This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
handy.h: Create nBIT_UMAX() 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
1381ccb1
KW
279/* The largest unsigned number that will fit into n bits */
280#define nBIT_UMAX(n) nBIT_MASK(n)
281
8d9433eb
KW
282/*
283=for apidoc Am|void|__ASSERT_|bool expr
284
285This is a helper macro to avoid preprocessor issues, replaced by nothing
286unless under DEBUGGING, where it expands to an assert of its argument,
287followed by a comma (hence the comma operator). If we just used a straight
288assert(), we would get a comma with nothing before it when not DEBUGGING.
289
290=cut
291
292We also use empty definition under Coverity since the __ASSERT__
293checks often check for things that Really Cannot Happen, and Coverity
294detects that and gets all excited. */
3e94db23 295
e7ae132e
KW
296#if defined(DEBUGGING) && !defined(__COVERITY__) \
297 && ! defined(PERL_SMALL_MACRO_BUFFER)
0f092d08
KW
298# define __ASSERT_(statement) assert(statement),
299#else
300# define __ASSERT_(statement)
301#endif
302
3fe05580 303/*
a4ee4fb5 304=head1 SV Manipulation Functions
3fe05580 305
3bb9fd01 306=for apidoc Ama|SV*|newSVpvs|"literal string"
1568d13a 307Like C<newSVpvn>, but takes a literal string instead of a
30a15352 308string/length pair.
3fe05580 309
3bb9fd01 310=for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
1568d13a 311Like C<newSVpvn_flags>, but takes a literal string instead of
30a15352 312a string/length pair.
84bafc02 313
3bb9fd01 314=for apidoc Ama|SV*|newSVpvs_share|"literal string"
1568d13a 315Like C<newSVpvn_share>, but takes a literal string instead of
30a15352 316a string/length pair and omits the hash parameter.
3fe05580 317
3bb9fd01 318=for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
1568d13a 319Like C<sv_catpvn_flags>, but takes a literal string instead
30a15352 320of a string/length pair.
9dcc53ea 321
3bb9fd01 322=for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
1568d13a 323Like C<sv_catpvn_nomg>, but takes a literal string instead of
0c395ea5 324a string/length pair.
9dcc53ea 325
3bb9fd01 326=for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
1568d13a 327Like C<sv_catpvn>, but takes a literal string instead of a
0c395ea5 328string/length pair.
3fe05580 329
3bb9fd01 330=for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
1568d13a 331Like C<sv_catpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
332string/length pair.
333
3bb9fd01 334=for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
1568d13a 335Like C<sv_setpvn>, but takes a literal string instead of a
0c395ea5 336string/length pair.
3fe05580 337
3bb9fd01 338=for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
1568d13a 339Like C<sv_setpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
340string/length pair.
341
3bb9fd01 342=for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
1568d13a 343Like C<sv_setref_pvn>, but takes a literal string instead of
0c395ea5 344a string/length pair.
9dcc53ea 345
3fe05580
MHM
346=head1 Memory Management
347
3bb9fd01 348=for apidoc Ama|char*|savepvs|"literal string"
1568d13a 349Like C<savepvn>, but takes a literal string instead of a
30a15352 350string/length pair.
3fe05580 351
3bb9fd01 352=for apidoc Ama|char*|savesharedpvs|"literal string"
9dcc53ea
Z
353A version of C<savepvs()> which allocates the duplicate string in memory
354which is shared between threads.
355
3fe05580
MHM
356=head1 GV Functions
357
3bb9fd01 358=for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
1568d13a 359Like C<gv_stashpvn>, but takes a literal string instead of a
0c395ea5 360string/length pair.
3fe05580
MHM
361
362=head1 Hash Manipulation Functions
363
3bb9fd01 364=for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
1568d13a 365Like C<hv_fetch>, but takes a literal string instead of a
0c395ea5 366string/length pair.
