * ASCII in the latter case) */
1474
1709d539
KW
1475# define _CC_WORDCHAR 0 /* \w and [:word:] */
1476# define _CC_DIGIT 1 /* \d and [:digit:] */
1477# define _CC_ALPHA 2 /* [:alpha:] */
1478# define _CC_LOWER 3 /* [:lower:] */
1479# define _CC_UPPER 4 /* [:upper:] */
1480# define _CC_PUNCT 5 /* [:punct:] */
1481# define _CC_PRINT 6 /* [:print:] */
15861f94 1482# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1483# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1484# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
779cf272 1485# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1486# define _CC_BLANK 11 /* [:blank:] */
1487# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1488# define _CC_CNTRL 13 /* [:cntrl:] */
1489# define _CC_ASCII 14 /* [:ascii:] */
1490# define _CC_VERTSPACE 15 /* \v */
1709d539 1491
a0947d7b
KW
1492# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1493
1709d539 1494/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1495 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1496# define _CC_IDFIRST 16
1497# define _CC_CHARNAME_CONT 17
1498# define _CC_NONLATIN1_FOLD 18
1499# define _CC_NONLATIN1_SIMPLE_FOLD 19
1500# define _CC_QUOTEMETA 20
1501# define _CC_NON_FINAL_FOLD 21
1502# define _CC_IS_IN_SOME_FOLD 22
2ae9030c
KW
1503# define _CC_BINDIGIT 23
1504# define _CC_OCTDIGIT 24
1505# define _CC_MNEMONIC_CNTRL 25
073c22b3
KW
1506
1507/* This next group is only used on EBCDIC platforms, so theoretically could be
1508 * shared with something entirely different that's only on ASCII platforms */
abb8abf6 1509# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
0654f0ab 1510/* Unused: 26-30
f4cdb42c
KW
1511 * If more bits are needed, one could add a second word for non-64bit
1512 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1513 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1514 * is used only for optimization (as of this writing), and differs in the
1515 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1516 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1517 * would just cause /i regexes which match them to run less efficiently.
1518 * Similarly the EBCDIC-only bits are used just for speed, and could be
1519 * replaced by other means */
96ac0975 1520
3a371f2f
KW
1521#if defined(PERL_CORE) || defined(PERL_EXT)
1522/* An enum version of the character class numbers, to help compilers
1523 * optimize */
1524typedef enum {
3a371f2f 1525 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1526 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1527 _CC_ENUM_ASCII = _CC_ASCII,
1528 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1529 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1530 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1531 _CC_ENUM_DIGIT = _CC_DIGIT,
1532 _CC_ENUM_GRAPH = _CC_GRAPH,
1533 _CC_ENUM_LOWER = _CC_LOWER,
1534 _CC_ENUM_PRINT = _CC_PRINT,
1535 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1536 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1537 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1538 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1539 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1540 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1541} _char_class_number;
1542#endif
1543
86f72d56 1544#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1545
6635f04f 1546START_EXTERN_C
96ac0975
NC
1547# ifdef DOINIT
1548EXTCONST U32 PL_charclass[] = {
1549# include "l1_char_class_tab.h"
1550};
1551
1552# else /* ! DOINIT */
1553EXTCONST U32 PL_charclass[];
1554# endif
6635f04f 1555END_EXTERN_C
96ac0975 1556
265c1f46 1557 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1558# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1559
1560 /* For internal core Perl use only: the base macro for defining macros like
1561 * isALPHA */
ff7ecfc3 1562# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1563 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1564
f4cdb42c
KW
1565 /* The mask for the _A versions of the macros; it just adds in the bit for
1566 * ASCII. */
1567# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1568
4650c663
KW
1569 /* For internal core Perl use only: the base macro for defining macros like
1570 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1571 * the ASCII bit are present */
b7d90381
KW
1572# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1573 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1574 == _CC_mask_A(classnum)))
f4cdb42c 1575
26c1d9d8
KW
1576/* On ASCII platforms certain classes form a single range. It's faster to
1577 * special case these. isDIGIT is a single range on all platforms */
b877c1ff
KW
1578# ifdef EBCDIC
1579# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1580# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1581# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1582# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1583# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1584# else
26c1d9d8 1585 /* By folding the upper and lowercase, we can use a single range */
b877c1ff 1586# define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
26c1d9d8 1587# define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
b877c1ff
KW
1588# define isLOWER_A(c) inRANGE(c, 'a', 'z')
1589# define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1590# define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1591# endif
15861f94 1592# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1593# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1594# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
b877c1ff 1595# define isDIGIT_A(c) inRANGE(c, '0', '9')
f4cdb42c
KW
1596# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1597# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
f4cdb42c 1598# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1599# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1600 */
d95f8b6a 1601# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1602# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1603# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1604# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1605
1606 /* continuation character for legal NAME in \N{NAME} */
1607# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1608
1609# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1610# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1611# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1612# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1613# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1614# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1615# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1616# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1617# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1618# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1619
0852beac
KW
1620# ifdef EBCDIC
1621# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1622# endif
1623
f12c0118
KW
1624 /* Participates in a single-character fold with a character above 255 */
1625# 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)))
1626
1627 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1628# 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 1629
4eeeb416 1630# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1631# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1632 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1633# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1634 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
5e6ebb12
KW
1635
1636/* is c a control character for which we have a mnemonic? */
1637# if defined(PERL_CORE) || defined(PERL_EXT)
1638# define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1639# endif
687c8d01 1640#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1641
1642 /* If we don't have perl.h, we are compiling a utility program. Below we
1643 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1644 * to it. Most are coded based on first principles. These are written to
74665a89 1645 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
