This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
handy.h: Create nBIT_MASK(n) macro
[perl5.git] / handy.h
CommitLineData
a0d0e21e 1/* handy.h
a687059c 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
da5d8dbb 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
a687059c 5 *
6e21c824
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8d063cd8 8 *
8d063cd8
LW
9 */
10
4650c663
KW
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
6a5bc5ac
KW
14#ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */
15#define PERL_HANDY_H_
9d745869 16
24792b8d
NC
17#ifndef PERL_CORE
18# define Null(type) ((type)NULL)
954c1994
GS
19
20/*
ccfc67b7 21=head1 Handy Values
954c1994 22
78342678 23=for apidoc AmnU||Nullch
72d33970
FC
24Null character pointer. (No longer available when C<PERL_CORE> is
25defined.)
2307c6d0 26
78342678 27=for apidoc AmnU||Nullsv
72d33970 28Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
29
30=cut
31*/
32
24792b8d
NC
33# define Nullch Null(char*)
34# define Nullfp Null(PerlIO*)
35# define Nullsv Null(SV*)
36#endif
8d063cd8 37
641d3f0b
PP
38#ifdef TRUE
39#undef TRUE
40#endif
41#ifdef FALSE
42#undef FALSE
43#endif
44#define TRUE (1)
45#define FALSE (0)
46
cf3f0ffb
DM
47/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
48 * (compiler permitting) that casting away const-ness will give a warning;
49 * e.g.:
50 *
51 * const SV *sv = ...;
52 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
53 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
54 */
55
b1bc3f34 56#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
6c2255e0 57# define MUTABLE_PTR(p) ({ void *p_ = (p); p_; })
b1bc3f34
NC
58#else
59# define MUTABLE_PTR(p) ((void *) (p))
60#endif
61
a062e10d 62#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 63#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 64#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 65#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 66#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 67#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 68
f789f6a4 69#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
70# include <stdbool.h>
71# ifndef HAS_BOOL
72# define HAS_BOOL 1
73# endif
74#endif
75
8e84507e 76/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
77 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
78 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 79 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
80 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
81 g++ can be identified by __GNUG__.
82 Andy Dougherty February 2000
5d94fbed 83*/
3609ea0d 84#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 85# ifndef PERL_BOOL_AS_CHAR
5d94fbed 86# ifndef HAS_BOOL
c1d22f6b 87# define HAS_BOOL 1
5d94fbed 88# endif
f789f6a4 89# endif
5d94fbed
AD
90#endif
91
92#ifndef HAS_BOOL
f789f6a4
FC
93# ifdef bool
94# undef bool
95# endif
70d5cb32 96# define bool char
c1d22f6b 97# define HAS_BOOL 1
a687059c 98#endif
0d3e774c 99
25ba28ce
KW
100/*
101=for apidoc Am|bool|cBOOL|bool expr
102
103Cast-to-bool. A simple S<C<(bool) I<expr>>> cast may not do the right thing:
104if C<bool> is defined as C<char>, for example, then the cast from C<int> is
105implementation-defined.
106
107C<(bool)!!(cbool)> in a ternary triggers a bug in xlc on AIX
108
109=cut
110*/
18f5643b 111#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 112
46c6c7e2 113/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
114 * XXX Should really be a Configure probe, with HAS__FUNCTION__
115 * and FUNCTION__ as results.
116 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
46c6c7e2
JH
117#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
118# define FUNCTION__ __func__
7adf2470 119#elif (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
07798b17 120# define FUNCTION__ ""
46c6c7e2 121#else
07798b17 122# define FUNCTION__ __FUNCTION__ /* Common extension. */
46c6c7e2
JH
123#endif
124
27d4fb96
PP
125/* XXX A note on the perl source internal type system. The
126 original intent was that I32 be *exactly* 32 bits.
127
128 Currently, we only guarantee that I32 is *at least* 32 bits.
129 Specifically, if int is 64 bits, then so is I32. (This is the case
130 for the Cray.) This has the advantage of meshing nicely with
131 standard library calls (where we pass an I32 and the library is
132 expecting an int), but the disadvantage that an I32 is not 32 bits.
133 Andy Dougherty August 1996
24fef2a7 134
dc45a647
MB
135 There is no guarantee that there is *any* integral type with
136 exactly 32 bits. It is perfectly legal for a system to have
137 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 138
dc45a647
MB
139 Similarly, there is no guarantee that I16 and U16 have exactly 16
140 bits.
693762b4 141
8e84507e
NIS
142 For dealing with issues that may arise from various 32/64-bit
143 systems, we will ask Configure to check out
8175356b 144
3609ea0d
JH
145 SHORTSIZE == sizeof(short)
146 INTSIZE == sizeof(int)
147 LONGSIZE == sizeof(long)
dc45a647 148 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 149 PTRSIZE == sizeof(void *)
dc45a647
MB
150 DOUBLESIZE == sizeof(double)
151 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 152
27d4fb96
PP
153*/
154
69512466
JH
155#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
156# include <inttypes.h>
dd0eed91
JH
157# ifdef INT32_MIN_BROKEN
158# undef INT32_MIN
159# define INT32_MIN (-2147483647-1)
160# endif
161# ifdef INT64_MIN_BROKEN
162# undef INT64_MIN
163# define INT64_MIN (-9223372036854775807LL-1)
164# endif
69512466
JH
165#endif
166
8175356b
JH
167typedef I8TYPE I8;
168typedef U8TYPE U8;
169typedef I16TYPE I16;
170typedef U16TYPE U16;
171typedef I32TYPE I32;
172typedef U32TYPE U32;
16d89be8 173
74b807c7 174#ifdef QUADKIND
8175356b
JH
175typedef I64TYPE I64;
176typedef U64TYPE U64;
16d89be8 177#endif
8175356b 178
d8668976 179#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 180
5ff3f7a4
GS
181/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
182 Please search CHAR_MAX in perl.h for further details. */
183#define U8_MAX UINT8_MAX
184#define U8_MIN UINT8_MIN
185
5ff3f7a4
GS
186#define I16_MAX INT16_MAX
187#define I16_MIN INT16_MIN
188#define U16_MAX UINT16_MAX
189#define U16_MIN UINT16_MIN
190
5ff3f7a4
GS
191#define I32_MAX INT32_MAX
192#define I32_MIN INT32_MIN
0e983133
GS
193#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
194# define U32_MAX UINT32_MAX
195#else
196# define U32_MAX 4294967295U
197#endif
5ff3f7a4
GS
198#define U32_MIN UINT32_MIN
199
200#else
201
5c9fa16e
KA
202/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
203 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
204#define U8_MAX PERL_UCHAR_MAX
205#define U8_MIN PERL_UCHAR_MIN
79072805 206
27d4fb96
PP
207#define I16_MAX PERL_SHORT_MAX
208#define I16_MIN PERL_SHORT_MIN
209#define U16_MAX PERL_USHORT_MAX
210#define U16_MIN PERL_USHORT_MIN
79072805 211
c4f23d77 212#if LONGSIZE > 4
27d4fb96
PP
213# define I32_MAX PERL_INT_MAX
214# define I32_MIN PERL_INT_MIN
215# define U32_MAX PERL_UINT_MAX
216# define U32_MIN PERL_UINT_MIN
79072805 217#else
27d4fb96
PP
218# define I32_MAX PERL_LONG_MAX
219# define I32_MIN PERL_LONG_MIN
220# define U32_MAX PERL_ULONG_MAX
221# define U32_MIN PERL_ULONG_MIN
79072805
LW
222#endif
223
5ff3f7a4
GS
224#endif
225
247cee9f
KW
226/* These C99 typedefs are useful sometimes for, say, loop variables whose
227 * maximum values are small, but for which speed trumps size. If we have a C99
228 * compiler, use that. Otherwise, a plain 'int' should be good enough.
229 *
230 * Restrict these to core for now until we are more certain this is a good
231 * idea. */
232#if defined(PERL_CORE) || defined(PERL_EXT)
233# ifdef I_STDINT
234 typedef int_fast8_t PERL_INT_FAST8_T;
235 typedef uint_fast8_t PERL_UINT_FAST8_T;
236 typedef int_fast16_t PERL_INT_FAST16_T;
237 typedef uint_fast16_t PERL_UINT_FAST16_T;
238# else
239 typedef int PERL_INT_FAST8_T;
240 typedef unsigned int PERL_UINT_FAST8_T;
241 typedef int PERL_INT_FAST16_T;
242 typedef unsigned int PERL_UINT_FAST16_T;
243# endif
244#endif
245
464decb6
KW
246/* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
247 * anyone is grepping for it */
248#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
fc36a67e
PP
249#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
250#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
251
88794300 252/* Unused by core; should be deprecated */
ff68c719 253#define Ctl(ch) ((ch) & 037)
8d063cd8 254
98fce2a4
KW
255#if defined(PERL_CORE) || defined(PERL_EXT)
256# ifndef MIN
257# define MIN(a,b) ((a) < (b) ? (a) : (b))
258# endif
259# ifndef MAX
260# define MAX(a,b) ((a) > (b) ? (a) : (b))
261# endif
262#endif
263
84ff4fa9
KW
264/* Returns a boolean as to whether the input unsigned number is a power of 2
265 * (2**0, 2**1, etc). In other words if it has just a single bit set.
266 * If not, subtracting 1 would leave the uppermost bit set, so the & would
267 * yield non-zero */
268#if defined(PERL_CORE) || defined(PERL_EXT)
011b1419 269# define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
84ff4fa9
KW
270#endif
271
d223e1ea
KW
272/* Returns a mask with the lowest n bits set */
273#ifdef HAS_LONG_LONG
274# define nBIT_MASK(n) ((1ULL << (n)) - 1)
275#else
276# define nBIT_MASK(n) ((1UL << (n)) - 1)
277#endif
278
8d9433eb
KW
279/*
280=for apidoc Am|void|__ASSERT_|bool expr
281
282This is a helper macro to avoid preprocessor issues, replaced by nothing
283unless under DEBUGGING, where it expands to an assert of its argument,
284followed by a comma (hence the comma operator). If we just used a straight
285assert(), we would get a comma with nothing before it when not DEBUGGING.
286
287=cut
288
289We also use empty definition under Coverity since the __ASSERT__
290checks often check for things that Really Cannot Happen, and Coverity
291detects that and gets all excited. */
3e94db23 292
e7ae132e
KW
293#if defined(DEBUGGING) && !defined(__COVERITY__) \
294 && ! defined(PERL_SMALL_MACRO_BUFFER)
0f092d08
KW
295# define __ASSERT_(statement) assert(statement),
296#else
297# define __ASSERT_(statement)
298#endif
299
3fe05580 300/*
a4ee4fb5 301=head1 SV Manipulation Functions
3fe05580 302
3bb9fd01 303=for apidoc Ama|SV*|newSVpvs|"literal string"
1568d13a 304Like C<newSVpvn>, but takes a literal string instead of a
30a15352 305string/length pair.
3fe05580 306
3bb9fd01 307=for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
1568d13a 308Like C<newSVpvn_flags>, but takes a literal string instead of
30a15352 309a string/length pair.
84bafc02 310
3bb9fd01 311=for apidoc Ama|SV*|newSVpvs_share|"literal string"
1568d13a 312Like C<newSVpvn_share>, but takes a literal string instead of
30a15352 313a string/length pair and omits the hash parameter.
3fe05580 314
3bb9fd01 315=for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
1568d13a 316Like C<sv_catpvn_flags>, but takes a literal string instead
30a15352 317of a string/length pair.
9dcc53ea 318
3bb9fd01 319=for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
1568d13a 320Like C<sv_catpvn_nomg>, but takes a literal string instead of
0c395ea5 321a string/length pair.
9dcc53ea 322
3bb9fd01 323=for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
1568d13a 324Like C<sv_catpvn>, but takes a literal string instead of a
0c395ea5 325string/length pair.
3fe05580 326
3bb9fd01 327=for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
1568d13a 328Like C<sv_catpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
329string/length pair.
330
3bb9fd01 331=for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
1568d13a 332Like C<sv_setpvn>, but takes a literal string instead of a
0c395ea5 333string/length pair.
3fe05580 334
3bb9fd01 335=for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
1568d13a 336Like C<sv_setpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
337string/length pair.
338
3bb9fd01 339=for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
1568d13a 340Like C<sv_setref_pvn>, but takes a literal string instead of
0c395ea5 341a string/length pair.
9dcc53ea 342
3fe05580
MHM
343=head1 Memory Management
344
3bb9fd01 345=for apidoc Ama|char*|savepvs|"literal string"
1568d13a 346Like C<savepvn>, but takes a literal string instead of a
30a15352 347string/length pair.
3fe05580 348
3bb9fd01 349=for apidoc Ama|char*|savesharedpvs|"literal string"
9dcc53ea
Z
350A version of C<savepvs()> which allocates the duplicate string in memory
351which is shared between threads.
352
3fe05580
MHM
353=head1 GV Functions
354
3bb9fd01 355=for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
1568d13a 356Like C<gv_stashpvn>, but takes a literal string instead of a
0c395ea5 357string/length pair.