3fe05580 367
3bb9fd01 368=for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
1568d13a 369Like C<hv_store>, but takes a literal string instead of a
0c395ea5 370string/length pair
3fe05580
MHM
371and omits the hash parameter.
372
510966aa
Z
373=head1 Lexer interface
374
3bb9fd01 375=for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
510966aa 376
1568d13a 377Like L</lex_stuff_pvn>, but takes a literal string instead of
0c395ea5 378a string/length pair.
510966aa 379
3fe05580
MHM
380=cut
381*/
382
a34e53fc
KW
383/*
384=head1 Handy Values
2efa8cc7 385
a34e53fc
KW
386=for apidoc Amu|pair|STR_WITH_LEN|"literal string"
387
388Returns two comma separated tokens of the input literal string, and its length.
389This is convenience macro which helps out in some API calls.
390Note that it can't be used as an argument to macros or functions that under
391some configurations might be macros, which means that it requires the full
392Perl_xxx(aTHX_ ...) form for any API calls where it's used.
393
394=cut
395*/
396
a34e53fc 397#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
ba3a79e7
GA
398
399/* STR_WITH_LEN() shortcuts */
400#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
401#define newSVpvs_flags(str,flags) \
402 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 403#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
404#define sv_catpvs_flags(sv, str, flags) \
405 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
406#define sv_catpvs_nomg(sv, str) \
407 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
408#define sv_catpvs(sv, str) \
409 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
410#define sv_catpvs_mg(sv, str) \
411 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 412#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
413#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
414#define sv_setref_pvs(rv, classname, str) \
415 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 416#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
417#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
418#define gv_stashpvs(str, create) \
419 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
420#define gv_fetchpvs(namebeg, add, sv_type) \
421 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
422#define gv_fetchpvn(namebeg, len, add, sv_type) \
423 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
424#define sv_catxmlpvs(dsv, str, utf8) \
425 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 426
ba3a79e7 427
510966aa
Z
428#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
429
b96d8cd9
NC
430#define get_cvs(str, flags) \
431 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 432
9b6e9510
KW
433/* internal helpers */
434#define PERL_RVS_TO_DECIMAL_(r,v,s) (((r)*1000000)+((v)*1000)+(s))
435#define PERL_DECIMAL_VERSION_ \
436 PERL_RVS_TO_DECIMAL_(PERL_REVISION, PERL_VERSION, PERL_SUBVERSION)
437
438/*
439=for apidoc AmR|bool|PERL_VERSION_EQ|const int r|const int v|const int s
440
441Returns whether or not the perl currently executing has the specified
442relationship to the perl given by the parameters. For example,
443
444 #if PERL_VERSION_GT(5,24,2)
445 code that will only be compiled on perls after v5.24.2
446 #else
447 fallback code
448 #endif
449
450Note that this is usable in making compile-time decisions
451
452The possible comparisons are C<PERL_VERSION_EQ>, C<PERL_VERSION_NE>,
453C<PERL_VERSION_GE>, C<PERL_VERSION_GT>, C<PERL_VERSION_LE>, and
454C<PERL_VERSION_LT>.
455
456=for apidoc AmRh|bool|PERL_VERSION_NE|const int r|const int v|const int s
457=for apidoc AmRh|bool|PERL_VERSION_GE|const int r|const int v|const int s
458=for apidoc AmRh|bool|PERL_VERSION_GT|const int r|const int v|const int s
459=for apidoc AmRh|bool|PERL_VERSION_LE|const int r|const int v|const int s
460=for apidoc AmRh|bool|PERL_VERSION_LT|const int r|const int v|const int s
461
462=cut
463*/
464
465# define PERL_VERSION_EQ(r,v,s) (PERL_DECIMAL_VERSION_ == PERL_RVS_TO_DECIMAL_(r,v,s))
466# define PERL_VERSION_NE(r,v,s) (! PERL_VERSION_EQ(r,v,s))
467# define PERL_VERSION_LT(r,v,s) (PERL_DECIMAL_VERSION_ < PERL_RVS_TO_DECIMAL_(r,v,s))
468# define PERL_VERSION_LE(r,v,s) (PERL_DECIMAL_VERSION_ <= PERL_RVS_TO_DECIMAL_(r,v,s))
469# define PERL_VERSION_GT(r,v,s) (! PERL_VERSION_LE(r,v,s))
470# define PERL_VERSION_GE(r,v,s) (! PERL_VERSION_LT(r,v,s))
471
954c1994 472/*
ccfc67b7
JH
473=head1 Miscellaneous Functions
474
954c1994 475=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
476Test two C<NUL>-terminated strings to see if they are different. Returns true
477or false.