182c4ace 1646# define isDIGIT_A(c) inRANGE(c, '0', '9')
0852beac 1647# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1648# define isSPACE_A(c) (isBLANK_A(c) \
1649 || (c) == '\n' \
1650 || (c) == '\r' \
1651 || (c) == '\v' \
0852beac 1652 || (c) == '\f')
74665a89
KW
1653 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1654 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1655 * performance (if optimization isn't on), and allow the same code to be
1656 * used for both platform types */
182c4ace
KW
1657# define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1658 || inRANGE((c), 'j', 'r') \
1659 || inRANGE((c), 's', 'z')
1660# define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1661 || inRANGE((c), 'J', 'R') \
1662 || inRANGE((c), 'S', 'Z')
a4d7a999
KW
1663# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1664# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1665# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1666# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
182c4ace
KW
1667# define isXDIGIT_A(c) ( isDIGIT_A(c) \
1668 || inRANGE((c), 'a', 'f') \
1669 || inRANGE((c), 'A', 'F')
74665a89
KW
1670# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1671 || (c) == '#' || (c) == '$' || (c) == '%' \
1672 || (c) == '&' || (c) == '\'' || (c) == '(' \
1673 || (c) == ')' || (c) == '*' || (c) == '+' \
1674 || (c) == ',' || (c) == '.' || (c) == '/' \
1675 || (c) == ':' || (c) == ';' || (c) == '<' \
1676 || (c) == '=' || (c) == '>' || (c) == '?' \
1677 || (c) == '@' || (c) == '[' || (c) == '\\' \
1678 || (c) == ']' || (c) == '^' || (c) == '_' \
1679 || (c) == '`' || (c) == '{' || (c) == '|' \
1680 || (c) == '}' || (c) == '~')
1681# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1682# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1683
0852beac 1684# ifdef EBCDIC
74665a89
KW
1685 /* The below is accurate for the 3 EBCDIC code pages traditionally
1686 * supported by perl. The only difference between them in the controls
1687 * is the position of \n, and that is represented symbolically below */
1688# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1689 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1690 || (c) == '\t' || (c) == '\v' \
182c4ace 1691 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
8ec0a736 1692 || (c) == 7F /* U+7F DEL */ \
182c4ace
KW
1693 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1694 DC[1-3] */ \
74665a89
KW
1695 || (c) == 0x18 /* U+18 CAN */ \
1696 || (c) == 0x19 /* U+19 EOM */ \
182c4ace 1697 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
74665a89
KW
1698 || (c) == 0x26 /* U+17 ETB */ \
1699 || (c) == 0x27 /* U+1B ESC */ \
1700 || (c) == 0x2D /* U+05 ENQ */ \
1701 || (c) == 0x2E /* U+06 ACK */ \
1702 || (c) == 0x32 /* U+16 SYN */ \
1703 || (c) == 0x37 /* U+04 EOT */ \
1704 || (c) == 0x3C /* U+14 DC4 */ \
1705 || (c) == 0x3D /* U+15 NAK */ \
1706 || (c) == 0x3F)/* U+1A SUB */
0852beac 1707# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1708# else /* isASCII is already defined for ASCII platforms, so can use that to
1709 define isCNTRL */
1710# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1711# endif
1712
3ffc8c70 1713 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1714 * during debugging, and it seems best to keep them. We may be called
1715 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1716 * do anything anyway, so make it not a problem */
1717# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1718# define NATIVE_TO_LATIN1(ch) (ch)
1719# endif
3ded5eb0
KW
1720# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1721# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1722# define isBLANK_L1(c) (isBLANK_A(c) \
1723 || (FITS_IN_8_BITS(c) \
1724 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1725# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1726# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1727# define isLOWER_L1(c) (isLOWER_A(c) \
1728 || (FITS_IN_8_BITS(c) \
ae683a5f 1729 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1730 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1731 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1732 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1733 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1734# define isPRINT_L1(c) (isPRINT_A(c) \
1735 || (FITS_IN_8_BITS(c) \
1736 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1737# define isPUNCT_L1(c) (isPUNCT_A(c) \
1738 || (FITS_IN_8_BITS(c) \
ae683a5f 1739 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1740 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1741 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1742 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1743 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1744 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1745 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1746# define isSPACE_L1(c) (isSPACE_A(c) \
1747 || (FITS_IN_8_BITS(c) \
ae683a5f 1748 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1749 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1750# define isUPPER_L1(c) (isUPPER_A(c) \
1751 || (FITS_IN_8_BITS(c) \
182c4ace
KW
1752 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1753 0xC0, 0xDE) \
3ded5eb0
KW
1754 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1755# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1756# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1757# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1758 || isBLANK_L1(c) \
1759 || (c) == '-' \
1760 || (c) == '(' \
1761 || (c) == ')')
1762 /* The following are not fully accurate in the above-ASCII range. I (khw)
1763 * don't think it's necessary to be so for the purposes where this gets
1764 * compiled */
1765# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1766# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1767
1768 /* And these aren't accurate at all. They are useful only for above
1769 * Latin1, which utilities and bootstrapping don't deal with */
1770# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1771# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1772# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1773
1774 /* Many of the macros later in this file are defined in terms of these. By
1775 * implementing them with a function, which converts the class number into
1776 * a call to the desired macro, all of the later ones work. However, that
1777 * function won't be actually defined when building a utility program (no
1778 * perl.h), and so a compiler error will be generated if one is attempted
1779 * to be used. And the above-Latin1 code points require Unicode tables to
1780 * be present, something unlikely to be the case when bootstrapping */
1781# define _generic_isCC(c, classnum) \
1782 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1783# define _generic_isCC_A(c, classnum) \
1784 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1785#endif /* End of no perl.h H_PERL */
8a58bdcf 1786
e66b99e9
KW
1787#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1788#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1789#define isASCII_A(c) isASCII(c)
1790#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1791#define isBLANK(c) isBLANK_A(c)
1792#define isCNTRL(c) isCNTRL_A(c)
1793#define isDIGIT(c) isDIGIT_A(c)
1794#define isGRAPH(c) isGRAPH_A(c)
1795#define isIDFIRST(c) isIDFIRST_A(c)
1796#define isLOWER(c) isLOWER_A(c)
1797#define isPRINT(c) isPRINT_A(c)
779cf272 1798#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1799#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1800#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1801#define isPUNCT(c) isPUNCT_A(c)
1802#define isSPACE(c) isSPACE_A(c)
1803#define isUPPER(c) isUPPER_A(c)
1804#define isWORDCHAR(c) isWORDCHAR_A(c)
1805#define isXDIGIT(c) isXDIGIT_A(c)
1806
1807/* ASCII casing. These could also be written as
1808 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1809 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1810 which uses table lookup and mask instead of subtraction. (This would
c5e9991e
KW
1811 work because the _MOD does not apply in the ASCII range).