3fe05580
MHM
358
359=head1 Hash Manipulation Functions
360
3bb9fd01 361=for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
1568d13a 362Like C<hv_fetch>, but takes a literal string instead of a
0c395ea5 363string/length pair.
3fe05580 364
3bb9fd01 365=for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
1568d13a 366Like C<hv_store>, but takes a literal string instead of a
0c395ea5 367string/length pair
3fe05580
MHM
368and omits the hash parameter.
369
510966aa
Z
370=head1 Lexer interface
371
3bb9fd01 372=for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
510966aa 373
1568d13a 374Like L</lex_stuff_pvn>, but takes a literal string instead of
0c395ea5 375a string/length pair.
510966aa 376
3fe05580
MHM
377=cut
378*/
379
a34e53fc
KW
380/*
381=head1 Handy Values
2efa8cc7 382
a34e53fc
KW
383=for apidoc Amu|pair|STR_WITH_LEN|"literal string"
384
385Returns two comma separated tokens of the input literal string, and its length.
386This is convenience macro which helps out in some API calls.
387Note that it can't be used as an argument to macros or functions that under
388some configurations might be macros, which means that it requires the full
389Perl_xxx(aTHX_ ...) form for any API calls where it's used.
390
391=cut
392*/
393
a34e53fc 394#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
ba3a79e7
GA
395
396/* STR_WITH_LEN() shortcuts */
397#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
398#define newSVpvs_flags(str,flags) \
399 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 400#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
401#define sv_catpvs_flags(sv, str, flags) \
402 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
403#define sv_catpvs_nomg(sv, str) \
404 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
405#define sv_catpvs(sv, str) \
406 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
407#define sv_catpvs_mg(sv, str) \
408 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 409#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
410#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
411#define sv_setref_pvs(rv, classname, str) \
412 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 413#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
414#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
415#define gv_stashpvs(str, create) \
416 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
417#define gv_fetchpvs(namebeg, add, sv_type) \
418 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
419#define gv_fetchpvn(namebeg, len, add, sv_type) \
420 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
421#define sv_catxmlpvs(dsv, str, utf8) \
422 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 423
ba3a79e7 424
510966aa
Z
425#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
426
b96d8cd9
NC
427#define get_cvs(str, flags) \
428 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 429
9b6e9510
KW
430/* internal helpers */
431#define PERL_RVS_TO_DECIMAL_(r,v,s) (((r)*1000000)+((v)*1000)+(s))
432#define PERL_DECIMAL_VERSION_ \
433 PERL_RVS_TO_DECIMAL_(PERL_REVISION, PERL_VERSION, PERL_SUBVERSION)
434
435/*
436=for apidoc AmR|bool|PERL_VERSION_EQ|const int r|const int v|const int s
437
438Returns whether or not the perl currently executing has the specified
439relationship to the perl given by the parameters. For example,
440
441 #if PERL_VERSION_GT(5,24,2)
442 code that will only be compiled on perls after v5.24.2
443 #else
444 fallback code
445 #endif
446
447Note that this is usable in making compile-time decisions
448
449The possible comparisons are C<PERL_VERSION_EQ>, C<PERL_VERSION_NE>,
450C<PERL_VERSION_GE>, C<PERL_VERSION_GT>, C<PERL_VERSION_LE>, and
451C<PERL_VERSION_LT>.
452
453=for apidoc AmRh|bool|PERL_VERSION_NE|const int r|const int v|const int s
454=for apidoc AmRh|bool|PERL_VERSION_GE|const int r|const int v|const int s
455=for apidoc AmRh|bool|PERL_VERSION_GT|const int r|const int v|const int s
456=for apidoc AmRh|bool|PERL_VERSION_LE|const int r|const int v|const int s
457=for apidoc AmRh|bool|PERL_VERSION_LT|const int r|const int v|const int s
458
459=cut
460*/
461
462# define PERL_VERSION_EQ(r,v,s) (PERL_DECIMAL_VERSION_ == PERL_RVS_TO_DECIMAL_(r,v,s))
463# define PERL_VERSION_NE(r,v,s) (! PERL_VERSION_EQ(r,v,s))
464# define PERL_VERSION_LT(r,v,s) (PERL_DECIMAL_VERSION_ < PERL_RVS_TO_DECIMAL_(r,v,s))
465# define PERL_VERSION_LE(r,v,s) (PERL_DECIMAL_VERSION_ <= PERL_RVS_TO_DECIMAL_(r,v,s))
466# define PERL_VERSION_GT(r,v,s) (! PERL_VERSION_LE(r,v,s))
467# define PERL_VERSION_GE(r,v,s) (! PERL_VERSION_LT(r,v,s))
468
954c1994 469/*
ccfc67b7
JH
470=head1 Miscellaneous Functions
471
954c1994 472=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
473Test two C<NUL>-terminated strings to see if they are different. Returns true
474or false.
954c1994
GS
475
476=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
477Test two C<NUL>-terminated strings to see if they are equal. Returns true or
478false.
954c1994
GS
479
480=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
481Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
482second, C<s2>. Returns true or false.
954c1994
GS
483
484=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
485Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
486equal to the second, C<s2>. Returns true or false.
954c1994
GS
487
488=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
489Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
490the second, C<s2>. Returns true or false.
954c1994
GS
491
492=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
493Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
494or equal to the second, C<s2>. Returns true or false.
954c1994
GS
495
496=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
497Test two C<NUL>-terminated strings to see if they are different. The C<len>
498parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
499wrapper for C<strncmp>).
500
501=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
502Test two C<NUL>-terminated strings to see if they are equal. The C<len>
503parameter indicates the number of bytes to compare. Returns true or false. (A
504wrapper for C<strncmp>).
954c1994 505
bd18bd40
KW
506=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
507Test two buffers (which may contain embedded C<NUL> characters, to see if they
508are equal. The C<len> parameter indicates the number of bytes to compare.
509Returns zero if equal, or non-zero if non-equal.
510
3bb9fd01 511=for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
512Like L</memEQ>, but the second string is a literal enclosed in double quotes,
513C<l1> gives the number of bytes in C<s1>.
514Returns zero if equal, or non-zero if non-equal.
515
bd18bd40
KW
516=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
517Test two buffers (which may contain embedded C<NUL> characters, to see if they
518are not equal. The C<len> parameter indicates the number of bytes to compare.
519Returns zero if non-equal, or non-zero if equal.
520
3bb9fd01 521=for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
522Like L</memNE>, but the second string is a literal enclosed in double quotes,
523C<l1> gives the number of bytes in C<s1>.
524Returns zero if non-equal, or zero if non-equal.
525
4aada8b9
KW
526=for apidoc Am|bool|memCHRs|"list"|char c
527Returns the position of the first occurence of the byte C<c> in the literal
528string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
529treated as unsigned char. Thus this macro can be used to determine if C<c> is
530in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
531is C<NUL> (and the set doesn't include C<NUL>).
532
954c1994 533=cut
fc169e00
KW
534
535New macros should use the following conventions for their names (which are
536based on the underlying C library functions):
537
538 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
539
540 Each has two main parameters, string-like operands that are compared
541 against each other, as specified by the macro name. Some macros may
542 additionally have one or potentially even two length parameters. If a length
543 parameter applies to both string parameters, it will be positioned third;
544 otherwise any length parameter immediately follows the string parameter it
545 applies to.
546
547 If the prefix to the name is 'str', the string parameter is a pointer to a C
548 language string. Such a string does not contain embedded NUL bytes; its
549 length may be unknown, but can be calculated by C<strlen()>, since it is
550 terminated by a NUL, which isn't included in its length.
551
a3815e44 552 The optional 'n' following 'str' means that there is a third parameter,
fc169e00
KW
553 giving the maximum number of bytes to look at in each string. Even if both
554 strings are longer than the length parameter, those extra bytes will be
555 unexamined.
556
557 The 's' suffix means that the 2nd byte string parameter is a literal C
558 double-quoted string. Its length will automatically be calculated by the
559 macro, so no length parameter will ever be needed for it.
560
561 If the prefix is 'mem', the string parameters don't have to be C strings;
562 they may contain embedded NUL bytes, do not necessarily have a terminating
563 NUL, and their lengths can be known only through other means, which in
564 practice are additional parameter(s) passed to the function. All 'mem'
565 functions have at least one length parameter. Barring any 'l' or 's' suffix,
566 there is a single length parameter, in position 3, which applies to both
567 string parameters. The 's' suffix means, as described above, that the 2nd
568 string is a literal double-quoted C string (hence its length is calculated by
569 the macro, and the length parameter to the function applies just to the first
570 string parameter, and hence is positioned just after it). An 'l' suffix
571 means that the 2nd string parameter has its own length parameter, and the
572 signature will look like memFOOl(s1, l1, s2, l2).
573
574 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
575 substring of the 1st string. 'P' if present indicates that the substring
576 must be a "proper" one in tha mathematical sense that the first one must be
577 strictly larger than the 2nd.
578
954c1994
GS
579*/
580
62946e08 581
75400963
KW
582#define strNE(s1,s2) (strcmp(s1,s2) != 0)
583#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
584#define strLT(s1,s2) (strcmp(s1,s2) < 0)
585#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
586#define strGT(s1,s2) (strcmp(s1,s2) > 0)
587#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 588
75400963
KW
589#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
590#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 591
9d3980bc
KW
592#define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
593#define memNE(s1,s2,l) (! memEQ(s1,s2,l))
36477c24 594
085b7534 595/* memEQ and memNE where second comparand is a string constant */
568a785a 596#define memEQs(s1, l, s2) \
777fa2cb 597 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 598#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 599
fdbb9a7c
KW
600/* Keep these private until we decide it was a good idea */
601#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
602
603#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
604
bdb7e3f0 605#define memBEGINs(s1, l, s2) \
30a6480c 606 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 607 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158 608#define memBEGINPs(s1, l, s2) \
30a6480c 609 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
de627158 610 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0 611#define memENDs(s1, l, s2) \
30a6480c 612 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 613 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424 614#define memENDPs(s1, l, s2) \
30a6480c 615 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
b80f8424 616 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 617#endif /* End of making macros private */
bdb7e3f0 618
062b6850
KW
619#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
620#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
621#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
622#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
623
4aada8b9
KW
624#define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
625
bbce6d69
PP
626/*
627 * Character classes.
628 *
629 * Unfortunately, the introduction of locales means that we
630 * can't trust isupper(), etc. to tell the truth. And when
631 * it comes to /\w+/ with tainting enabled, we *must* be able
632 * to trust our character classes.
633 *
634 * Therefore, the default tests in the text of Perl will be
635 * independent of locale. Any code that wants to depend on
636 * the current locale will use the tests that begin with "lc".
637 */
638
2304df62
AD
639#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
640# ifndef CTYPE256
641# define CTYPE256
642# endif
643#endif
644
954c1994 645/*
ccfc67b7 646
dcccc8ff 647=head1 Character classification
243effed
KW
648This section is about functions (really macros) that classify characters
649into types, such as punctuation versus alphabetic, etc. Most of these are
650analogous to regular expression character classes. (See
651L<perlrecharclass/POSIX Character Classes>.) There are several variants for
652each class. (Not all macros have all variants; each item below lists the
653ones valid for it.) None are affected by C<use bytes>, and only the ones
654with C<LC> in the name are affected by the current locale.
655
d713f9d9
KW
656The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
657treating it as a code point, and returns a boolean as to whether or not the
658character represented by it is (or on non-ASCII platforms, corresponds to) an
6aff1f14
KW
659ASCII character in the named class based on platform, Unicode, and Perl rules.
660If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 661
c98722a4 662Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
663with no suffix C<"_A">. This variant is used to emphasize by its name that
664only ASCII-range characters can return TRUE.
4b9734bf 665
d60679e1 666Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
667onto the platform. That is, the code points that are ASCII are unaffected,
668since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
669as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
670true when called with the code point 0xDF, which is a word character in both
4650c663 671ASCII and EBCDIC (though it represents different characters in each).
d713f9d9
KW
672If the input is a number that doesn't fit in an octet, FALSE is returned.
673(Perl's documentation uses a colloquial definition of Latin-1, to include all
674code points below 256.)
243effed 675
d713f9d9
KW
676Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
677inputs below 256, but if the code point is larger than 255, Unicode rules are
678used to determine if it is in the character class. For example,
d0da05db 679C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 680WITH MACRON in Unicode, and is a word character.
243effed 681
059703b0
KW
682Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
683but are used for UTF-8 encoded strings. The two forms are different names for
684the same thing. Each call to one of these classifies the first character of
685the string starting at C<p>. The second parameter, C<e>, points to anywhere in
686the string beyond the first character, up to one byte past the end of the
687entire string. Although both variants are identical, the suffix C<_safe> in
688one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
689provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
690C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
691some way, the program may croak, or the function may return FALSE, at the
692discretion of the implementation, and subject to change in future releases.
243effed 693
d713f9d9
KW
694Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
695but the result is based on the current locale, which is what C<LC> in the name
696stands for. If Perl can determine that the current locale is a UTF-8 locale,
697it uses the published Unicode rules; otherwise, it uses the C library function
698that gives the named classification. For example, C<isDIGIT_LC()> when not in
699a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
700returned if the input won't fit into an octet. On some platforms where the C
701library function is known to be defective, Perl changes its result to follow
702the POSIX standard's rules.