954c1994
GS
478
479=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
480Test two C<NUL>-terminated strings to see if they are equal. Returns true or
481false.
954c1994
GS
482
483=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
484Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
485second, C<s2>. Returns true or false.
954c1994
GS
486
487=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
488Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
489equal to the second, C<s2>. Returns true or false.
954c1994
GS
490
491=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
492Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
493the second, C<s2>. Returns true or false.
954c1994
GS
494
495=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
496Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
497or equal to the second, C<s2>. Returns true or false.
954c1994
GS
498
499=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
500Test two C<NUL>-terminated strings to see if they are different. The C<len>
501parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
502wrapper for C<strncmp>).
503
504=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
505Test two C<NUL>-terminated strings to see if they are equal. The C<len>
506parameter indicates the number of bytes to compare. Returns true or false. (A
507wrapper for C<strncmp>).
954c1994 508
bd18bd40
KW
509=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
510Test two buffers (which may contain embedded C<NUL> characters, to see if they
511are equal. The C<len> parameter indicates the number of bytes to compare.
512Returns zero if equal, or non-zero if non-equal.
513
3bb9fd01 514=for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
515Like L</memEQ>, but the second string is a literal enclosed in double quotes,
516C<l1> gives the number of bytes in C<s1>.
517Returns zero if equal, or non-zero if non-equal.
518
bd18bd40
KW
519=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
520Test two buffers (which may contain embedded C<NUL> characters, to see if they
521are not equal. The C<len> parameter indicates the number of bytes to compare.
522Returns zero if non-equal, or non-zero if equal.
523
3bb9fd01 524=for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
525Like L</memNE>, but the second string is a literal enclosed in double quotes,
526C<l1> gives the number of bytes in C<s1>.
527Returns zero if non-equal, or zero if non-equal.
528
4aada8b9
KW
529=for apidoc Am|bool|memCHRs|"list"|char c
530Returns the position of the first occurence of the byte C<c> in the literal
531string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
532treated as unsigned char. Thus this macro can be used to determine if C<c> is
533in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
534is C<NUL> (and the set doesn't include C<NUL>).
535
954c1994 536=cut
fc169e00
KW
537
538New macros should use the following conventions for their names (which are
539based on the underlying C library functions):
540
541 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
542
543 Each has two main parameters, string-like operands that are compared
544 against each other, as specified by the macro name. Some macros may
545 additionally have one or potentially even two length parameters. If a length
546 parameter applies to both string parameters, it will be positioned third;
547 otherwise any length parameter immediately follows the string parameter it
548 applies to.
549
550 If the prefix to the name is 'str', the string parameter is a pointer to a C
551 language string. Such a string does not contain embedded NUL bytes; its
552 length may be unknown, but can be calculated by C<strlen()>, since it is
553 terminated by a NUL, which isn't included in its length.
554
a3815e44 555 The optional 'n' following 'str' means that there is a third parameter,
fc169e00
KW
556 giving the maximum number of bytes to look at in each string. Even if both
557 strings are longer than the length parameter, those extra bytes will be
558 unexamined.
559
560 The 's' suffix means that the 2nd byte string parameter is a literal C
561 double-quoted string. Its length will automatically be calculated by the
562 macro, so no length parameter will ever be needed for it.