1812
1813 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1814 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1815 there non-ASCII invariants, and all of them are controls.) */
68067e4e
DM
1816#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1817#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1818
25200305
KW
1819/* In the ASCII range, these are equivalent to what they're here defined to be.
1820 * But by creating these definitions, other code doesn't have to be aware of
c5e9991e
KW
1821 * this detail. Actually this works for all UTF-8 invariants, not just the
1822 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
25200305 1823#define toFOLD(c) toLOWER(c)
25200305
KW
1824#define toTITLE(c) toUPPER(c)
1825
c753c8d3
KW
1826#define toLOWER_A(c) toLOWER(c)
1827#define toUPPER_A(c) toUPPER(c)
25200305
KW
1828#define toFOLD_A(c) toFOLD(c)
1829#define toTITLE_A(c) toTITLE(c)
1a0901db 1830
4650c663 1831/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1832#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1833 ? (c) \
f4cd282c 1834 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1835#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1836
1a0901db 1837/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1838 * all mapped to one of them, and these need special handling; returns the
1839 * input itself if is out-of-range */
b2bf251f 1840#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1841 ? (c) \
f4cd282c 1842 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1843#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1844
beab9ebe
KW
1845/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1846
1847/* For internal core Perl use only: the base macro for defining macros like
1848 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1849 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1850 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1851 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1852 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1853 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1854 * code points under locale rules, use the macros based on _generic_LC_uvchr
1855 * instead of this. */
beab9ebe
KW
1856#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1857 (! FITS_IN_8_BITS(c) \
1858 ? 0 \
31f05a37
KW
1859 : IN_UTF8_CTYPE_LOCALE \
1860 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1861 : cBOOL(non_utf8))
1862
1863/* For internal core Perl use only: a helper macro for defining macros like
1864 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1865 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1866 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1867 * _generic_LC_base for more info */
1868#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1869 _generic_LC_base(c,utf8_locale_classnum, \
1870 non_utf8_func( (_LC_CAST) (c)))
1871
1872/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1873 * 'c' is the platform's native underscore character */
1874#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1875 _generic_LC_base(c, utf8_locale_classnum, \
1876 (non_utf8_func( (_LC_CAST) (c)) \
1877 || (char)(c) == '_'))
1878
1879/* These next three are also for internal core Perl use only: case-change
247985d4
KW
1880 * helper macros. The reason for using the PL_latin arrays is in case the
1881 * system function is defective; it ensures uniform results that conform to the
b257a28c 1882 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
beab9ebe
KW
1883#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1884 ? (c) \
31f05a37
KW
1885 : (IN_UTF8_CTYPE_LOCALE) \
1886 ? PL_latin1_lc[ (U8) (c) ] \
5a10328c 1887 : (cast)function((cast)(c)))
beab9ebe 1888
31f05a37
KW
1889/* Note that the result can be larger than a byte in a UTF-8 locale. It
1890 * returns a single value, so can't adequately return the upper case of LATIN
1891 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1892 * values "SS"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1893 * otherwise returns its input. It does not handle the anomalies in UTF-8
1894 * Turkic locales. */
beab9ebe
KW
1895#define _generic_toUPPER_LC(c, function, cast) \
1896 (! FITS_IN_8_BITS(c) \
1897 ? (c) \
31f05a37 1898 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1899 ? (cast)function((cast)(c)) \
31f05a37
KW
1900 : ((((U8)(c)) == MICRO_SIGN) \
1901 ? GREEK_CAPITAL_LETTER_MU \
1902 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1903 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1904 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1905 ? (__ASSERT_(0) (c)) \
1906 : PL_mod_latin1_uc[ (U8) (c) ])))))
1907
1908/* Note that the result can be larger than a byte in a UTF-8 locale. It
1909 * returns a single value, so can't adequately return the fold case of LATIN
1910 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1911 * values "ss"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1912 * otherwise returns its input. It does not handle the anomalies in UTF-8
1913 * Turkic locales */
beab9ebe 1914#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1915 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1916 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1917 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1918 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1919 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1920
84061b6a 1921/* Use the libc versions for these if available. */
f05550c0 1922#if defined(HAS_ISASCII)
84061b6a
KW
1923# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1924#else
1925# define isASCII_LC(c) isASCII(c)
1926#endif
1927
f05550c0 1928#if defined(HAS_ISBLANK)
84061b6a 1929# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1930#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1931# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1932#endif
1933
f05550c0 1934#define _LC_CAST U8
bbce6d69 1935
f05550c0 1936#ifdef WIN32
375f5f06
KW
1937 /* The Windows functions don't bother to follow the POSIX standard, which
1938 * for example says that something can't both be a printable and a control.