243effed 703
d713f9d9
KW
704Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
705than 256, but for larger ones it returns the Unicode classification of the code
706point.
243effed 707
059703b0
KW
708Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
709C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
710are different names for the same thing. Each call to one of these classifies
711the first character of the string starting at C<p>. The second parameter,
712C<e>, points to anywhere in the string beyond the first character, up to one
713byte past the end of the entire string. Although both variants are identical,
714the suffix C<_safe> in one name emphasizes that it will not attempt to read
715beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
716is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
717character is malformed in some way, the program may croak, or the function may
718return FALSE, at the discretion of the implementation, and subject to change in
719future releases.
ccfc67b7 720
d713f9d9
KW
721=for apidoc Am|bool|isALPHA|int ch
722Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
723analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
724See the L<top of this section|/Character classification> for an explanation of
725variants
059703b0
KW
726C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>,
727C<isALPHA_utf8_safe>, C<isALPHA_LC>, C<isALPHA_LC_uvchr>, C<isALPHA_LC_utf8>,
728and C<isALPHA_LC_utf8_safe>.
8a58bdcf 729
f16858ed
KW
730=cut
731
732Here and below, we add the protoypes of these macros for downstream programs
733that would be interested in them, such as Devel::PPPort
734
735=for apidoc Amh|bool|isALPHA_A|int ch
736=for apidoc Amh|bool|isALPHA_L1|int ch
737=for apidoc Amh|bool|isALPHA_uvchr|int ch
738=for apidoc Amh|bool|isALPHA_utf8_safe|U8 * s|U8 * end
d23c7e08 739=for apidoc Amh|bool|isALPHA_utf8|U8 * s|U8 * end
f16858ed
KW
740=for apidoc Amh|bool|isALPHA_LC|int ch
741=for apidoc Amh|bool|isALPHA_LC_uvchr|int ch
742=for apidoc Amh|bool|isALPHA_LC_utf8_safe|U8 * s| U8 *end
743
d713f9d9
KW
744=for apidoc Am|bool|isALPHANUMERIC|int ch
745Returns a boolean indicating whether the specified character is one of
746C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
747See the L<top of this section|/Character classification> for an explanation of
748variants
d0da05db 749C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
059703b0
KW
750C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>,
751C<isALPHANUMERIC_LC_uvchr>, C<isALPHANUMERIC_LC_utf8>, and
752C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 753
255b632a
KW
754A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
755this corresponds to the C language alphanumeric definition). Also
756there are the variants
757C<isALNUMC_A>, C<isALNUMC_L1>
758C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
759
f16858ed
KW
760=for apidoc Amh|bool|isALPHANUMERIC_A|int ch
761=for apidoc Amh|bool|isALPHANUMERIC_L1|int ch
762=for apidoc Amh|bool|isALPHANUMERIC_uvchr|int ch
763=for apidoc Amh|bool|isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
d23c7e08 764=for apidoc Amh|bool|isALPHANUMERIC_utf8|U8 * s|U8 * end
f16858ed
KW
765=for apidoc Amh|bool|isALPHANUMERIC_LC|int ch
766=for apidoc Amh|bool|isALPHANUMERIC_LC_uvchr|int ch
767=for apidoc Amh|bool|isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
768=for apidoc Amh|bool|isALNUMC|int ch
769=for apidoc Amh|bool|isALNUMC_A|int ch
770=for apidoc Amh|bool|isALNUMC_L1|int ch
771=for apidoc Amh|bool|isALNUMC_LC|int ch
772=for apidoc Amh|bool|isALNUMC_LC_uvchr|int ch
773
d713f9d9 774=for apidoc Am|bool|isASCII|int ch
8a58bdcf 775Returns a boolean indicating whether the specified character is one of the 128
243effed 776characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 777On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
778character corresponds to an ASCII character. Variants C<isASCII_A()> and
779C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
780See the L<top of this section|/Character classification> for an explanation of
781variants
059703b0
KW
782C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_utf8_safe>, C<isASCII_LC>,
783C<isASCII_LC_uvchr>, C<isASCII_LC_utf8>, and C<isASCII_LC_utf8_safe>.
784Note, however, that some platforms do not have the C library routine
785C<isascii()>. In these cases, the variants whose names contain C<LC> are the
786same as the corresponding ones without.
243effed 787
f16858ed
KW
788=for apidoc Amh|bool|isASCII_A|int ch
789=for apidoc Amh|bool|isASCII_L1|int ch
790=for apidoc Amh|bool|isASCII_uvchr|int ch
791=for apidoc Amh|bool|isASCII_utf8_safe|U8 * s|U8 * end
d23c7e08 792=for apidoc Amh|bool|isASCII_utf8|U8 * s|U8 * end
f16858ed
KW
793=for apidoc Amh|bool|isASCII_LC|int ch
794=for apidoc Amh|bool|isASCII_LC_uvchr|int ch
795=for apidoc Amh|bool|isASCII_LC_utf8_safe|U8 * s| U8 *end
796
d98532ea
KW
797Also note, that because all ASCII characters are UTF-8 invariant (meaning they
798have the exact same representation (always a single byte) whether encoded in
799UTF-8 or not), C<isASCII> will give the correct results when called with any
059703b0
KW
800byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
801C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
d98532ea 802
243effed
KW
803=for apidoc Am|bool|isBLANK|char ch
804Returns a boolean indicating whether the specified character is a
6aff1f14 805character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
806See the L<top of this section|/Character classification> for an explanation of
807variants
059703b0
KW
808C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>,
809C<isBLANK_utf8_safe>, C<isBLANK_LC>, C<isBLANK_LC_uvchr>, C<isBLANK_LC_utf8>,
810and C<isBLANK_LC_utf8_safe>. Note,
da8c1a98
KW
811however, that some platforms do not have the C library routine
812C<isblank()>. In these cases, the variants whose names contain C<LC> are
813the same as the corresponding ones without.
243effed 814
f16858ed
KW
815=for apidoc Amh|bool|isBLANK_A|int ch
816=for apidoc Amh|bool|isBLANK_L1|int ch
817=for apidoc Amh|bool|isBLANK_uvchr|int ch
818=for apidoc Amh|bool|isBLANK_utf8_safe|U8 * s|U8 * end
d23c7e08 819=for apidoc Amh|bool|isBLANK_utf8|U8 * s|U8 * end
f16858ed
KW
820=for apidoc Amh|bool|isBLANK_LC|int ch
821=for apidoc Amh|bool|isBLANK_LC_uvchr|int ch
822=for apidoc Amh|bool|isBLANK_LC_utf8_safe|U8 * s| U8 *end
823
243effed
KW
824=for apidoc Am|bool|isCNTRL|char ch
825Returns a boolean indicating whether the specified character is a
6aff1f14 826control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
827See the L<top of this section|/Character classification> for an explanation of
828variants
059703b0
KW
829C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>,
830C<isCNTRL_utf8_safe>, C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, C<isCNTRL_LC_utf8>
831and C<isCNTRL_LC_utf8_safe>. On EBCDIC
da8c1a98 832platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 833
f16858ed
KW
834=for apidoc Amh|bool|isCNTRL_A|int ch
835=for apidoc Amh|bool|isCNTRL_L1|int ch
836=for apidoc Amh|bool|isCNTRL_uvchr|int ch
837=for apidoc Amh|bool|isCNTRL_utf8_safe|U8 * s|U8 * end
d23c7e08 838=for apidoc Amh|bool|isCNTRL_utf8|U8 * s|U8 * end
f16858ed
KW
839=for apidoc Amh|bool|isCNTRL_LC|int ch
840=for apidoc Amh|bool|isCNTRL_LC_uvchr|int ch
841=for apidoc Amh|bool|isCNTRL_LC_utf8_safe|U8 * s| U8 *end
842
954c1994 843=for apidoc Am|bool|isDIGIT|char ch
2787a470 844Returns a boolean indicating whether the specified character is a
6aff1f14 845digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 846Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
847See the L<top of this section|/Character classification> for an explanation of
848variants
059703b0
KW
849C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>,
850C<isDIGIT_LC_uvchr>, C<isDIGIT_LC_utf8>, and C<isDIGIT_LC_utf8_safe>.
243effed 851
f16858ed
KW
852=for apidoc Amh|bool|isDIGIT_A|int ch
853=for apidoc Amh|bool|isDIGIT_L1|int ch
854=for apidoc Amh|bool|isDIGIT_uvchr|int ch
855=for apidoc Amh|bool|isDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 856=for apidoc Amh|bool|isDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
857=for apidoc Amh|bool|isDIGIT_LC|int ch
858=for apidoc Amh|bool|isDIGIT_LC_uvchr|int ch
859=for apidoc Amh|bool|isDIGIT_LC_utf8_safe|U8 * s| U8 *end
860
243effed
KW
861=for apidoc Am|bool|isGRAPH|char ch
862Returns a boolean indicating whether the specified character is a
6aff1f14 863graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 864See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
865variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>,
866C<isGRAPH_utf8_safe>, C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>,
867C<isGRAPH_LC_utf8_safe>, and C<isGRAPH_LC_utf8_safe>.
954c1994 868
f16858ed
KW
869=for apidoc Amh|bool|isGRAPH_A|int ch
870=for apidoc Amh|bool|isGRAPH_L1|int ch
871=for apidoc Amh|bool|isGRAPH_uvchr|int ch
872=for apidoc Amh|bool|isGRAPH_utf8_safe|U8 * s|U8 * end
d23c7e08 873=for apidoc Amh|bool|isGRAPH_utf8|U8 * s|U8 * end
f16858ed
KW
874=for apidoc Amh|bool|isGRAPH_LC|int ch
875=for apidoc Amh|bool|isGRAPH_LC_uvchr|int ch
876=for apidoc Amh|bool|isGRAPH_LC_utf8_safe|U8 * s| U8 *end
877
0c82b6df 878=for apidoc Am|bool|isLOWER|char ch
2787a470 879Returns a boolean indicating whether the specified character is a
6aff1f14 880lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
881See the L<top of this section|/Character classification> for an explanation of
882variants
059703b0
KW
883C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>,
884C<isLOWER_utf8_safe>, C<isLOWER_LC>, C<isLOWER_LC_uvchr>, C<isLOWER_LC_utf8>,
885and C<isLOWER_LC_utf8_safe>.
0c82b6df 886
f16858ed
KW
887=for apidoc Amh|bool|isLOWER_A|int ch
888=for apidoc Amh|bool|isLOWER_L1|int ch
889=for apidoc Amh|bool|isLOWER_uvchr|int ch
890=for apidoc Amh|bool|isLOWER_utf8_safe|U8 * s|U8 * end
d23c7e08 891=for apidoc Amh|bool|isLOWER_utf8|U8 * s|U8 * end
f16858ed
KW
892=for apidoc Amh|bool|isLOWER_LC|int ch
893=for apidoc Amh|bool|isLOWER_LC_uvchr|int ch
894=for apidoc Amh|bool|isLOWER_LC_utf8_safe|U8 * s| U8 *end
895
c99e91e9 896=for apidoc Am|bool|isOCTAL|char ch
2787a470 897Returns a boolean indicating whether the specified character is an
6aff1f14 898octal digit, [0-7].
243effed
KW
899The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
900C<isOCTAL>.
901
f16858ed
KW
902=for apidoc Amh|bool|isOCTAL_A|int ch
903=for apidoc Amh|bool|isOCTAL_L1|int ch
904
243effed
KW
905=for apidoc Am|bool|isPUNCT|char ch
906Returns a boolean indicating whether the specified character is a
6aff1f14
KW
907punctuation character, analogous to C<m/[[:punct:]]/>.
908Note that the definition of what is punctuation isn't as
243effed
KW
909straightforward as one might desire. See L<perlrecharclass/POSIX Character
910Classes> for details.
dcccc8ff 911See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
912variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>,
913C<isPUNCT_utf8_safe>, C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, C<isPUNCT_LC_utf8>,
914and C<isPUNCT_LC_utf8_safe>.
c99e91e9 915
f16858ed
KW
916=for apidoc Amh|bool|isPUNCT_A|int ch
917=for apidoc Amh|bool|isPUNCT_L1|int ch
918=for apidoc Amh|bool|isPUNCT_uvchr|int ch
919=for apidoc Amh|bool|isPUNCT_utf8_safe|U8 * s|U8 * end
d23c7e08 920=for apidoc Amh|bool|isPUNCT_utf8|U8 * s|U8 * end
f16858ed
KW
921=for apidoc Amh|bool|isPUNCT_LC|int ch
922=for apidoc Amh|bool|isPUNCT_LC_uvchr|int ch
923=for apidoc Amh|bool|isPUNCT_LC_utf8_safe|U8 * s| U8 *end
924
0c82b6df 925=for apidoc Am|bool|isSPACE|char ch
2787a470 926Returns a boolean indicating whether the specified character is a
6aff1f14 927whitespace character. This is analogous
398d098a 928to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 929this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
930locale forms of this macro (the ones with C<LC> in their names) matched
931precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
932in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
933(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
934See the L<top of this section|/Character classification> for an explanation of
935variants
059703b0
KW
936C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>,
937C<isSPACE_utf8_safe>, C<isSPACE_LC>, C<isSPACE_LC_uvchr>, C<isSPACE_LC_utf8>,
938and C<isSPACE_LC_utf8_safe>.