563
564 If the prefix is 'mem', the string parameters don't have to be C strings;
565 they may contain embedded NUL bytes, do not necessarily have a terminating
566 NUL, and their lengths can be known only through other means, which in
567 practice are additional parameter(s) passed to the function. All 'mem'
568 functions have at least one length parameter. Barring any 'l' or 's' suffix,
569 there is a single length parameter, in position 3, which applies to both
570 string parameters. The 's' suffix means, as described above, that the 2nd
571 string is a literal double-quoted C string (hence its length is calculated by
572 the macro, and the length parameter to the function applies just to the first
573 string parameter, and hence is positioned just after it). An 'l' suffix
574 means that the 2nd string parameter has its own length parameter, and the
575 signature will look like memFOOl(s1, l1, s2, l2).
576
577 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
578 substring of the 1st string. 'P' if present indicates that the substring
579 must be a "proper" one in tha mathematical sense that the first one must be
580 strictly larger than the 2nd.
581
954c1994
GS
582*/
583
62946e08 584
75400963
KW
585#define strNE(s1,s2) (strcmp(s1,s2) != 0)
586#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
587#define strLT(s1,s2) (strcmp(s1,s2) < 0)
588#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
589#define strGT(s1,s2) (strcmp(s1,s2) > 0)
590#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 591
75400963
KW
592#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
593#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 594
9d3980bc
KW
595#define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
596#define memNE(s1,s2,l) (! memEQ(s1,s2,l))
36477c24 597
085b7534 598/* memEQ and memNE where second comparand is a string constant */
568a785a 599#define memEQs(s1, l, s2) \
777fa2cb 600 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 601#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 602
fdbb9a7c
KW
603/* Keep these private until we decide it was a good idea */
604#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
605
606#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
607
bdb7e3f0 608#define memBEGINs(s1, l, s2) \
30a6480c 609 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 610 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158 611#define memBEGINPs(s1, l, s2) \
30a6480c 612 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
de627158 613 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0 614#define memENDs(s1, l, s2) \
30a6480c 615 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 616 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424 617#define memENDPs(s1, l, s2) \
30a6480c 618 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
b80f8424 619 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 620#endif /* End of making macros private */
bdb7e3f0 621
062b6850
KW
622#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
623#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
624#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
625#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
626
4aada8b9
KW
627#define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
628
bbce6d69
PP
629/*
630 * Character classes.
631 *
632 * Unfortunately, the introduction of locales means that we
633 * can't trust isupper(), etc. to tell the truth. And when
634 * it comes to /\w+/ with tainting enabled, we *must* be able
635 * to trust our character classes.
636 *
637 * Therefore, the default tests in the text of Perl will be
638 * independent of locale. Any code that wants to depend on
639 * the current locale will use the tests that begin with "lc".
640 */
641
2304df62
AD
642#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
643# ifndef CTYPE256
644# define CTYPE256
645# endif
646#endif
647
954c1994 648/*
ccfc67b7 649
dcccc8ff 650=head1 Character classification
243effed
KW
651This section is about functions (really macros) that classify characters
652into types, such as punctuation versus alphabetic, etc. Most of these are
653analogous to regular expression character classes. (See
654L<perlrecharclass/POSIX Character Classes>.) There are several variants for
655each class. (Not all macros have all variants; each item below lists the
656ones valid for it.) None are affected by C<use bytes>, and only the ones
657with C<LC> in the name are affected by the current locale.
658
d713f9d9
KW
659The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
660treating it as a code point, and returns a boolean as to whether or not the
661character represented by it is (or on non-ASCII platforms, corresponds to) an
6aff1f14
KW
662ASCII character in the named class based on platform, Unicode, and Perl rules.
663If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 664
c98722a4 665Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
666with no suffix C<"_A">. This variant is used to emphasize by its name that
667only ASCII-range characters can return TRUE.