1939 * But Windows treats the \t control as a printable, and does such things
1940 * as making superscripts into both digits and punctuation. This tames
1941 * these flaws by assuming that the definitions of both controls and space
1942 * are correct, and then making sure that other definitions don't have
1943 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1944 * Not all possible weirdnesses are checked for, just the ones that were
1945 * detected on actual Microsoft code pages */
1946
b7d90381
KW
1947# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1948# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1949
1950# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1951 && isALPHANUMERIC_LC(c))
1952# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1953 ! isPUNCT_LC(c))
1954# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1955 isALPHANUMERIC_LC(c))
1956# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1957# define isIDFIRST_LC(c) (((c) == '_') \
1958 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1959# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1960# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1961# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1962# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1963# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1964# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1965 && isALPHANUMERIC_LC(c))
f05550c0
BF
1966
1967# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1968# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1969# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1970
1971#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1972 /* For most other platforms */
beab9ebe 1973
f05550c0
BF
1974# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1975# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1976# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1977# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1978# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1979# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1980# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1981# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1982# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1983# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1984# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1985# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1986# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1987
1988
1989# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1990# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1991# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1992
1993#else /* The final fallback position */
1994
b7d90381
KW
1995# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1996# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1997# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1998# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1999# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 2000# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
2001# define isLOWER_LC(c) (isascii(c) && islower(c))
2002# define isPRINT_LC(c) (isascii(c) && isprint(c))
2003# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
2004# define isSPACE_LC(c) (isascii(c) && isspace(c))
2005# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 2006# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 2007# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
2008
2009# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
2010# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
2011# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 2012
f05550c0 2013#endif
55204971 2014
eba68aa0
KW
2015#define isIDCONT(c) isWORDCHAR(c)
2016#define isIDCONT_A(c) isWORDCHAR_A(c)
2017#define isIDCONT_L1(c) isWORDCHAR_L1(c)
2018#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 2019#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 2020
4650c663 2021/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
2022 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
2023 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
2024 * classes where there is a macro or function 'above_latin1' that takes the
2025 * single argument 'c' and returns the desired value. These exist for those
2366ba44
KW
2026 * classes which have simple definitions, avoiding the overhead of an inversion
2027 * list binary search. _generic_invlist_uvchr() can be used
4650c663 2028 * for classes where that overhead is faster than a direct lookup.
d0da05db 2029 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
2030 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
2031 * have duplicate checks here, so could create versions of the macros that
2032 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
2033
2034/* Note that all ignore 'use bytes' */
1e222e4f
KW
2035#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
2036 ? _generic_isCC(c, classnum) \
cd500f2f 2037 : above_latin1(c))
2366ba44 2038#define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
1e222e4f 2039 ? _generic_isCC(c, classnum) \
922e8cb4 2040 : _is_uni_FOO(classnum, c))
2366ba44
KW
2041#define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
2042#define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
d0da05db
KW
2043#define isASCII_uvchr(c) isASCII(c)
2044#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
2045#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2366ba44
KW
2046#define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
2047#define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
1e222e4f
KW
2048#define isIDCONT_uvchr(c) \
2049 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
2050#define isIDFIRST_uvchr(c) \
2051 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2366ba44
KW
2052#define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
2053#define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
d0da05db 2054
2366ba44 2055#define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
d0da05db
KW
2056#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
2057#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
2058
2366ba44 2059#define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
d0da05db 2060#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2366ba44 2061#define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
d0da05db
KW
2062#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
2063
2064#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
2065#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
2066#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
2067#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
2068
2069/* For backwards compatibility, even though '_uni' should mean official Unicode
2070 * code points, in Perl it means native for those below 256 */
2071#define isALPHA_uni(c) isALPHA_uvchr(c)
2072#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
2073#define isASCII_uni(c) isASCII_uvchr(c)
2074#define isBLANK_uni(c) isBLANK_uvchr(c)
2075#define isCNTRL_uni(c) isCNTRL_uvchr(c)
2076#define isDIGIT_uni(c) isDIGIT_uvchr(c)
2077#define isGRAPH_uni(c) isGRAPH_uvchr(c)
2078#define isIDCONT_uni(c) isIDCONT_uvchr(c)
2079#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
2080#define isLOWER_uni(c) isLOWER_uvchr(c)
2081#define isPRINT_uni(c) isPRINT_uvchr(c)
2082#define isPUNCT_uni(c) isPUNCT_uvchr(c)
2083#define isSPACE_uni(c) isSPACE_uvchr(c)
2084#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
2085#define isUPPER_uni(c) isUPPER_uvchr(c)
2086#define isVERTWS_uni(c) isVERTWS_uvchr(c)
2087#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2088#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2089#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2090#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2091#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2092#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 2093
4650c663
KW
2094/* For internal core Perl use only: the base macros for defining macros like
2095 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 2096 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 2097 * simple class definitions; the other for more complex. These are like
d0da05db 2098 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
2099#define _generic_LC_uvchr(latin1, above_latin1, c) \
2100 (c < 256 ? latin1(c) : above_latin1(c))
2366ba44 2101#define _generic_LC_invlist_uvchr(latin1, classnum, c) \
cd500f2f
KW
2102 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2103
2366ba44
KW
2104#define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2105#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
cd500f2f 2106 _CC_ALPHANUMERIC, c)
b7d90381
KW
2107#define isASCII_LC_uvchr(c) isASCII_LC(c)
2108#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2109 is_HORIZWS_cp_high, c)
feeab5a9 2110#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2366ba44
KW
2111#define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2112#define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 2113#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 2114 _is_uni_perl_idcont, c)
b7d90381 2115#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f 2116 _is_uni_perl_idstart, c)
2366ba44
KW
2117#define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2118#define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 2119#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2366ba44 2120#define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 2121#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 2122 is_XPERLSPACE_cp_high, c)
2366ba44
KW
2123#define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2124#define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
cd500f2f 2125 _CC_WORDCHAR, c)
b7d90381
KW
2126#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2127 is_XDIGIT_cp_high, c)
e712593e 2128
b7d90381 2129#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 2130
da8c1a98
KW
2131/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2132 * they don't otherwise go out of their way to look for malformed UTF-8. If
2133 * they can return accurate results without knowing if the input is otherwise
2134 * malformed, they do so. For example isASCII is accurate in spite of any
2135 * non-length malformations because it looks only at a single byte. Likewise
2136 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2137 * variant ones return FALSE. But, if the input has to be well-formed in order
2138 * for the results to be accurate, the macros will test and if malformed will
2139 * call a routine to die
2140 *
2141 * Except for toke.c, the macros do assume that e > p, asserting that on
2142 * DEBUGGING builds. Much code that calls these depends on this being true,
2143 * for other reasons. toke.c is treated specially as using the regular
2144 * assertion breaks it in many ways. All strings that these operate on there
2145 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2146 * bunch of code in toke.c assumes that this is true, so the assertion allows
2147 * for that */
2148#ifdef PERL_IN_TOKE_C
2149# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2150#else
2151# define _utf8_safe_assert(p,e) ((e) > (p))
2152#endif
2153
2154#define _generic_utf8_safe(classnum, p, e, above_latin1) \
c81b3562
KW
2155 ((! _utf8_safe_assert(p, e)) \
2156 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2157 : (UTF8_IS_INVARIANT(*(p))) \
da8c1a98
KW
2158 ? _generic_isCC(*(p), classnum) \
2159 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2160 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2161 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2162 classnum) \
2163 : (_force_out_malformed_utf8_message( \
2164 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2165 : above_latin1))
b7d90381
KW
2166/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2167 * 'above_latin1' can be a macro */
da8c1a98
KW
2168#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2169 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2366ba44 2170#define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2171 _generic_utf8_safe(classnum, p, e, \
2172 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2173 ? (_force_out_malformed_utf8_message( \
2174 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2175 : above_latin1(p)))
2366ba44
KW
2176/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2177 * 'above_latin1' parameter */
2178#define _generic_invlist_utf8_safe(classnum, p, e) \
2179 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
922e8cb4 2180
cc8ab7c0 2181/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
2182 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2183 * class is TRUE for. Hence it can skip the tests for this range.