0c82b6df 939
f16858ed
KW
940=for apidoc Amh|bool|isSPACE_A|int ch
941=for apidoc Amh|bool|isSPACE_L1|int ch
942=for apidoc Amh|bool|isSPACE_uvchr|int ch
943=for apidoc Amh|bool|isSPACE_utf8_safe|U8 * s|U8 * end
d23c7e08 944=for apidoc Amh|bool|isSPACE_utf8|U8 * s|U8 * end
f16858ed
KW
945=for apidoc Amh|bool|isSPACE_LC|int ch
946=for apidoc Amh|bool|isSPACE_LC_uvchr|int ch
947=for apidoc Amh|bool|isSPACE_LC_utf8_safe|U8 * s| U8 *end
948
398d098a
KW
949=for apidoc Am|bool|isPSXSPC|char ch
950(short for Posix Space)
779cf272
KW
951Starting in 5.18, this is identical in all its forms to the
952corresponding C<isSPACE()> macros.
398d098a
KW
953The locale forms of this macro are identical to their corresponding
954C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
955non-locale forms differ from their C<isSPACE()> forms only in that the
956C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
957Otherwise they are identical. Thus this macro is analogous to what
958C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 959See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
960variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
961C<isPSXSPC_utf8_safe>, C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>,
962C<isPSXSPC_LC_utf8>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 963
f16858ed
KW
964=for apidoc Amh|bool|isPSXSPC_A|int ch
965=for apidoc Amh|bool|isPSXSPC_L1|int ch
966=for apidoc Amh|bool|isPSXSPC_uvchr|int ch
967=for apidoc Amh|bool|isPSXSPC_utf8_safe|U8 * s|U8 * end
d23c7e08 968=for apidoc Amh|bool|isPSXSPC_utf8|U8 * s|U8 * end
f16858ed
KW
969=for apidoc Amh|bool|isPSXSPC_LC|int ch
970=for apidoc Amh|bool|isPSXSPC_LC_uvchr|int ch
971=for apidoc Amh|bool|isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
972
954c1994 973=for apidoc Am|bool|isUPPER|char ch
2787a470 974Returns a boolean indicating whether the specified character is an
6aff1f14 975uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 976See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
977variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>,
978C<isUPPER_utf8_safe>, C<isUPPER_LC>, C<isUPPER_LC_uvchr>, C<isUPPER_LC_utf8>,
979and C<isUPPER_LC_utf8_safe>.
954c1994 980
f16858ed
KW
981=for apidoc Amh|bool|isUPPER_A|int ch
982=for apidoc Amh|bool|isUPPER_L1|int ch
983=for apidoc Amh|bool|isUPPER_uvchr|int ch
984=for apidoc Amh|bool|isUPPER_utf8_safe|U8 * s|U8 * end
d23c7e08 985=for apidoc Amh|bool|isUPPER_utf8|U8 * s|U8 * end
f16858ed
KW
986=for apidoc Amh|bool|isUPPER_LC|int ch
987=for apidoc Amh|bool|isUPPER_LC_uvchr|int ch
988=for apidoc Amh|bool|isUPPER_LC_utf8_safe|U8 * s| U8 *end
989
243effed 990=for apidoc Am|bool|isPRINT|char ch
8eea39dd 991Returns a boolean indicating whether the specified character is a
6aff1f14 992printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
993See the L<top of this section|/Character classification> for an explanation of
994variants
059703b0
KW
995C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>,
996C<isPRINT_utf8_safe>, C<isPRINT_LC>, C<isPRINT_LC_uvchr>, C<isPRINT_LC_utf8>,
997and C<isPRINT_LC_utf8_safe>.
243effed 998
f16858ed
KW
999=for apidoc Amh|bool|isPRINT_A|int ch
1000=for apidoc Amh|bool|isPRINT_L1|int ch
1001=for apidoc Amh|bool|isPRINT_uvchr|int ch
1002=for apidoc Amh|bool|isPRINT_utf8_safe|U8 * s|U8 * end
d23c7e08 1003=for apidoc Amh|bool|isPRINT_utf8|U8 * s|U8 * end
f16858ed
KW
1004=for apidoc Amh|bool|isPRINT_LC|int ch
1005=for apidoc Amh|bool|isPRINT_LC_uvchr|int ch
1006=for apidoc Amh|bool|isPRINT_LC_utf8_safe|U8 * s| U8 *end
1007
243effed
KW
1008=for apidoc Am|bool|isWORDCHAR|char ch
1009Returns a boolean indicating whether the specified character is a character
1010that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1011in a regular expression. A word character is an alphabetic character, a
1012decimal digit, a connecting punctuation character (such as an underscore), or
1013a "mark" character that attaches to one of those (like some sort of accent).
1014C<isALNUM()> is a synonym provided for backward compatibility, even though a
1015word character includes more than the standard C language meaning of
1016alphanumeric.
dcccc8ff 1017See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
1018variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1019C<isWORDCHAR_utf8>, and C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>,
1020C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and C<isWORDCHAR_LC_utf8_safe>
1021are also as described there, but additionally include the platform's native
1022underscore.
8a58bdcf 1023
f16858ed
KW
1024=for apidoc Amh|bool|isWORDCHAR_A|int ch
1025=for apidoc Amh|bool|isWORDCHAR_L1|int ch
1026=for apidoc Amh|bool|isWORDCHAR_uvchr|int ch
1027=for apidoc Amh|bool|isWORDCHAR_utf8_safe|U8 * s|U8 * end
d23c7e08 1028=for apidoc Amh|bool|isWORDCHAR_utf8|U8 * s|U8 * end
f16858ed
KW
1029=for apidoc Amh|bool|isWORDCHAR_LC|int ch
1030=for apidoc Amh|bool|isWORDCHAR_LC_uvchr|int ch
1031=for apidoc Amh|bool|isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
1032=for apidoc Amh|bool|isALNUM|int ch
1033=for apidoc Amh|bool|isALNUM_A|int ch
1034=for apidoc Amh|bool|isALNUM_LC|int ch
1035=for apidoc Amh|bool|isALNUM_LC_uvchr|int ch
1036
8a58bdcf
KW
1037=for apidoc Am|bool|isXDIGIT|char ch
1038Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
1039digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1040and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
1041See the L<top of this section|/Character classification> for an explanation of
1042variants
059703b0
KW
1043C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>,
1044C<isXDIGIT_LC_uvchr>, C<isXDIGIT_LC_utf8>, and C<isXDIGIT_LC_utf8_safe>.
243effed 1045
f16858ed
KW
1046=for apidoc Amh|bool|isXDIGIT_A|int ch
1047=for apidoc Amh|bool|isXDIGIT_L1|int ch
1048=for apidoc Amh|bool|isXDIGIT_uvchr|int ch
1049=for apidoc Amh|bool|isXDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 1050=for apidoc Amh|bool|isXDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
1051=for apidoc Amh|bool|isXDIGIT_LC|int ch
1052=for apidoc Amh|bool|isXDIGIT_LC_uvchr|int ch
1053=for apidoc Amh|bool|isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1054
3c3ecf18
KW
1055=for apidoc Am|bool|isIDFIRST|char ch
1056Returns a boolean indicating whether the specified character can be the first
1057character of an identifier. This is very close to, but not quite the same as
1058the official Unicode property C<XID_Start>. The difference is that this
1059returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
1060See the L<top of this section|/Character classification> for an explanation of
1061variants
059703b0
KW
1062C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
1063C<isIDFIRST_utf8_safe>, C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>,
1064C<isIDFIRST_LC_utf8>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18 1065
f16858ed
KW
1066=for apidoc Amh|bool|isIDFIRST_A|int ch
1067=for apidoc Amh|bool|isIDFIRST_L1|int ch
1068=for apidoc Amh|bool|isIDFIRST_uvchr|int ch
1069=for apidoc Amh|bool|isIDFIRST_utf8_safe|U8 * s|U8 * end
d23c7e08 1070=for apidoc Amh|bool|isIDFIRST_utf8|U8 * s|U8 * end
f16858ed
KW
1071=for apidoc Amh|bool|isIDFIRST_LC|int ch
1072=for apidoc Amh|bool|isIDFIRST_LC_uvchr|int ch
1073=for apidoc Amh|bool|isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1074
3c3ecf18
KW
1075=for apidoc Am|bool|isIDCONT|char ch
1076Returns a boolean indicating whether the specified character can be the
1077second or succeeding character of an identifier. This is very close to, but
1078not quite the same as the official Unicode property C<XID_Continue>. The
1079difference is that this returns true only if the input character also matches
dcccc8ff 1080L</isWORDCHAR>. See the L<top of this section|/Character classification> for
059703b0
KW
1081an explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
1082C<isIDCONT_utf8>, C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>,
1083C<isIDCONT_LC_utf8>, and C<isIDCONT_LC_utf8_safe>.
3c3ecf18 1084
f16858ed
KW
1085=for apidoc Amh|bool|isIDCONT_A|int ch
1086=for apidoc Amh|bool|isIDCONT_L1|int ch
1087=for apidoc Amh|bool|isIDCONT_uvchr|int ch
1088=for apidoc Amh|bool|isIDCONT_utf8_safe|U8 * s|U8 * end
d23c7e08 1089=for apidoc Amh|bool|isIDCONT_utf8|U8 * s|U8 * end
f16858ed
KW
1090=for apidoc Amh|bool|isIDCONT_LC|int ch
1091=for apidoc Amh|bool|isIDCONT_LC_uvchr|int ch
1092=for apidoc Amh|bool|isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1093
243effed 1094=head1 Miscellaneous Functions
8eea39dd 1095
95a59cab 1096=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 1097Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
1098Behaviour is only well defined when isXDIGIT(*str) is true.
1099
e7c1e6c1 1100=head1 Character case changing
21da7284
KW
1101Perl uses "full" Unicode case mappings. This means that converting a single
1102character to another case may result in a sequence of more than one character.
1103For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1104character sequence C<SS>. This presents some complications The lowercase of
1105all characters in the range 0..255 is a single character, and thus
1106C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1107return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1108an API that does allow every possible legal result to be returned.) Likewise
1109no other function that is crippled by not being able to give the correct
1110results for the full range of possible inputs has been implemented here.
e7c1e6c1 1111
d713f9d9 1112=for apidoc Am|U8|toUPPER|int ch
1f607577
KW
1113Converts the specified character to uppercase. If the input is anything but an
1114ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 1115C<toUPPER_A> is equivalent.
954c1994 1116
d0da05db
KW
1117=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1118Converts the code point C<cp> to its uppercase version, and
1119stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1120point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1121that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1122bytes since the uppercase version may be longer than the original character.
1123
1124The first code point of the uppercased version is returned
21da7284
KW
1125(but note, as explained at L<the top of this section|/Character case
1126changing>, that there may be more.)
1f607577 1127
059703b0 1128=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1129Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1130extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
1131stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1132that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1133bytes since the uppercase version may be longer than the original character.
1134
1135The first code point of the uppercased version is returned
21da7284
KW
1136(but note, as explained at L<the top of this section|/Character case
1137changing>, that there may be more).
1f607577 1138
059703b0
KW
1139It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1140S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1141the UTF-8 for the input character is malformed in some way, the program may
1142croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1143of the implementation, and subject to change in future releases.
a239b1e2 1144
059703b0
KW
1145=for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1146Same as L</toUPPER_utf8>.
1f607577 1147
25200305
KW
1148=for apidoc Am|U8|toFOLD|U8 ch
1149Converts the specified character to foldcase. If the input is anything but an
1150ASCII uppercase character, that input character itself is returned. Variant
1151C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 1152Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 1153
d0da05db
KW
1154=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1155Converts the code point C<cp> to its foldcase version, and
1156stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1157point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1158that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1159bytes since the foldcase version may be longer than the original character.
1160
1161The first code point of the foldcased version is returned
21da7284
KW
1162(but note, as explained at L<the top of this section|/Character case
1163changing>, that there may be more).
1f607577 1164
059703b0 1165=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1166Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1167extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
1168stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1169that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1170bytes since the foldcase version may be longer than the original character.
1171
1172The first code point of the foldcased version is returned
21da7284
KW
1173(but note, as explained at L<the top of this section|/Character case
1174changing>, that there may be more).
1f607577 1175
059703b0 1176It will not attempt
a239b1e2
KW
1177to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1178true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1179input character is malformed in some way, the program may croak, or the
1180function may return the REPLACEMENT CHARACTER, at the discretion of the
1181implementation, and subject to change in future releases.
1182
059703b0
KW
1183=for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1184Same as L</toFOLD_utf8>.
1f607577
KW
1185
1186=for apidoc Am|U8|toLOWER|U8 ch
1187Converts the specified character to lowercase. If the input is anything but an
1188ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 1189C<toLOWER_A> is equivalent.
954c1994 1190
1f607577 1191=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
1192Converts the specified Latin1 character to lowercase. The results are
1193undefined if the input doesn't fit in a byte.
1f607577
KW
1194
1195=for apidoc Am|U8|toLOWER_LC|U8 ch
1196Converts the specified character to lowercase using the current locale's rules,
1197if possible; otherwise returns the input character itself.