4b9734bf 668
d60679e1 669Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
670onto the platform. That is, the code points that are ASCII are unaffected,
671since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
672as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
673true when called with the code point 0xDF, which is a word character in both
4650c663 674ASCII and EBCDIC (though it represents different characters in each).
d713f9d9
KW
675If the input is a number that doesn't fit in an octet, FALSE is returned.
676(Perl's documentation uses a colloquial definition of Latin-1, to include all
677code points below 256.)
243effed 678
d713f9d9
KW
679Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
680inputs below 256, but if the code point is larger than 255, Unicode rules are
681used to determine if it is in the character class. For example,
d0da05db 682C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 683WITH MACRON in Unicode, and is a word character.
243effed 684
059703b0
KW
685Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
686but are used for UTF-8 encoded strings. The two forms are different names for
687the same thing. Each call to one of these classifies the first character of
688the string starting at C<p>. The second parameter, C<e>, points to anywhere in
689the string beyond the first character, up to one byte past the end of the
690entire string. Although both variants are identical, the suffix C<_safe> in
691one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
692provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
693C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
694some way, the program may croak, or the function may return FALSE, at the
695discretion of the implementation, and subject to change in future releases.
243effed 696
d713f9d9
KW
697Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
698but the result is based on the current locale, which is what C<LC> in the name
699stands for. If Perl can determine that the current locale is a UTF-8 locale,
700it uses the published Unicode rules; otherwise, it uses the C library function
701that gives the named classification. For example, C<isDIGIT_LC()> when not in
702a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
703returned if the input won't fit into an octet. On some platforms where the C
704library function is known to be defective, Perl changes its result to follow
705the POSIX standard's rules.
243effed 706
d713f9d9
KW
707Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
708than 256, but for larger ones it returns the Unicode classification of the code
709point.
243effed 710
059703b0
KW
711Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
712C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
713are different names for the same thing. Each call to one of these classifies
714the first character of the string starting at C<p>. The second parameter,
715C<e>, points to anywhere in the string beyond the first character, up to one
716byte past the end of the entire string. Although both variants are identical,
717the suffix C<_safe> in one name emphasizes that it will not attempt to read
718beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
719is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
720character is malformed in some way, the program may croak, or the function may
721return FALSE, at the discretion of the implementation, and subject to change in
722future releases.
ccfc67b7 723
d713f9d9
KW
724=for apidoc Am|bool|isALPHA|int ch
725Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
726analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
727See the L<top of this section|/Character classification> for an explanation of
728variants
059703b0
KW
729C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>,
730C<isALPHA_utf8_safe>, C<isALPHA_LC>, C<isALPHA_LC_uvchr>, C<isALPHA_LC_utf8>,
731and C<isALPHA_LC_utf8_safe>.
8a58bdcf 732
f16858ed
KW
733=cut
734
735Here and below, we add the protoypes of these macros for downstream programs
736that would be interested in them, such as Devel::PPPort
737
738=for apidoc Amh|bool|isALPHA_A|int ch
739=for apidoc Amh|bool|isALPHA_L1|int ch
740=for apidoc Amh|bool|isALPHA_uvchr|int ch