2184 * 'above_latin1' should include its arguments */
da8c1a98
KW
2185#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2186 (__ASSERT_(_utf8_safe_assert(p, e)) \
2187 (UTF8_IS_INVARIANT(*(p))) \
2188 ? _generic_isCC(*(p), classnum) \
2189 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2190 ? 0 /* Note that doesn't check validity for latin1 */ \
2191 : above_latin1)
2192
84238efa 2193
059703b0
KW
2194#define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2195#define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2196#define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2197#define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2198#define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2199#define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2200#define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2201#define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2202#define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2203#define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2204#define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2205#define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2206#define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2207#define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2208#define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2209#define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2210#define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2211#define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
e8fa43e2 2212
2366ba44 2213#define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
da8c1a98 2214#define isALPHANUMERIC_utf8_safe(p, e) \
2366ba44 2215 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
da8c1a98
KW
2216#define isASCII_utf8_safe(p, e) \
2217 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2218 * works */ \
2219 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2220#define isBLANK_utf8_safe(p, e) \
2366ba44 2221 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
da8c1a98 2222
e8fa43e2
KW
2223#ifdef EBCDIC
2224 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2225 * more efficient macro instead of the more general one */
da8c1a98 2226# define isCNTRL_utf8_safe(p, e) \
56d02b8c 2227 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 2228#else
da8c1a98 2229# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
2230#endif
2231
da8c1a98
KW
2232#define isDIGIT_utf8_safe(p, e) \
2233 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2366ba44
KW
2234 _is_utf8_FOO(_CC_DIGIT, p, e))
2235#define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
da8c1a98 2236#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
dd1a3ba7 2237 _is_utf8_perl_idcont, p, e)
e5dcd934 2238
c11ff943
KW
2239/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2240 * IDFIRST is an alnum. See
8034715d 2241 * https://github.com/Perl/perl5/issues/10275 for more detail than you
f91dcd13
KW
2242 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2243 * This used to be not the XID version, but we decided to go with the more
2244 * modern Unicode definition */
da8c1a98
KW
2245#define isIDFIRST_utf8_safe(p, e) \
2246 _generic_func_utf8_safe(_CC_IDFIRST, \
dd1a3ba7 2247 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
da8c1a98 2248
2366ba44
KW
2249#define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2250#define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
da8c1a98 2251#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2366ba44 2252#define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
da8c1a98 2253#define isSPACE_utf8_safe(p, e) \
2366ba44
KW
2254 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2255#define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
da8c1a98 2256#define isVERTWS_utf8_safe(p, e) \
2366ba44 2257 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
da8c1a98 2258#define isWORDCHAR_utf8_safe(p, e) \
2366ba44 2259 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
da8c1a98
KW
2260#define isXDIGIT_utf8_safe(p, e) \
2261 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2262 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2263 ? (_force_out_malformed_utf8_message( \
2264 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2265 : is_XDIGIT_high(p)))
a0ed51b3 2266
059703b0
KW
2267#define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2268#define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2269#define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2270#define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2e8adce6 2271
567b353c 2272/* For internal core use only, subject to change */
059703b0
KW
2273#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2274#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2275#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2276#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
a1a5ec35
KW
2277
2278#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2279#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2280#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2281#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 2282
059703b0
KW
2283#define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2284#define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2285#define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2286#define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2287#define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2288#define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2289#define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2290#define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2291#define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2292#define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2293#define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2294#define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2295#define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2296#define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2297#define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2298#define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2299#define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
34aeb2e9 2300
da8c1a98
KW
2301/* For internal core Perl use only: the base macros for defining macros like
2302 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2303 * point in 'p' is within the 0-255 range, it uses locale rules from the
2304 * passed-in 'macro' parameter */
2305#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2306 (__ASSERT_(_utf8_safe_assert(p, e)) \
2307 (UTF8_IS_INVARIANT(*(p))) \
2308 ? macro(*(p)) \
2309 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2310 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2311 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2312 : (_force_out_malformed_utf8_message( \
2313 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2314 : above_latin1))
2315
2366ba44
KW
2316#define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2317 _generic_LC_utf8_safe(macro, p, e, \
2318 _is_utf8_FOO(classnum, p, e))
da8c1a98
KW
2319
2320#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2321 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2322
2366ba44 2323#define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2324 _generic_LC_utf8_safe(classnum, p, e, \
2325 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2326 ? (_force_out_malformed_utf8_message( \
2327 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2328 : above_latin1(p)))
2329
2330#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2366ba44 2331 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
da8c1a98
KW
2332 _CC_ALPHANUMERIC, p, e)
2333#define isALPHA_LC_utf8_safe(p, e) \