1198
d0da05db
KW
1199=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1200Converts the code point C<cp> to its lowercase version, and
1201stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1202point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1203that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1204bytes since the lowercase version may be longer than the original character.
1205
1206The first code point of the lowercased version is returned
21da7284
KW
1207(but note, as explained at L<the top of this section|/Character case
1208changing>, that there may be more).
1f607577 1209
bd350c85 1210=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1211Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1212extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
1213stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1214that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1215bytes since the lowercase version may be longer than the original character.
1216
1217The first code point of the lowercased version is returned
21da7284
KW
1218(but note, as explained at L<the top of this section|/Character case
1219changing>, that there may be more).
059703b0
KW
1220It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1221S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1222the UTF-8 for the input character is malformed in some way, the program may
1223croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1224of the implementation, and subject to change in future releases.
1f607577 1225
059703b0
KW
1226=for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1227Same as L</toLOWER_utf8>.
1f607577 1228
25200305
KW
1229=for apidoc Am|U8|toTITLE|U8 ch
1230Converts the specified character to titlecase. If the input is anything but an
1231ASCII lowercase character, that input character itself is returned. Variant
b7d90381 1232C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 1233range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 1234not a concept used in locale handling, so there is no functionality for that.)
25200305 1235
d0da05db
KW
1236=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1237Converts the code point C<cp> to its titlecase version, and
1238stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1239point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1240that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1241bytes since the titlecase version may be longer than the original character.
1242
1243The first code point of the titlecased version is returned
21da7284
KW
1244(but note, as explained at L<the top of this section|/Character case
1245changing>, that there may be more).
1f607577 1246
059703b0 1247=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1248Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1249extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1250stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1251that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1252bytes since the titlecase version may be longer than the original character.
1253
1254The first code point of the titlecased version is returned
21da7284
KW
1255(but note, as explained at L<the top of this section|/Character case
1256changing>, that there may be more).
1f607577 1257
059703b0 1258It will not attempt
a239b1e2
KW
1259to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1260true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1261input character is malformed in some way, the program may croak, or the
1262function may return the REPLACEMENT CHARACTER, at the discretion of the
1263implementation, and subject to change in future releases.
1264
059703b0
KW
1265=for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1266Same as L</toTITLE_utf8>.
1f607577 1267
954c1994 1268=cut
353c9b6f 1269
d0da05db 1270XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1271really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1272and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1273
8a58bdcf 1274Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1275patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1276
954c1994
GS
1277*/
1278
8f5283f4
KW
1279/*
1280 void below because that's the best fit, and works for Devel::PPPort
1281=for apidoc AmnU|void|WIDEST_UTYPE
1282
1283Yields the widest unsigned integer type on the platform, currently either
1284C<U32> or C<64>. This can be used in declarations such as
1285
1286 WIDEST_UTYPE my_uv;
1287
1288or casts
1289
1290 my_uv = (WIDEST_UTYPE) val;
1291
1292=cut
1293
1294*/
de9e2639
KW
1295#ifdef QUADKIND
1296# define WIDEST_UTYPE U64
7c062697
KW
1297#else
1298# define WIDEST_UTYPE U32
1299#endif
1300
3912bc88
KW
1301/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1302 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1303 * bits of information are lost even on a 64-bit machine, but to get the
1304 * compiler to optimize it out if possible. This is because Configure makes
1305 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1306 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1307 *
1308 * For Coverity, be always true, because otherwise Coverity thinks
1309 * it finds several expressions that are always true, independent
1310 * of operands. Well, they are, but that is kind of the point.
220c71bf 1311 */
7e75d1a1 1312#ifndef __COVERITY__
6c5b02ac
KW
1313 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1314 * pointer) */
1315#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1316 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1317#else
1318#define FITS_IN_8_BITS(c) (1)
1319#endif
cf301eb7 1320
45f4bb73 1321/* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
833b0f46 1322 * Written this way so that after optimization, only one conditional test is
76d3ad4c
KW
1323 * needed. (The NV casts stop any warnings about comparison always being true
1324 * if called with an unsigned. The cast preserves the sign, which is all we
1325 * care about.) */
1326#define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1327 __ASSERT_((NV) (n) >= 0) \
1328 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
833b0f46 1329
94250c4f
KW
1330/* Returns true if c is in the range l..u, where 'l' is non-negative
1331 * Written this way so that after optimization, only one conditional test is
4758c20d 1332 * needed. */
1eaefa6e 1333#define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
4758c20d 1334 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
4758c20d
KW
1335 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1336 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
45f4bb73 1337 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
305fe86e 1338
41f43cc2 1339#ifdef EBCDIC
b6340bd0 1340# ifndef _ALL_SOURCE
0852beac
KW
1341 /* The native libc isascii() et.al. functions return the wrong results
1342 * on at least z/OS unless this is defined. */
b6340bd0
KW
1343# error _ALL_SOURCE should probably be defined
1344# endif
41f43cc2 1345#else
0852beac
KW
1346 /* There is a simple definition of ASCII for ASCII platforms. But the
1347 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1348 * other macros below.
3f3c579d
KW
1349 *
1350 * The cast here is used instead of '(c) >= 0', because some compilers emit
1351 * a warning that that test is always true when the parameter is an
1352 * unsigned type. khw supposes that it could be written as
1353 * && ((c) == '\0' || (c) > 0)
1354 * to avoid the message, but the cast will likely avoid extra branches even
1355 * with stupid compilers.
1356 *
1357 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1358 * a pointer) */
9c903d59 1359# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1360#endif
1361
38694112
KW
1362/* Take the eight possible bit patterns of the lower 3 bits and you get the
1363 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1364 * can be ignored. If the rest match '0', we have an octal */
1365#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1366
9fb1bf9d 1367#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1368
a500dc72
KW
1369/* Character class numbers. For internal core Perl use only. The ones less
1370 * than 32 are used in PL_charclass[] and the ones up through the one that
1371 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1372 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1373 * their actual definitions are here. If that file has a name not used here,
1374 * it won't compile.
1709d539
KW
1375 *
1376 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561 1377 * frequency of their use. This gives a slight edge to exiting a loop earlier
58a3ba2c
KW
1378 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1379 * real Posix character class, and some (small) inefficiencies in regular
1380 * expression handling would be introduced by putting it in the middle of those
1381 * that are. Also, cntrl and ascii come after the others as it may be useful
1382 * to group these which have no members that match above Latin1, (or above
1383 * ASCII in the latter case) */
1384
1709d539
KW
1385# define _CC_WORDCHAR 0 /* \w and [:word:] */
1386# define _CC_DIGIT 1 /* \d and [:digit:] */
1387# define _CC_ALPHA 2 /* [:alpha:] */
1388# define _CC_LOWER 3 /* [:lower:] */
1389# define _CC_UPPER 4 /* [:upper:] */
1390# define _CC_PUNCT 5 /* [:punct:] */
1391# define _CC_PRINT 6 /* [:print:] */
15861f94 1392# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1393# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1394# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
779cf272 1395# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1396# define _CC_BLANK 11 /* [:blank:] */
1397# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1398# define _CC_CNTRL 13 /* [:cntrl:] */
1399# define _CC_ASCII 14 /* [:ascii:] */
1400# define _CC_VERTSPACE 15 /* \v */
1709d539 1401
a0947d7b
KW
1402# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1403
1709d539 1404/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1405 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1406# define _CC_IDFIRST 16
1407# define _CC_CHARNAME_CONT 17
1408# define _CC_NONLATIN1_FOLD 18
1409# define _CC_NONLATIN1_SIMPLE_FOLD 19
1410# define _CC_QUOTEMETA 20
1411# define _CC_NON_FINAL_FOLD 21
1412# define _CC_IS_IN_SOME_FOLD 22
2ae9030c
KW
1413# define _CC_BINDIGIT 23
1414# define _CC_OCTDIGIT 24
1415# define _CC_MNEMONIC_CNTRL 25
073c22b3
KW
1416
1417/* This next group is only used on EBCDIC platforms, so theoretically could be
1418 * shared with something entirely different that's only on ASCII platforms */
abb8abf6 1419# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
0654f0ab 1420/* Unused: 26-30
f4cdb42c
KW
1421 * If more bits are needed, one could add a second word for non-64bit
1422 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1423 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1424 * is used only for optimization (as of this writing), and differs in the
1425 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1426 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1427 * would just cause /i regexes which match them to run less efficiently.
1428 * Similarly the EBCDIC-only bits are used just for speed, and could be
1429 * replaced by other means */
96ac0975 1430
3a371f2f
KW
1431#if defined(PERL_CORE) || defined(PERL_EXT)
1432/* An enum version of the character class numbers, to help compilers
1433 * optimize */
1434typedef enum {
3a371f2f 1435 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1436 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1437 _CC_ENUM_ASCII = _CC_ASCII,
1438 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1439 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1440 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1441 _CC_ENUM_DIGIT = _CC_DIGIT,
1442 _CC_ENUM_GRAPH = _CC_GRAPH,
1443 _CC_ENUM_LOWER = _CC_LOWER,
1444 _CC_ENUM_PRINT = _CC_PRINT,
1445 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1446 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1447 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1448 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1449 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1450 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1451} _char_class_number;
1452#endif
1453
86f72d56 1454#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1455
6635f04f 1456START_EXTERN_C
96ac0975
NC
1457# ifdef DOINIT
1458EXTCONST U32 PL_charclass[] = {
1459# include "l1_char_class_tab.h"
1460};
1461
1462# else /* ! DOINIT */
1463EXTCONST U32 PL_charclass[];
1464# endif
6635f04f 1465END_EXTERN_C
96ac0975 1466
265c1f46 1467 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1468# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1469
1470 /* For internal core Perl use only: the base macro for defining macros like
1471 * isALPHA */
ff7ecfc3 1472# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1473 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1474
f4cdb42c
KW
1475 /* The mask for the _A versions of the macros; it just adds in the bit for
1476 * ASCII. */
1477# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1478
4650c663
KW
1479 /* For internal core Perl use only: the base macro for defining macros like
1480 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1481 * the ASCII bit are present */
b7d90381
KW
1482# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1483 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1484 == _CC_mask_A(classnum)))
f4cdb42c 1485
26c1d9d8
KW
1486/* On ASCII platforms certain classes form a single range. It's faster to
1487 * special case these. isDIGIT is a single range on all platforms */
b877c1ff
KW
1488# ifdef EBCDIC
1489# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1490# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1491# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1492# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1493# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1494# else
26c1d9d8 1495 /* By folding the upper and lowercase, we can use a single range */
b877c1ff 1496# define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
26c1d9d8 1497# define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
b877c1ff
KW
1498# define isLOWER_A(c) inRANGE(c, 'a', 'z')
1499# define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1500# define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1501# endif
15861f94 1502# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1503# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1504# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
b877c1ff 1505# define isDIGIT_A(c) inRANGE(c, '0', '9')
f4cdb42c
KW
1506# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1507# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
f4cdb42c 1508# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1509# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1510 */
d95f8b6a 1511# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1512# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1513# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1514# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1515
1516 /* continuation character for legal NAME in \N{NAME} */
1517# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1518
1519# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1520# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1521# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1522# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1523# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1524# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1525# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1526# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1527# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1528# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1529
0852beac
KW
1530# ifdef EBCDIC
1531# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1532# endif
1533
f12c0118
KW
1534 /* Participates in a single-character fold with a character above 255 */
1535# 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)))
1536
1537 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1538# 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 1539
4eeeb416 1540# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1541# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1542 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1543# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1544 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
5e6ebb12
KW
1545
1546/* is c a control character for which we have a mnemonic? */
1547# if defined(PERL_CORE) || defined(PERL_EXT)
1548# define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1549# endif
687c8d01 1550#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1551
1552 /* If we don't have perl.h, we are compiling a utility program. Below we
1553 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1554 * to it. Most are coded based on first principles. These are written to
74665a89 1555 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
182c4ace 1556# define isDIGIT_A(c) inRANGE(c, '0', '9')
0852beac 1557# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1558# define isSPACE_A(c) (isBLANK_A(c) \
1559 || (c) == '\n' \
1560 || (c) == '\r' \
1561 || (c) == '\v' \
0852beac 1562 || (c) == '\f')
74665a89
KW
1563 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1564 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1565 * performance (if optimization isn't on), and allow the same code to be
1566 * used for both platform types */
182c4ace
KW
1567# define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1568 || inRANGE((c), 'j', 'r') \
1569 || inRANGE((c), 's', 'z')
1570# define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1571 || inRANGE((c), 'J', 'R') \
1572 || inRANGE((c), 'S', 'Z')
a4d7a999
KW