741=for apidoc Amh|bool|isALPHA_utf8_safe|U8 * s|U8 * end
d23c7e08 742=for apidoc Amh|bool|isALPHA_utf8|U8 * s|U8 * end
f16858ed
KW
743=for apidoc Amh|bool|isALPHA_LC|int ch
744=for apidoc Amh|bool|isALPHA_LC_uvchr|int ch
745=for apidoc Amh|bool|isALPHA_LC_utf8_safe|U8 * s| U8 *end
746
d713f9d9
KW
747=for apidoc Am|bool|isALPHANUMERIC|int ch
748Returns a boolean indicating whether the specified character is one of
749C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
750See the L<top of this section|/Character classification> for an explanation of
751variants
d0da05db 752C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
059703b0
KW
753C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>,
754C<isALPHANUMERIC_LC_uvchr>, C<isALPHANUMERIC_LC_utf8>, and
755C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 756
255b632a
KW
757A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
758this corresponds to the C language alphanumeric definition). Also
759there are the variants
760C<isALNUMC_A>, C<isALNUMC_L1>
761C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
762
f16858ed
KW
763=for apidoc Amh|bool|isALPHANUMERIC_A|int ch
764=for apidoc Amh|bool|isALPHANUMERIC_L1|int ch
765=for apidoc Amh|bool|isALPHANUMERIC_uvchr|int ch
766=for apidoc Amh|bool|isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
d23c7e08 767=for apidoc Amh|bool|isALPHANUMERIC_utf8|U8 * s|U8 * end
f16858ed
KW
768=for apidoc Amh|bool|isALPHANUMERIC_LC|int ch
769=for apidoc Amh|bool|isALPHANUMERIC_LC_uvchr|int ch
770=for apidoc Amh|bool|isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
771=for apidoc Amh|bool|isALNUMC|int ch
772=for apidoc Amh|bool|isALNUMC_A|int ch
773=for apidoc Amh|bool|isALNUMC_L1|int ch
774=for apidoc Amh|bool|isALNUMC_LC|int ch
775=for apidoc Amh|bool|isALNUMC_LC_uvchr|int ch
776
d713f9d9 777=for apidoc Am|bool|isASCII|int ch
8a58bdcf 778Returns a boolean indicating whether the specified character is one of the 128
243effed 779characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 780On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
781character corresponds to an ASCII character. Variants C<isASCII_A()> and
782C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
783See the L<top of this section|/Character classification> for an explanation of
784variants
059703b0
KW
785C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_utf8_safe>, C<isASCII_LC>,
786C<isASCII_LC_uvchr>, C<isASCII_LC_utf8>, and C<isASCII_LC_utf8_safe>.
787Note, however, that some platforms do not have the C library routine
788C<isascii()>. In these cases, the variants whose names contain C<LC> are the
789same as the corresponding ones without.
243effed 790
f16858ed
KW
791=for apidoc Amh|bool|isASCII_A|int ch
792=for apidoc Amh|bool|isASCII_L1|int ch
793=for apidoc Amh|bool|isASCII_uvchr|int ch
794=for apidoc Amh|bool|isASCII_utf8_safe|U8 * s|U8 * end
d23c7e08 795=for apidoc Amh|bool|isASCII_utf8|U8 * s|U8 * end
f16858ed
KW
796=for apidoc Amh|bool|isASCII_LC|int ch
797=for apidoc Amh|bool|isASCII_LC_uvchr|int ch
798=for apidoc Amh|bool|isASCII_LC_utf8_safe|U8 * s| U8 *end
799
d98532ea
KW
800Also note, that because all ASCII characters are UTF-8 invariant (meaning they
801have the exact same representation (always a single byte) whether encoded in
802UTF-8 or not), C<isASCII> will give the correct results when called with any
059703b0
KW
803byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
804C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
d98532ea 805
243effed
KW
806=for apidoc Am|bool|isBLANK|char ch
807Returns a boolean indicating whether the specified character is a
6aff1f14 808character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
809See the L<top of this section|/Character classification> for an explanation of
810variants
059703b0
KW
811C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>,
812C<isBLANK_utf8_safe>, C<isBLANK_LC>, C<isBLANK_LC_uvchr>, C<isBLANK_LC_utf8>,
813and C<isBLANK_LC_utf8_safe>. Note,
da8c1a98
KW
814however, that some platforms do not have the C library routine
815C<isblank()>. In these cases, the variants whose names contain C<LC> are
816the same as the corresponding ones without.