2366ba44 2334 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
da8c1a98
KW
2335#define isASCII_LC_utf8_safe(p, e) \
2336 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2337#define isBLANK_LC_utf8_safe(p, e) \
2366ba44 2338 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
da8c1a98
KW
2339#define isCNTRL_LC_utf8_safe(p, e) \
2340 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2341#define isDIGIT_LC_utf8_safe(p, e) \
2366ba44 2342 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
da8c1a98 2343#define isGRAPH_LC_utf8_safe(p, e) \
2366ba44 2344 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
da8c1a98
KW
2345#define isIDCONT_LC_utf8_safe(p, e) \
2346 _generic_LC_func_utf8_safe(isIDCONT_LC, \
dd1a3ba7 2347 _is_utf8_perl_idcont, p, e)
da8c1a98
KW
2348#define isIDFIRST_LC_utf8_safe(p, e) \
2349 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
dd1a3ba7 2350 _is_utf8_perl_idstart, p, e)
da8c1a98 2351#define isLOWER_LC_utf8_safe(p, e) \
2366ba44 2352 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
da8c1a98 2353#define isPRINT_LC_utf8_safe(p, e) \
2366ba44 2354 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
da8c1a98
KW
2355#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2356#define isPUNCT_LC_utf8_safe(p, e) \
2366ba44 2357 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
da8c1a98 2358#define isSPACE_LC_utf8_safe(p, e) \
2366ba44 2359 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
da8c1a98 2360#define isUPPER_LC_utf8_safe(p, e) \
2366ba44 2361 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
da8c1a98 2362#define isWORDCHAR_LC_utf8_safe(p, e) \
2366ba44 2363 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
da8c1a98 2364#define isXDIGIT_LC_utf8_safe(p, e) \
2366ba44 2365 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2366
fbc19f27
KW
2367/* Macros for backwards compatibility and for completeness when the ASCII and
2368 * Latin1 values are identical */
b7d90381
KW
2369#define isALPHAU(c) isALPHA_L1(c)
2370#define isDIGIT_L1(c) isDIGIT_A(c)
2371#define isOCTAL(c) isOCTAL_A(c)
2372#define isOCTAL_L1(c) isOCTAL_A(c)
2373#define isXDIGIT_L1(c) isXDIGIT_A(c)
2374#define isALNUM(c) isWORDCHAR(c)
a377c856 2375#define isALNUM_A(c) isALNUM(c)
b7d90381
KW
2376#define isALNUMU(c) isWORDCHAR_L1(c)
2377#define isALNUM_LC(c) isWORDCHAR_LC(c)
2378#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2379#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
059703b0 2380#define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
4c1d9526 2381#define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
059703b0 2382#define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
4c1d9526 2383#define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
b7d90381
KW
2384#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2385#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2386#define isALNUMC(c) isALPHANUMERIC(c)
2387#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2388#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2389#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
059703b0 2390#define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
4c1d9526
KW
2391#define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2392#define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
fbc19f27 2393
2bd1cbf6
KW
2394/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2395 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2396 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2397 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2398 * the outlier from the block that contains the other controls, just like
2399 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2400 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2401 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2402 * etc. */
2bd1cbf6 2403#ifndef EBCDIC
75763b3a 2404# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2405#else
75763b3a
KW
2406# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2407 ((isPRINT_A(c)) \
2408 ? (UNLIKELY((c) == '?') \
2409 ? QUESTION_MARK_CTRL \
2410 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2411 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2412 ? '?' \
2413 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2414#endif
bbce6d69 2415
dea28490
JJ
2416/* Line numbers are unsigned, 32 bits. */
2417typedef U32 line_t;
e5dcd934 2418#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2419
91152fc1
DG
2420/* Helpful alias for version prescan */
2421#define is_LAX_VERSION(a,b) \
2422 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2423
2424#define is_STRICT_VERSION(a,b) \
2425 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2426
2427#define BADVERSION(a,b,c) \
2428 if (b) { \
2429 *b = c; \
2430 } \
2431 return a;
8c52afec 2432
1ce77b7d
KW
2433/* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2434 * a-f) to its numeric value without using any branches. The input is
2435 * validated only by an assert() in DEBUGGING builds.
2436 *
2437 * It works by right shifting and isolating the bit that is 0 for the digits,
2438 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2439 * position, and then to the eights position. Both are added together to form
2440 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2441 * final four bits of the input to form the correct value. */
2442#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2443 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2444 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2445 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2446
2447/* The argument is a string pointer, which is advanced. */
2448#define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
95a59cab 2449
cb27eebd
KW
2450/* Converts a character known to represent an octal digit (0-7) to its numeric
2451 * value. The input is validated only by an assert() in DEBUGGING builds. In
2452 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2453#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2454
305b8651
KW
2455/* Efficiently returns a boolean as to if two native characters are equivalent
2456 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2457 * the ALPHA in the name is to remind you of that. This is asserted() in
2458 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2459 * works (on valid input) for both non- and UTF-8-encoded bytes.
2460 *
2461 * When one of the inputs is a compile-time constant and gets folded by the
2462 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2463 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2464 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2465 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2466 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2467 * just a single 0, in the bit position where the upper- and lowercase differ.
2468 * */
305b8651
KW
2469#define isALPHA_FOLD_EQ(c1, c2) \
2470 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2471 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2472#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2473
8e84507e 2474/*
70b05c7c 2475=for apidoc_section Memory Management
ccfc67b7 2476
a02a5408 2477=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2478The XSUB-writer's interface to the C C<malloc> function.
2479
596f7718 2480Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2481
c5008215
JC
2482In 5.9.3, Newx() and friends replace the older New() API, and drops
2483the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2484themselves. This aid has been superseded by a new build option,
d10b4965 2485PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2486there for use in XS modules supporting older perls.
2487
a02a5408 2488=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2489The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2490cast. See also C<L</Newx>>.