1573# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1574# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1575# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1576# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
182c4ace
KW
1577# define isXDIGIT_A(c) ( isDIGIT_A(c) \
1578 || inRANGE((c), 'a', 'f') \
1579 || inRANGE((c), 'A', 'F')
74665a89
KW
1580# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1581 || (c) == '#' || (c) == '$' || (c) == '%' \
1582 || (c) == '&' || (c) == '\'' || (c) == '(' \
1583 || (c) == ')' || (c) == '*' || (c) == '+' \
1584 || (c) == ',' || (c) == '.' || (c) == '/' \
1585 || (c) == ':' || (c) == ';' || (c) == '<' \
1586 || (c) == '=' || (c) == '>' || (c) == '?' \
1587 || (c) == '@' || (c) == '[' || (c) == '\\' \
1588 || (c) == ']' || (c) == '^' || (c) == '_' \
1589 || (c) == '`' || (c) == '{' || (c) == '|' \
1590 || (c) == '}' || (c) == '~')
1591# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1592# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1593
0852beac 1594# ifdef EBCDIC
74665a89
KW
1595 /* The below is accurate for the 3 EBCDIC code pages traditionally
1596 * supported by perl. The only difference between them in the controls
1597 * is the position of \n, and that is represented symbolically below */
1598# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1599 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1600 || (c) == '\t' || (c) == '\v' \
182c4ace 1601 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
8ec0a736 1602 || (c) == 7F /* U+7F DEL */ \
182c4ace
KW
1603 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1604 DC[1-3] */ \
74665a89
KW
1605 || (c) == 0x18 /* U+18 CAN */ \
1606 || (c) == 0x19 /* U+19 EOM */ \
182c4ace 1607 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
74665a89
KW
1608 || (c) == 0x26 /* U+17 ETB */ \
1609 || (c) == 0x27 /* U+1B ESC */ \
1610 || (c) == 0x2D /* U+05 ENQ */ \
1611 || (c) == 0x2E /* U+06 ACK */ \
1612 || (c) == 0x32 /* U+16 SYN */ \
1613 || (c) == 0x37 /* U+04 EOT */ \
1614 || (c) == 0x3C /* U+14 DC4 */ \
1615 || (c) == 0x3D /* U+15 NAK */ \
1616 || (c) == 0x3F)/* U+1A SUB */
0852beac 1617# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1618# else /* isASCII is already defined for ASCII platforms, so can use that to
1619 define isCNTRL */
1620# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1621# endif
1622
3ffc8c70 1623 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1624 * during debugging, and it seems best to keep them. We may be called
1625 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1626 * do anything anyway, so make it not a problem */
1627# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1628# define NATIVE_TO_LATIN1(ch) (ch)
1629# endif
3ded5eb0
KW
1630# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1631# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1632# define isBLANK_L1(c) (isBLANK_A(c) \
1633 || (FITS_IN_8_BITS(c) \
1634 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1635# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1636# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1637# define isLOWER_L1(c) (isLOWER_A(c) \
1638 || (FITS_IN_8_BITS(c) \
ae683a5f 1639 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1640 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1641 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1642 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1643 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1644# define isPRINT_L1(c) (isPRINT_A(c) \
1645 || (FITS_IN_8_BITS(c) \
1646 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1647# define isPUNCT_L1(c) (isPUNCT_A(c) \
1648 || (FITS_IN_8_BITS(c) \
ae683a5f 1649 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1650 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1651 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1652 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1653 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1654 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1655 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1656# define isSPACE_L1(c) (isSPACE_A(c) \
1657 || (FITS_IN_8_BITS(c) \
ae683a5f 1658 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1659 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1660# define isUPPER_L1(c) (isUPPER_A(c) \
1661 || (FITS_IN_8_BITS(c) \
182c4ace
KW
1662 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1663 0xC0, 0xDE) \
3ded5eb0
KW
1664 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1665# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1666# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1667# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1668 || isBLANK_L1(c) \
1669 || (c) == '-' \
1670 || (c) == '(' \
1671 || (c) == ')')
1672 /* The following are not fully accurate in the above-ASCII range. I (khw)
1673 * don't think it's necessary to be so for the purposes where this gets
1674 * compiled */
1675# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1676# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1677
1678 /* And these aren't accurate at all. They are useful only for above
1679 * Latin1, which utilities and bootstrapping don't deal with */
1680# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1681# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1682# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1683
1684 /* Many of the macros later in this file are defined in terms of these. By
1685 * implementing them with a function, which converts the class number into
1686 * a call to the desired macro, all of the later ones work. However, that
1687 * function won't be actually defined when building a utility program (no
1688 * perl.h), and so a compiler error will be generated if one is attempted
1689 * to be used. And the above-Latin1 code points require Unicode tables to
1690 * be present, something unlikely to be the case when bootstrapping */
1691# define _generic_isCC(c, classnum) \
1692 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1693# define _generic_isCC_A(c, classnum) \
1694 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1695#endif /* End of no perl.h H_PERL */
8a58bdcf 1696
e66b99e9
KW
1697#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1698#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1699#define isASCII_A(c) isASCII(c)
1700#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1701#define isBLANK(c) isBLANK_A(c)
1702#define isCNTRL(c) isCNTRL_A(c)
1703#define isDIGIT(c) isDIGIT_A(c)
1704#define isGRAPH(c) isGRAPH_A(c)
1705#define isIDFIRST(c) isIDFIRST_A(c)
1706#define isLOWER(c) isLOWER_A(c)
1707#define isPRINT(c) isPRINT_A(c)
779cf272 1708#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1709#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1710#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1711#define isPUNCT(c) isPUNCT_A(c)
1712#define isSPACE(c) isSPACE_A(c)
1713#define isUPPER(c) isUPPER_A(c)
1714#define isWORDCHAR(c) isWORDCHAR_A(c)
1715#define isXDIGIT(c) isXDIGIT_A(c)
1716
1717/* ASCII casing. These could also be written as
1718 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1719 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1720 which uses table lookup and mask instead of subtraction. (This would
c5e9991e
KW
1721 work because the _MOD does not apply in the ASCII range).
1722
1723 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1724 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1725 there non-ASCII invariants, and all of them are controls.) */
68067e4e
DM
1726#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1727#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1728
25200305
KW
1729/* In the ASCII range, these are equivalent to what they're here defined to be.
1730 * But by creating these definitions, other code doesn't have to be aware of
c5e9991e
KW
1731 * this detail. Actually this works for all UTF-8 invariants, not just the
1732 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
25200305 1733#define toFOLD(c) toLOWER(c)
25200305
KW
1734#define toTITLE(c) toUPPER(c)
1735
c753c8d3
KW
1736#define toLOWER_A(c) toLOWER(c)
1737#define toUPPER_A(c) toUPPER(c)
25200305
KW
1738#define toFOLD_A(c) toFOLD(c)
1739#define toTITLE_A(c) toTITLE(c)
1a0901db 1740
4650c663 1741/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1742#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1743 ? (c) \
f4cd282c 1744 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1745#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1746
1a0901db 1747/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1748 * all mapped to one of them, and these need special handling; returns the
1749 * input itself if is out-of-range */
b2bf251f 1750#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1751 ? (c) \
f4cd282c 1752 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1753#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1754
beab9ebe
KW
1755/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1756
1757/* For internal core Perl use only: the base macro for defining macros like
1758 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1759 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1760 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1761 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1762 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1763 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1764 * code points under locale rules, use the macros based on _generic_LC_uvchr
1765 * instead of this. */
beab9ebe
KW
1766#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1767 (! FITS_IN_8_BITS(c) \
1768 ? 0 \
31f05a37
KW
1769 : IN_UTF8_CTYPE_LOCALE \
1770 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1771 : cBOOL(non_utf8))
1772
1773/* For internal core Perl use only: a helper macro for defining macros like
1774 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1775 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1776 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1777 * _generic_LC_base for more info */
1778#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1779 _generic_LC_base(c,utf8_locale_classnum, \
1780 non_utf8_func( (_LC_CAST) (c)))
1781
1782/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1783 * 'c' is the platform's native underscore character */
1784#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1785 _generic_LC_base(c, utf8_locale_classnum, \
1786 (non_utf8_func( (_LC_CAST) (c)) \
1787 || (char)(c) == '_'))
1788
1789/* These next three are also for internal core Perl use only: case-change
247985d4
KW
1790 * helper macros. The reason for using the PL_latin arrays is in case the
1791 * system function is defective; it ensures uniform results that conform to the
b257a28c 1792 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
beab9ebe
KW
1793#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1794 ? (c) \
31f05a37
KW
1795 : (IN_UTF8_CTYPE_LOCALE) \
1796 ? PL_latin1_lc[ (U8) (c) ] \
5a10328c 1797 : (cast)function((cast)(c)))
beab9ebe 1798
31f05a37
KW
1799/* Note that the result can be larger than a byte in a UTF-8 locale. It
1800 * returns a single value, so can't adequately return the upper case of LATIN
1801 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1802 * values "SS"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1803 * otherwise returns its input. It does not handle the anomalies in UTF-8
1804 * Turkic locales. */
beab9ebe
KW
1805#define _generic_toUPPER_LC(c, function, cast) \
1806 (! FITS_IN_8_BITS(c) \
1807 ? (c) \
31f05a37 1808 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1809 ? (cast)function((cast)(c)) \
31f05a37
KW
1810 : ((((U8)(c)) == MICRO_SIGN) \
1811 ? GREEK_CAPITAL_LETTER_MU \
1812 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1813 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1814 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1815 ? (__ASSERT_(0) (c)) \
1816 : PL_mod_latin1_uc[ (U8) (c) ])))))
1817
1818/* Note that the result can be larger than a byte in a UTF-8 locale. It
1819 * returns a single value, so can't adequately return the fold case of LATIN
1820 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1821 * values "ss"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1822 * otherwise returns its input. It does not handle the anomalies in UTF-8
1823 * Turkic locales */
beab9ebe 1824#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1825 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1826 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1827 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1828 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1829 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1830
84061b6a 1831/* Use the libc versions for these if available. */
f05550c0 1832#if defined(HAS_ISASCII)
84061b6a
KW
1833# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1834#else
1835# define isASCII_LC(c) isASCII(c)
1836#endif
1837
f05550c0 1838#if defined(HAS_ISBLANK)
84061b6a 1839# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1840#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1841# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1842#endif
1843
f05550c0 1844#define _LC_CAST U8
bbce6d69 1845
f05550c0 1846#ifdef WIN32
375f5f06
KW
1847 /* The Windows functions don't bother to follow the POSIX standard, which
1848 * for example says that something can't both be a printable and a control.
1849 * But Windows treats the \t control as a printable, and does such things
1850 * as making superscripts into both digits and punctuation. This tames
1851 * these flaws by assuming that the definitions of both controls and space
1852 * are correct, and then making sure that other definitions don't have
1853 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1854 * Not all possible weirdnesses are checked for, just the ones that were
1855 * detected on actual Microsoft code pages */
1856
b7d90381
KW
1857# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1858# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1859
1860# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1861 && isALPHANUMERIC_LC(c))
1862# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1863 ! isPUNCT_LC(c))
1864# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1865 isALPHANUMERIC_LC(c))
1866# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1867# define isIDFIRST_LC(c) (((c) == '_') \
1868 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1869# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1870# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1871# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1872# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1873# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1874# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1875 && isALPHANUMERIC_LC(c))
f05550c0
BF
1876
1877# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1878# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1879# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1880
1881#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1882 /* For most other platforms */
beab9ebe 1883
f05550c0
BF
1884# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1885# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1886# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1887# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1888# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1889# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1890# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1891# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1892# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1893# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1894# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1895# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1896# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1897
1898
1899# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1900# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1901# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1902
1903#else /* The final fallback position */
1904
b7d90381
KW
1905# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1906# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1907# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1908# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1909# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1910# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1911# define isLOWER_LC(c) (isascii(c) && islower(c))
1912# define isPRINT_LC(c) (isascii(c) && isprint(c))
1913# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1914# define isSPACE_LC(c) (isascii(c) && isspace(c))
1915# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1916# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1917# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1918
1919# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1920# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1921# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1922
f05550c0 1923#endif
55204971 1924
eba68aa0
KW
1925#define isIDCONT(c) isWORDCHAR(c)
1926#define isIDCONT_A(c) isWORDCHAR_A(c)
1927#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1928#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1929#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1930
4650c663 1931/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1932 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1933 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1934 * classes where there is a macro or function 'above_latin1' that takes the
1935 * single argument 'c' and returns the desired value. These exist for those
2366ba44
KW
1936 * classes which have simple definitions, avoiding the overhead of an inversion
1937 * list binary search. _generic_invlist_uvchr() can be used
4650c663 1938 * for classes where that overhead is faster than a direct lookup.