243effed 817
f16858ed
KW
818=for apidoc Amh|bool|isBLANK_A|int ch
819=for apidoc Amh|bool|isBLANK_L1|int ch
820=for apidoc Amh|bool|isBLANK_uvchr|int ch
821=for apidoc Amh|bool|isBLANK_utf8_safe|U8 * s|U8 * end
d23c7e08 822=for apidoc Amh|bool|isBLANK_utf8|U8 * s|U8 * end
f16858ed
KW
823=for apidoc Amh|bool|isBLANK_LC|int ch
824=for apidoc Amh|bool|isBLANK_LC_uvchr|int ch
825=for apidoc Amh|bool|isBLANK_LC_utf8_safe|U8 * s| U8 *end
826
243effed
KW
827=for apidoc Am|bool|isCNTRL|char ch
828Returns a boolean indicating whether the specified character is a
6aff1f14 829control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
830See the L<top of this section|/Character classification> for an explanation of
831variants
059703b0
KW
832C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>,
833C<isCNTRL_utf8_safe>, C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, C<isCNTRL_LC_utf8>
834and C<isCNTRL_LC_utf8_safe>. On EBCDIC
da8c1a98 835platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 836
f16858ed
KW
837=for apidoc Amh|bool|isCNTRL_A|int ch
838=for apidoc Amh|bool|isCNTRL_L1|int ch
839=for apidoc Amh|bool|isCNTRL_uvchr|int ch
840=for apidoc Amh|bool|isCNTRL_utf8_safe|U8 * s|U8 * end
d23c7e08 841=for apidoc Amh|bool|isCNTRL_utf8|U8 * s|U8 * end
f16858ed
KW
842=for apidoc Amh|bool|isCNTRL_LC|int ch
843=for apidoc Amh|bool|isCNTRL_LC_uvchr|int ch
844=for apidoc Amh|bool|isCNTRL_LC_utf8_safe|U8 * s| U8 *end
845
954c1994 846=for apidoc Am|bool|isDIGIT|char ch
2787a470 847Returns a boolean indicating whether the specified character is a
6aff1f14 848digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 849Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
850See the L<top of this section|/Character classification> for an explanation of
851variants
059703b0
KW
852C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>,
853C<isDIGIT_LC_uvchr>, C<isDIGIT_LC_utf8>, and C<isDIGIT_LC_utf8_safe>.
243effed 854
f16858ed
KW
855=for apidoc Amh|bool|isDIGIT_A|int ch
856=for apidoc Amh|bool|isDIGIT_L1|int ch
857=for apidoc Amh|bool|isDIGIT_uvchr|int ch
858=for apidoc Amh|bool|isDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 859=for apidoc Amh|bool|isDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
860=for apidoc Amh|bool|isDIGIT_LC|int ch
861=for apidoc Amh|bool|isDIGIT_LC_uvchr|int ch
862=for apidoc Amh|bool|isDIGIT_LC_utf8_safe|U8 * s| U8 *end
863
243effed
KW
864=for apidoc Am|bool|isGRAPH|char ch
865Returns a boolean indicating whether the specified character is a
6aff1f14 866graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 867See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
868variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>,
869C<isGRAPH_utf8_safe>, C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>,
870C<isGRAPH_LC_utf8_safe>, and C<isGRAPH_LC_utf8_safe>.
954c1994 871
f16858ed
KW
872=for apidoc Amh|bool|isGRAPH_A|int ch
873=for apidoc Amh|bool|isGRAPH_L1|int ch
874=for apidoc Amh|bool|isGRAPH_uvchr|int ch
875=for apidoc Amh|bool|isGRAPH_utf8_safe|U8 * s|U8 * end
d23c7e08 876=for apidoc Amh|bool|isGRAPH_utf8|U8 * s|U8 * end
f16858ed
KW
877=for apidoc Amh|bool|isGRAPH_LC|int ch
878=for apidoc Amh|bool|isGRAPH_LC_uvchr|int ch
879=for apidoc Amh|bool|isGRAPH_LC_utf8_safe|U8 * s| U8 *end
880
0c82b6df 881=for apidoc Am|bool|isLOWER|char ch
2787a470 882Returns a boolean indicating whether the specified character is a
6aff1f14 883lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
884See the L<top of this section|/Character classification> for an explanation of
885variants
059703b0
KW
886C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>,
887C<isLOWER_utf8_safe>, C<isLOWER_LC>, C<isLOWER_LC_uvchr>, C<isLOWER_LC_utf8>,
888and C<isLOWER_LC_utf8_safe>.