954c1994 2491
596f7718 2492Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2493
a02a5408 2494=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2495The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2496memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2497
596f7718 2498Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2499
954c1994
GS
2500=for apidoc Am|void|Renew|void* ptr|int nitems|type
2501The XSUB-writer's interface to the C C<realloc> function.
2502
596f7718 2503Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2504
954c1994
GS
2505=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2506The XSUB-writer's interface to the C C<realloc> function, with
2507cast.
2508
596f7718 2509Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2510
49b8b560 2511=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2512The XSUB-writer's interface to the C C<free> function.
2513
596f7718 2514This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2515
51b56f5c 2516=for apidoc_section String Handling
954c1994
GS
2517=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2518The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2519source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2520C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2521
e90e2364 2522=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2523Like C<Move> but returns C<dest>. Useful
72d33970 2524for encouraging compilers to tail-call
e90e2364
NC
2525optimise.
2526
954c1994
GS
2527=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2528The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2529source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2530C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2531
e90e2364
NC
2532=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2533
796b6530 2534Like C<Copy> but returns C<dest>. Useful
72d33970 2535for encouraging compilers to tail-call
e90e2364
NC
2536optimise.
2537
954c1994
GS
2538=for apidoc Am|void|Zero|void* dest|int nitems|type
2539
2540The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2541destination, C<nitems> is the number of items, and C<type> is the type.
2542
e90e2364
NC
2543=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2544
72d33970
FC
2545Like C<Zero> but returns dest. Useful
2546for encouraging compilers to tail-call
e90e2364
NC
2547optimise.
2548
51b56f5c 2549=for apidoc_section Utility Functions
da5d8dbb 2550=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2551This is an architecture-independent macro to copy one structure to another.
954c1994 2552
7e337ee0
JH
2553=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2554
2555Fill up memory with a byte pattern (a byte repeated over and over
2556again) that hopefully catches attempts to access uninitialized memory.
2557
2558=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2559
2560PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2561
1c12ffb4 2562=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2563
2564PoisonWith(0xEF) for catching access to freed memory.
2565
9965345d
JH
2566=for apidoc Am|void|Poison|void* dest|int nitems|type
2567
7e337ee0 2568PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2569
2570=cut */
954c1994 2571
561b68a9
SH
2572/* Maintained for backwards-compatibility only. Use newSV() instead. */
2573#ifndef PERL_CORE
ff06c60c 2574#define NEWSV(x,len) newSV(len)
561b68a9 2575#endif
ff06c60c 2576
b7112dce 2577#define MEM_SIZE_MAX ((MEM_SIZE)-1)
19a94d75 2578
a500027b 2579#define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
e6bdf523 2580
27d5b266 2581#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2582
2583/* This expression will be constant-folded at compile time. It checks
2584 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2585 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2586 * As well as avoiding the need for a run-time check in some cases, it's
2587 * designed to avoid compiler warnings like:
2588 * comparison is always false due to limited range of data type
73e8ff00
DM
2589 * It's mathematically equivalent to
2590 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2591 */
2592
2593# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
445198b9
LM
2594 ( sizeof(MEM_SIZE) < sizeof(n) \
2595 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 2596
88f9f128 2597/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2598 * compiler warnings. We *want* to write the expression as
2599 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2600 * (for some compile-time constant C), but even when the LHS
2601 * constant-folds to false at compile-time, g++ insists on emitting
2602 * warnings about the RHS (e.g. "comparison is always false"), so instead
2603 * we write it as
e6bdf523 2604 *
d98e5cde 2605 * (cond ? n : X) > C
88f9f128 2606 *
d98e5cde
DM
2607 * where X is a constant with X > C always false. Choosing a value for X
2608 * is tricky. If 0, some compilers will complain about 0 > C always being
2609 * false; if 1, Coverity complains when n happens to be the constant value
2610 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2611 * for X and hope that nothing else whines.
e6bdf523
DM
2612 */
2613
2614# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2615 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2616 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2617
2618# define MEM_WRAP_CHECK(n,t) \
2619 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2620 && (croak_memory_wrap(),0))
2621
2622# define MEM_WRAP_CHECK_1(n,t,a) \
2623 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2624 && (Perl_croak_nocontext("%s",(a)),0))
2625
814eedc8
DD
2626/* "a" arg must be a string literal */
2627# define MEM_WRAP_CHECK_s(n,t,a) \
2628 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2629 && (Perl_croak_nocontext("" a ""),0))
2630
8b44ba4c 2631#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2632
a500027b 2633#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
2634#else
2635
410319be
NC
2636#define MEM_WRAP_CHECK(n,t)
2637#define MEM_WRAP_CHECK_1(n,t,a)
814eedc8 2638#define MEM_WRAP_CHECK_s(n,t,a)
8b44ba4c
NC
2639#define MEM_WRAP_CHECK_(n,t)
2640
a500027b 2641#define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
27d5b266 2642
1936d2a7 2643#endif
8b44ba4c 2644
fe4f188c 2645#ifdef PERL_MEM_LOG
46c6c7e2 2646/*
9f653bb5 2647 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2648 * go through functions, which are handy for debugging breakpoints, but
2649 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2650 * line number, and C function name if available) passed in. This info can
2651 * then be used for logging the calls, for which one gets a sample
73d1d973 2652 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2653 *
46c6c7e2 2654 * Known problems:
94e892a6 2655 * - not all memory allocs get logged, only those
46c6c7e2 2656 * that go through Newx() and derivatives (while all
94e892a6 2657 * Safefrees do get logged)
46c6c7e2
JH
2658 * - __FILE__ and __LINE__ do not work everywhere
2659 * - __func__ or __FUNCTION__ even less so
2660 * - I think more goes on after the perlio frees but
2661 * the thing is that STDERR gets closed (as do all
2662 * the file descriptors)
2663 * - no deeper calling stack than the caller of the Newx()
2664 * or the kind, but do I look like a C reflection/introspection
2665 * utility to you?