d0da05db 1939 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1940 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1941 * have duplicate checks here, so could create versions of the macros that
1942 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1943
1944/* Note that all ignore 'use bytes' */
1e222e4f
KW
1945#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1946 ? _generic_isCC(c, classnum) \
cd500f2f 1947 : above_latin1(c))
2366ba44 1948#define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
1e222e4f 1949 ? _generic_isCC(c, classnum) \
922e8cb4 1950 : _is_uni_FOO(classnum, c))
2366ba44
KW
1951#define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
1952#define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
d0da05db
KW
1953#define isASCII_uvchr(c) isASCII(c)
1954#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1955#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2366ba44
KW
1956#define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
1957#define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1958#define isIDCONT_uvchr(c) \
1959 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1960#define isIDFIRST_uvchr(c) \
1961 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2366ba44
KW
1962#define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
1963#define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
d0da05db 1964
2366ba44 1965#define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
d0da05db
KW
1966#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1967#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1968
2366ba44 1969#define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
d0da05db 1970#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2366ba44 1971#define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
d0da05db
KW
1972#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1973
1974#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1975#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1976#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1977#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1978
1979/* For backwards compatibility, even though '_uni' should mean official Unicode
1980 * code points, in Perl it means native for those below 256 */
1981#define isALPHA_uni(c) isALPHA_uvchr(c)
1982#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1983#define isASCII_uni(c) isASCII_uvchr(c)
1984#define isBLANK_uni(c) isBLANK_uvchr(c)
1985#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1986#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1987#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1988#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1989#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1990#define isLOWER_uni(c) isLOWER_uvchr(c)
1991#define isPRINT_uni(c) isPRINT_uvchr(c)
1992#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1993#define isSPACE_uni(c) isSPACE_uvchr(c)
1994#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1995#define isUPPER_uni(c) isUPPER_uvchr(c)
1996#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1997#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1998#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1999#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2000#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2001#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2002#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 2003
4650c663
KW
2004/* For internal core Perl use only: the base macros for defining macros like
2005 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 2006 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 2007 * simple class definitions; the other for more complex. These are like
d0da05db 2008 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
2009#define _generic_LC_uvchr(latin1, above_latin1, c) \
2010 (c < 256 ? latin1(c) : above_latin1(c))
2366ba44 2011#define _generic_LC_invlist_uvchr(latin1, classnum, c) \
cd500f2f
KW
2012 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2013
2366ba44
KW
2014#define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2015#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
cd500f2f 2016 _CC_ALPHANUMERIC, c)
b7d90381
KW
2017#define isASCII_LC_uvchr(c) isASCII_LC(c)
2018#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2019 is_HORIZWS_cp_high, c)
feeab5a9 2020#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2366ba44
KW
2021#define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2022#define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 2023#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 2024 _is_uni_perl_idcont, c)
b7d90381 2025#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f 2026 _is_uni_perl_idstart, c)
2366ba44
KW
2027#define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2028#define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 2029#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2366ba44 2030#define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 2031#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 2032 is_XPERLSPACE_cp_high, c)
2366ba44
KW
2033#define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2034#define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
cd500f2f 2035 _CC_WORDCHAR, c)
b7d90381
KW
2036#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2037 is_XDIGIT_cp_high, c)
e712593e 2038
b7d90381 2039#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 2040
da8c1a98
KW
2041/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2042 * they don't otherwise go out of their way to look for malformed UTF-8. If
2043 * they can return accurate results without knowing if the input is otherwise
2044 * malformed, they do so. For example isASCII is accurate in spite of any
2045 * non-length malformations because it looks only at a single byte. Likewise
2046 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2047 * variant ones return FALSE. But, if the input has to be well-formed in order
2048 * for the results to be accurate, the macros will test and if malformed will
2049 * call a routine to die
2050 *
2051 * Except for toke.c, the macros do assume that e > p, asserting that on
2052 * DEBUGGING builds. Much code that calls these depends on this being true,
2053 * for other reasons. toke.c is treated specially as using the regular
2054 * assertion breaks it in many ways. All strings that these operate on there
2055 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2056 * bunch of code in toke.c assumes that this is true, so the assertion allows
2057 * for that */
2058#ifdef PERL_IN_TOKE_C
2059# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2060#else
2061# define _utf8_safe_assert(p,e) ((e) > (p))
2062#endif
2063
2064#define _generic_utf8_safe(classnum, p, e, above_latin1) \
c81b3562
KW
2065 ((! _utf8_safe_assert(p, e)) \
2066 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2067 : (UTF8_IS_INVARIANT(*(p))) \
da8c1a98
KW
2068 ? _generic_isCC(*(p), classnum) \
2069 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2070 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2071 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2072 classnum) \
2073 : (_force_out_malformed_utf8_message( \
2074 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2075 : above_latin1))
b7d90381
KW
2076/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2077 * 'above_latin1' can be a macro */
da8c1a98
KW
2078#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2079 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2366ba44 2080#define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2081 _generic_utf8_safe(classnum, p, e, \
2082 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2083 ? (_force_out_malformed_utf8_message( \
2084 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2085 : above_latin1(p)))
2366ba44
KW
2086/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2087 * 'above_latin1' parameter */
2088#define _generic_invlist_utf8_safe(classnum, p, e) \
2089 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
922e8cb4 2090
cc8ab7c0 2091/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
2092 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2093 * class is TRUE for. Hence it can skip the tests for this range.
2094 * 'above_latin1' should include its arguments */
da8c1a98
KW
2095#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2096 (__ASSERT_(_utf8_safe_assert(p, e)) \
2097 (UTF8_IS_INVARIANT(*(p))) \
2098 ? _generic_isCC(*(p), classnum) \
2099 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2100 ? 0 /* Note that doesn't check validity for latin1 */ \
2101 : above_latin1)
2102
84238efa 2103
059703b0
KW
2104#define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2105#define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2106#define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2107#define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2108#define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2109#define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2110#define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2111#define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2112#define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2113#define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2114#define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2115#define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2116#define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2117#define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2118#define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2119#define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2120#define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2121#define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
e8fa43e2 2122
2366ba44 2123#define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
da8c1a98 2124#define isALPHANUMERIC_utf8_safe(p, e) \
2366ba44 2125 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
da8c1a98
KW
2126#define isASCII_utf8_safe(p, e) \
2127 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2128 * works */ \
2129 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2130#define isBLANK_utf8_safe(p, e) \
2366ba44 2131 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
da8c1a98 2132
e8fa43e2
KW
2133#ifdef EBCDIC
2134 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2135 * more efficient macro instead of the more general one */
da8c1a98 2136# define isCNTRL_utf8_safe(p, e) \
56d02b8c 2137 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 2138#else
da8c1a98 2139# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
2140#endif
2141
da8c1a98
KW
2142#define isDIGIT_utf8_safe(p, e) \
2143 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2366ba44
KW
2144 _is_utf8_FOO(_CC_DIGIT, p, e))
2145#define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
da8c1a98 2146#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
dd1a3ba7 2147 _is_utf8_perl_idcont, p, e)
e5dcd934 2148
c11ff943
KW
2149/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2150 * IDFIRST is an alnum. See
8034715d 2151 * https://github.com/Perl/perl5/issues/10275 for more detail than you
f91dcd13
KW
2152 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2153 * This used to be not the XID version, but we decided to go with the more
2154 * modern Unicode definition */
da8c1a98
KW
2155#define isIDFIRST_utf8_safe(p, e) \
2156 _generic_func_utf8_safe(_CC_IDFIRST, \
dd1a3ba7 2157 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
da8c1a98 2158
2366ba44
KW
2159#define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2160#define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
da8c1a98 2161#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2366ba44 2162#define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
da8c1a98 2163#define isSPACE_utf8_safe(p, e) \
2366ba44
KW
2164 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2165#define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
da8c1a98 2166#define isVERTWS_utf8_safe(p, e) \
2366ba44 2167 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
da8c1a98 2168#define isWORDCHAR_utf8_safe(p, e) \
2366ba44 2169 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
da8c1a98
KW
2170#define isXDIGIT_utf8_safe(p, e) \
2171 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2172 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2173 ? (_force_out_malformed_utf8_message( \
2174 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2175 : is_XDIGIT_high(p)))
a0ed51b3 2176
059703b0
KW
2177#define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2178#define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2179#define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2180#define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2e8adce6 2181
567b353c 2182/* For internal core use only, subject to change */
059703b0
KW
2183#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2184#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2185#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2186#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
a1a5ec35
KW
2187
2188#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2189#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2190#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2191#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 2192
059703b0
KW
2193#define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2194#define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2195#define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2196#define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2197#define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2198#define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2199#define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2200#define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2201#define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2202#define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2203#define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2204#define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2205#define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2206#define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2207#define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2208#define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2209#define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
34aeb2e9 2210
da8c1a98
KW
2211/* For internal core Perl use only: the base macros for defining macros like
2212 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2213 * point in 'p' is within the 0-255 range, it uses locale rules from the
2214 * passed-in 'macro' parameter */
2215#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2216 (__ASSERT_(_utf8_safe_assert(p, e)) \
2217 (UTF8_IS_INVARIANT(*(p))) \
2218 ? macro(*(p)) \
2219 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2220 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2221 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2222 : (_force_out_malformed_utf8_message( \
2223 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2224 : above_latin1))
2225
2366ba44
KW
2226#define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2227 _generic_LC_utf8_safe(macro, p, e, \
2228 _is_utf8_FOO(classnum, p, e))
da8c1a98
KW
2229
2230#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2231 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2232
2366ba44 2233#define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2234 _generic_LC_utf8_safe(classnum, p, e, \
2235 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2236 ? (_force_out_malformed_utf8_message( \
2237 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2238 : above_latin1(p)))
2239
2240#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2366ba44 2241 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
da8c1a98
KW
2242 _CC_ALPHANUMERIC, p, e)
2243#define isALPHA_LC_utf8_safe(p, e) \
2366ba44 2244 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
da8c1a98
KW
2245#define isASCII_LC_utf8_safe(p, e) \
2246 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2247#define isBLANK_LC_utf8_safe(p, e) \
2366ba44 2248 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
da8c1a98
KW
2249#define isCNTRL_LC_utf8_safe(p, e) \
2250 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2251#define isDIGIT_LC_utf8_safe(p, e) \
2366ba44 2252 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
da8c1a98 2253#define isGRAPH_LC_utf8_safe(p, e) \
2366ba44 2254 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
da8c1a98
KW
2255#define isIDCONT_LC_utf8_safe(p, e) \
2256 _generic_LC_func_utf8_safe(isIDCONT_LC, \
dd1a3ba7 2257 _is_utf8_perl_idcont, p, e)
da8c1a98
KW
2258#define isIDFIRST_LC_utf8_safe(p, e) \
2259 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
dd1a3ba7 2260 _is_utf8_perl_idstart, p, e)
da8c1a98 2261#define isLOWER_LC_utf8_safe(p, e) \
2366ba44 2262 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
da8c1a98 2263#define isPRINT_LC_utf8_safe(p, e) \
2366ba44 2264 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
da8c1a98
KW
2265#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2266#define isPUNCT_LC_utf8_safe(p, e) \
2366ba44 2267 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
da8c1a98 2268#define isSPACE_LC_utf8_safe(p, e) \
2366ba44 2269 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
da8c1a98 2270#define isUPPER_LC_utf8_safe(p, e) \
2366ba44 2271 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
da8c1a98 2272#define isWORDCHAR_LC_utf8_safe(p, e) \
2366ba44 2273 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
da8c1a98 2274#define isXDIGIT_LC_utf8_safe(p, e) \
2366ba44 2275 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2276
fbc19f27
KW
2277/* Macros for backwards compatibility and for completeness when the ASCII and
2278 * Latin1 values are identical */
b7d90381
KW
2279#define isALPHAU(c) isALPHA_L1(c)
2280#define isDIGIT_L1(c) isDIGIT_A(c)
2281#define isOCTAL(c) isOCTAL_A(c)
2282#define isOCTAL_L1(c) isOCTAL_A(c)
2283#define isXDIGIT_L1(c) isXDIGIT_A(c)
2284#define isALNUM(c) isWORDCHAR(c)
a377c856 2285#define isALNUM_A(c) isALNUM(c)
b7d90381
KW
2286#define isALNUMU(c) isWORDCHAR_L1(c)
2287#define isALNUM_LC(c) isWORDCHAR_LC(c)
2288#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2289#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
059703b0 2290#define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
4c1d9526 2291#define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
059703b0 2292#define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
4c1d9526 2293#define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
b7d90381
KW
2294#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2295#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2296#define isALNUMC(c) isALPHANUMERIC(c)
2297#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2298#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2299#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
059703b0 2300#define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
4c1d9526
KW
2301#define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2302#define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
fbc19f27 2303
2bd1cbf6
KW
2304/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2305 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2306 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2307 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2308 * the outlier from the block that contains the other controls, just like
2309 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2310 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2311 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2312 * etc. */
2bd1cbf6 2313#ifndef EBCDIC
75763b3a 2314# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2315#else
75763b3a
KW
2316# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2317 ((isPRINT_A(c)) \
2318 ? (UNLIKELY((c) == '?') \
2319 ? QUESTION_MARK_CTRL \
2320 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2321 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2322 ? '?' \
2323 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2324#endif
bbce6d69 2325
dea28490
JJ
2326/* Line numbers are unsigned, 32 bits. */
2327typedef U32 line_t;
e5dcd934 2328#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2329
91152fc1
DG
2330/* Helpful alias for version prescan */
2331#define is_LAX_VERSION(a,b) \
2332 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2333
2334#define is_STRICT_VERSION(a,b) \
2335 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2336
2337#define BADVERSION(a,b,c) \
2338 if (b) { \
2339 *b = c; \
2340 } \
2341 return a;
8c52afec 2342
1ce77b7d
KW
2343/* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2344 * a-f) to its numeric value without using any branches. The input is
2345 * validated only by an assert() in DEBUGGING builds.
2346 *
2347 * It works by right shifting and isolating the bit that is 0 for the digits,
2348 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2349 * position, and then to the eights position. Both are added together to form
2350 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2351 * final four bits of the input to form the correct value. */
2352#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2353 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2354 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2355 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2356
2357/* The argument is a string pointer, which is advanced. */
2358#define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
95a59cab 2359
cb27eebd
KW
2360/* Converts a character known to represent an octal digit (0-7) to its numeric
2361 * value. The input is validated only by an assert() in DEBUGGING builds. In
2362 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2363#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2364
305b8651
KW
2365/* Efficiently returns a boolean as to if two native characters are equivalent
2366 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2367 * the ALPHA in the name is to remind you of that. This is asserted() in
2368 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2369 * works (on valid input) for both non- and UTF-8-encoded bytes.
2370 *
2371 * When one of the inputs is a compile-time constant and gets folded by the
2372 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2373 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2374 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2375 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2376 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2377 * just a single 0, in the bit position where the upper- and lowercase differ.
2378 * */
305b8651
KW
2379#define isALPHA_FOLD_EQ(c1, c2) \
2380 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2381 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2382#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2383
8e84507e 2384/*
ccfc67b7
JH
2385=head1 Memory Management
2386
a02a5408 2387=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2388The XSUB-writer's interface to the C C<malloc> function.
2389
596f7718 2390Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2391
c5008215
JC
2392In 5.9.3, Newx() and friends replace the older New() API, and drops
2393the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2394themselves. This aid has been superseded by a new build option,
d10b4965 2395PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2396there for use in XS modules supporting older perls.
2397
a02a5408 2398=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2399The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2400cast. See also C<L</Newx>>.
954c1994 2401
596f7718 2402Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2403
a02a5408 2404=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2405The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2406memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2407
596f7718 2408Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2409
954c1994
GS
2410=for apidoc Am|void|Renew|void* ptr|int nitems|type
2411The XSUB-writer's interface to the C C<realloc> function.
2412
596f7718 2413Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2414
954c1994
GS
2415=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2416The XSUB-writer's interface to the C C<realloc> function, with
2417cast.
2418
596f7718 2419Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2420
49b8b560 2421=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2422The XSUB-writer's interface to the C C<free> function.
2423
596f7718 2424This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2425
954c1994
GS
2426=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2427The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2428source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2429C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2430
e90e2364 2431=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2432Like C<Move> but returns C<dest>. Useful
72d33970 2433for encouraging compilers to tail-call
e90e2364
NC
2434optimise.
2435
954c1994
GS
2436=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2437The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2438source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2439C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2440
e90e2364
NC
2441=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2442
796b6530 2443Like C<Copy> but returns C<dest>. Useful
72d33970 2444for encouraging compilers to tail-call
e90e2364
NC
2445optimise.
2446
954c1994
GS
2447=for apidoc Am|void|Zero|void* dest|int nitems|type
2448
2449The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2450destination, C<nitems> is the number of items, and C<type> is the type.
2451
e90e2364
NC
2452=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2453
72d33970
FC
2454Like C<Zero> but returns dest. Useful
2455for encouraging compilers to tail-call
e90e2364
NC
2456optimise.
2457
da5d8dbb 2458=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2459This is an architecture-independent macro to copy one structure to another.
954c1994 2460
7e337ee0
JH
2461=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2462
2463Fill up memory with a byte pattern (a byte repeated over and over
2464again) that hopefully catches attempts to access uninitialized memory.
2465
2466=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2467
2468PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2469
1c12ffb4 2470=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2471
2472PoisonWith(0xEF) for catching access to freed memory.
2473
9965345d
JH
2474=for apidoc Am|void|Poison|void* dest|int nitems|type
2475
7e337ee0 2476PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2477
2478=cut */
954c1994 2479
561b68a9
SH
2480/* Maintained for backwards-compatibility only. Use newSV() instead. */
2481#ifndef PERL_CORE
ff06c60c 2482#define NEWSV(x,len) newSV(len)
561b68a9 2483#endif
ff06c60c 2484
b7112dce 2485#define MEM_SIZE_MAX ((MEM_SIZE)-1)
19a94d75 2486
a500027b 2487#define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
e6bdf523 2488
27d5b266 2489#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2490
2491/* This expression will be constant-folded at compile time. It checks
2492 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2493 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2494 * As well as avoiding the need for a run-time check in some cases, it's
2495 * designed to avoid compiler warnings like:
2496 * comparison is always false due to limited range of data type
73e8ff00
DM
2497 * It's mathematically equivalent to
2498 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2499 */
2500
2501# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
445198b9
LM
2502 ( sizeof(MEM_SIZE) < sizeof(n) \
2503 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 2504
88f9f128 2505/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2506 * compiler warnings. We *want* to write the expression as
2507 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2508 * (for some compile-time constant C), but even when the LHS
2509 * constant-folds to false at compile-time, g++ insists on emitting
2510 * warnings about the RHS (e.g. "comparison is always false"), so instead
2511 * we write it as
e6bdf523 2512 *
d98e5cde 2513 * (cond ? n : X) > C
88f9f128 2514 *
d98e5cde
DM
2515 * where X is a constant with X > C always false. Choosing a value for X
2516 * is tricky. If 0, some compilers will complain about 0 > C always being
2517 * false; if 1, Coverity complains when n happens to be the constant value
2518 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2519 * for X and hope that nothing else whines.
e6bdf523
DM
2520 */
2521
2522# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2523 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2524 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2525
2526# define MEM_WRAP_CHECK(n,t) \
2527 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2528 && (croak_memory_wrap(),0))
2529
2530# define MEM_WRAP_CHECK_1(n,t,a) \
2531 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2532 && (Perl_croak_nocontext("%s",(a)),0))
2533
814eedc8
DD
2534/* "a" arg must be a string literal */
2535# define MEM_WRAP_CHECK_s(n,t,a) \
2536 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2537 && (Perl_croak_nocontext("" a ""),0))
2538
8b44ba4c 2539#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2540
a500027b 2541#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
2542#else
2543
410319be
NC
2544#define MEM_WRAP_CHECK(n,t)
2545#define MEM_WRAP_CHECK_1(n,t,a)
814eedc8 2546#define MEM_WRAP_CHECK_s(n,t,a)
8b44ba4c
NC
2547#define MEM_WRAP_CHECK_(n,t)
2548
a500027b 2549#define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
27d5b266 2550
1936d2a7 2551#endif
8b44ba4c 2552
fe4f188c 2553#ifdef PERL_MEM_LOG
46c6c7e2 2554/*
9f653bb5 2555 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2556 * go through functions, which are handy for debugging breakpoints, but
2557 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2558 * line number, and C function name if available) passed in. This info can
2559 * then be used for logging the calls, for which one gets a sample
73d1d973 2560 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2561 *
46c6c7e2 2562 * Known problems:
94e892a6 2563 * - not all memory allocs get logged, only those
46c6c7e2 2564 * that go through Newx() and derivatives (while all
94e892a6 2565 * Safefrees do get logged)
46c6c7e2
JH
2566 * - __FILE__ and __LINE__ do not work everywhere
2567 * - __func__ or __FUNCTION__ even less so
2568 * - I think more goes on after the perlio frees but
2569 * the thing is that STDERR gets closed (as do all
2570 * the file descriptors)
2571 * - no deeper calling stack than the caller of the Newx()
2572 * or the kind, but do I look like a C reflection/introspection
2573 * utility to you?
2574 * - the function prototypes for the logging functions
2575 * probably should maybe be somewhere else than handy.h
2576 * - one could consider inlining (macrofying) the logging
2577 * for speed, but I am too lazy
2578 * - one could imagine recording the allocations in a hash,
2579 * (keyed by the allocation address?), and maintain that
2580 * through reallocs and frees, but how to do that without
2581 * any News() happening...?
73d1d973 2582 * - lots of -Ddefines to get useful/controllable output
b953482e 2583 * - lots of ENV reads
46c6c7e2
JH
2584 */
2585
0b0ab801 2586# ifdef PERL_CORE
73d1d973 2587# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2588enum mem_log_type {
2589 MLT_ALLOC,
2590 MLT_REALLOC,
d7a2c63c
MHM
2591 MLT_FREE,
2592 MLT_NEW_SV,
2593 MLT_DEL_SV
0b0ab801
MHM
2594};
2595# endif
12754f92 2596# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2597void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2598void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2599# endif
0b0ab801
MHM
2600# endif
2601
fe4f188c
JH
2602#endif
2603
2604#ifdef PERL_MEM_LOG
d1401ee9
MHM
2605#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2606#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2607#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2608#endif
2609
2610#ifndef MEM_LOG_ALLOC
2611#define MEM_LOG_ALLOC(n,t,a) (a)
2612#endif
2613#ifndef MEM_LOG_REALLOC
2614#define MEM_LOG_REALLOC(n,t,v,a) (a)
2615#endif
2616#ifndef MEM_LOG_FREE
2617#define MEM_LOG_FREE(a) (a)
2618#endif
2619
d1401ee9
MHM
2620#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2621#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2622#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2623
2624#ifndef PERL_CORE
a02a5408
JC
2625/* pre 5.9.x compatibility */
2626#define New(x,v,n,t) Newx(v,n,t)
2627#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2628#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2629#endif
a02a5408 2630
ff68c719 2631#define Renew(v,n,t) \
d1401ee9 2632 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2633#define Renewc(v,n,t,c) \
d1401ee9 2634 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2635
2636#ifdef PERL_POISON
2637#define Safefree(d) \
06c0cc96 2638 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2639#else
fe4f188c 2640#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2641#endif
55497cff 2642
dbb57106
YO
2643/* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2644 * as it handles cases like constant string arguments without throwing warnings *
2645 * the cast is required, as is the inequality check, to avoid warnings */
45908e4d 2646#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2647
45908e4d
YO
2648
2649#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)))
2650#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)))
2651#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2652
bdd1531d 2653/* Like above, but returns a pointer to 'd' */
45908e4d
YO
2654#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)))
2655#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 2656#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2657
7e337ee0
JH
2658#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2659#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2660#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2661#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2662
caa674f3
DD
2663#ifdef PERL_POISON
2664# define PERL_POISON_EXPR(x) x
2665#else
2666# define PERL_POISON_EXPR(x)
2667#endif
2668
ff68c719 2669#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2670
1b7e2294
KW
2671/*
2672=head1 Handy Values
2673
2674=for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2675
2676Returns the number of elements in the input C array (so you want your
2677zero-based indices to be less than but not equal to).
2678
2679=for apidoc Am|void *|C_ARRAY_END|void *a
2680
2681Returns a pointer to one element past the final element of the input C array.
2682
2683=cut
2684
2685C_ARRAY_END is one past the last: half-open/half-closed range, not
2686last-inclusive range.
2687*/
622913ab 2688#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2689#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2690
2cc61e15
DD
2691#ifdef NEED_VA_COPY
2692# ifdef va_copy
2693# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2694# elif defined(__va_copy)
2695# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2696# else
07798b17 2697# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2698# endif
2699#endif
2700
472d47bc
SB
2701/* convenience debug macros */
2702#ifdef USE_ITHREADS
2703#define pTHX_FORMAT "Perl interpreter: 0x%p"
2704#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2705#define pTHX_VALUE_ (void *)my_perl,
2706#define pTHX_VALUE (void *)my_perl
2707#define pTHX__VALUE_ ,(void *)my_perl,
2708#define pTHX__VALUE ,(void *)my_perl
472d47bc 2709#else
3609ea0d 2710#define pTHX_FORMAT
472d47bc 2711#define pTHX__FORMAT
3609ea0d 2712#define pTHX_VALUE_
472d47bc 2713#define pTHX_VALUE
3609ea0d 2714#define pTHX__VALUE_
472d47bc
SB
2715#define pTHX__VALUE
2716#endif /* USE_ITHREADS */
3609ea0d 2717
2acdbac1
NC
2718/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2719 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2720 CPAN::Unpack show any users outside the core. */
2721#ifdef PERL_CORE
dc6e8de0
A
2722# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2723 "Use of " s " is deprecated")
c9680906
A
2724# define deprecate_disappears_in(when,message) \
2725 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2726 message ", and will disappear in Perl " when)
ac641426
A
2727# define deprecate_fatal_in(when,message) \
2728 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2729 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2730#endif
2731
dfff4baf
BF
2732/* Internal macros to deal with gids and uids */
2733#ifdef PERL_CORE
2734
2735# if Uid_t_size > IVSIZE
2736# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2737# define SvUID(sv) SvNV(sv)
07798b17
AC
2738# elif Uid_t_sign <= 0
2739# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2740# define SvUID(sv) SvIV(sv)
dfff4baf 2741# else
07798b17
AC
2742# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2743# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2744# endif /* Uid_t_size */
2745
2746# if Gid_t_size > IVSIZE
2747# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2748# define SvGID(sv) SvNV(sv)
07798b17
AC
2749# elif Gid_t_sign <= 0
2750# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2751# define SvGID(sv) SvIV(sv)
dfff4baf 2752# else
07798b17
AC
2753# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2754# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2755# endif /* Gid_t_size */
2756
2757#endif
2758
6a5bc5ac 2759#endif /* PERL_HANDY_H_ */
9d745869 2760
e9a8c099 2761/*
14d04a33 2762 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2763 */