2666 * - the function prototypes for the logging functions
2667 * probably should maybe be somewhere else than handy.h
2668 * - one could consider inlining (macrofying) the logging
2669 * for speed, but I am too lazy
2670 * - one could imagine recording the allocations in a hash,
2671 * (keyed by the allocation address?), and maintain that
2672 * through reallocs and frees, but how to do that without
2673 * any News() happening...?
73d1d973 2674 * - lots of -Ddefines to get useful/controllable output
b953482e 2675 * - lots of ENV reads
46c6c7e2
JH
2676 */
2677
0b0ab801 2678# ifdef PERL_CORE
73d1d973 2679# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2680enum mem_log_type {
2681 MLT_ALLOC,
2682 MLT_REALLOC,
d7a2c63c
MHM
2683 MLT_FREE,
2684 MLT_NEW_SV,
2685 MLT_DEL_SV
0b0ab801
MHM
2686};
2687# endif
12754f92 2688# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2689void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2690void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2691# endif
0b0ab801
MHM
2692# endif
2693
fe4f188c
JH
2694#endif
2695
2696#ifdef PERL_MEM_LOG
d1401ee9
MHM
2697#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2698#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2699#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2700#endif
2701
2702#ifndef MEM_LOG_ALLOC
2703#define MEM_LOG_ALLOC(n,t,a) (a)
2704#endif
2705#ifndef MEM_LOG_REALLOC
2706#define MEM_LOG_REALLOC(n,t,v,a) (a)
2707#endif
2708#ifndef MEM_LOG_FREE
2709#define MEM_LOG_FREE(a) (a)
2710#endif
2711
d1401ee9
MHM
2712#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2713#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2714#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2715
2716#ifndef PERL_CORE
a02a5408
JC
2717/* pre 5.9.x compatibility */
2718#define New(x,v,n,t) Newx(v,n,t)
2719#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2720#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2721#endif
a02a5408 2722
ff68c719 2723#define Renew(v,n,t) \
d1401ee9 2724 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2725#define Renewc(v,n,t,c) \
d1401ee9 2726 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2727
2728#ifdef PERL_POISON
2729#define Safefree(d) \
06c0cc96 2730 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2731#else
fe4f188c 2732#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2733#endif
55497cff 2734
dbb57106
YO
2735/* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2736 * as it handles cases like constant string arguments without throwing warnings *
2737 * the cast is required, as is the inequality check, to avoid warnings */
45908e4d 2738#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2739
45908e4d
YO
2740
2741#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)))
2742#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)))
2743#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2744
bdd1531d 2745/* Like above, but returns a pointer to 'd' */
45908e4d
YO
2746#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)))
2747#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 2748#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2749
7e337ee0
JH
2750#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2751#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2752#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2753#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2754
caa674f3
DD
2755#ifdef PERL_POISON
2756# define PERL_POISON_EXPR(x) x
2757#else
2758# define PERL_POISON_EXPR(x)
2759#endif
2760
ff68c719 2761#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2762
1b7e2294 2763/*
51b56f5c 2764=for apidoc_section Utility Functions
1b7e2294
KW
2765
2766=for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2767
2768Returns the number of elements in the input C array (so you want your
2769zero-based indices to be less than but not equal to).
2770
2771=for apidoc Am|void *|C_ARRAY_END|void *a
2772
2773Returns a pointer to one element past the final element of the input C array.
2774
2775=cut
2776
2777C_ARRAY_END is one past the last: half-open/half-closed range, not
2778last-inclusive range.
2779*/
622913ab 2780#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2781#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2782
2cc61e15
DD
2783#ifdef NEED_VA_COPY
2784# ifdef va_copy
2785# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2786# elif defined(__va_copy)
2787# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2788# else
07798b17 2789# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2790# endif
2791#endif
2792
472d47bc
SB
2793/* convenience debug macros */
2794#ifdef USE_ITHREADS
2795#define pTHX_FORMAT "Perl interpreter: 0x%p"
2796#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2797#define pTHX_VALUE_ (void *)my_perl,
2798#define pTHX_VALUE (void *)my_perl
2799#define pTHX__VALUE_ ,(void *)my_perl,
2800#define pTHX__VALUE ,(void *)my_perl
472d47bc 2801#else
3609ea0d 2802#define pTHX_FORMAT
472d47bc 2803#define pTHX__FORMAT
3609ea0d 2804#define pTHX_VALUE_
472d47bc 2805#define pTHX_VALUE
3609ea0d 2806#define pTHX__VALUE_
472d47bc
SB
2807#define pTHX__VALUE
2808#endif /* USE_ITHREADS */
3609ea0d 2809
2acdbac1
NC
2810/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2811 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2812 CPAN::Unpack show any users outside the core. */
2813#ifdef PERL_CORE
dc6e8de0
A
2814# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2815 "Use of " s " is deprecated")
c9680906
A
2816# define deprecate_disappears_in(when,message) \
2817 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2818 message ", and will disappear in Perl " when)
ac641426
A
2819# define deprecate_fatal_in(when,message) \
2820 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2821 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2822#endif
2823
dfff4baf
BF
2824/* Internal macros to deal with gids and uids */
2825#ifdef PERL_CORE
2826
2827# if Uid_t_size > IVSIZE
2828# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2829# define SvUID(sv) SvNV(sv)
07798b17
AC
2830# elif Uid_t_sign <= 0
2831# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2832# define SvUID(sv) SvIV(sv)
dfff4baf 2833# else
07798b17
AC
2834# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2835# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2836# endif /* Uid_t_size */
2837
2838# if Gid_t_size > IVSIZE
2839# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2840# define SvGID(sv) SvNV(sv)
07798b17
AC
2841# elif Gid_t_sign <= 0
2842# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2843# define SvGID(sv) SvIV(sv)
dfff4baf 2844# else
07798b17
AC
2845# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2846# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2847# endif /* Gid_t_size */
2848
2849#endif
2850
6a5bc5ac 2851#endif /* PERL_HANDY_H_ */
9d745869 2852
e9a8c099 2853/*
14d04a33 2854 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2855 */