0c82b6df 889
f16858ed
KW
890=for apidoc Amh|bool|isLOWER_A|int ch
891=for apidoc Amh|bool|isLOWER_L1|int ch
892=for apidoc Amh|bool|isLOWER_uvchr|int ch
893=for apidoc Amh|bool|isLOWER_utf8_safe|U8 * s|U8 * end
d23c7e08 894=for apidoc Amh|bool|isLOWER_utf8|U8 * s|U8 * end
f16858ed
KW
895=for apidoc Amh|bool|isLOWER_LC|int ch
896=for apidoc Amh|bool|isLOWER_LC_uvchr|int ch
897=for apidoc Amh|bool|isLOWER_LC_utf8_safe|U8 * s| U8 *end
898
c99e91e9 899=for apidoc Am|bool|isOCTAL|char ch
2787a470 900Returns a boolean indicating whether the specified character is an
6aff1f14 901octal digit, [0-7].
243effed
KW
902The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
903C<isOCTAL>.
904
f16858ed
KW
905=for apidoc Amh|bool|isOCTAL_A|int ch
906=for apidoc Amh|bool|isOCTAL_L1|int ch
907
243effed
KW
908=for apidoc Am|bool|isPUNCT|char ch
909Returns a boolean indicating whether the specified character is a
6aff1f14
KW
910punctuation character, analogous to C<m/[[:punct:]]/>.
911Note that the definition of what is punctuation isn't as
243effed
KW
912straightforward as one might desire. See L<perlrecharclass/POSIX Character
913Classes> for details.
dcccc8ff 914See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
915variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>,
916C<isPUNCT_utf8_safe>, C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, C<isPUNCT_LC_utf8>,
917and C<isPUNCT_LC_utf8_safe>.
c99e91e9 918
f16858ed
KW
919=for apidoc Amh|bool|isPUNCT_A|int ch
920=for apidoc Amh|bool|isPUNCT_L1|int ch
921=for apidoc Amh|bool|isPUNCT_uvchr|int ch
922=for apidoc Amh|bool|isPUNCT_utf8_safe|U8 * s|U8 * end
d23c7e08 923=for apidoc Amh|bool|isPUNCT_utf8|U8 * s|U8 * end
f16858ed
KW
924=for apidoc Amh|bool|isPUNCT_LC|int ch
925=for apidoc Amh|bool|isPUNCT_LC_uvchr|int ch
926=for apidoc Amh|bool|isPUNCT_LC_utf8_safe|U8 * s| U8 *end
927
0c82b6df 928=for apidoc Am|bool|isSPACE|char ch
2787a470 929Returns a boolean indicating whether the specified character is a
6aff1f14 930whitespace character. This is analogous
398d098a 931to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 932this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
933locale forms of this macro (the ones with C<LC> in their names) matched
934precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
935in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
936(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
937See the L<top of this section|/Character classification> for an explanation of
938variants
059703b0
KW
939C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>,
940C<isSPACE_utf8_safe>, C<isSPACE_LC>, C<isSPACE_LC_uvchr>, C<isSPACE_LC_utf8>,
941and C<isSPACE_LC_utf8_safe>.
0c82b6df 942
f16858ed
KW
943=for apidoc Amh|bool|isSPACE_A|int ch
944=for apidoc Amh|bool|isSPACE_L1|int ch
945=for apidoc Amh|bool|isSPACE_uvchr|int ch
946=for apidoc Amh|bool|isSPACE_utf8_safe|U8 * s|U8 * end
d23c7e08 947=for apidoc Amh|bool|isSPACE_utf8|U8 * s|U8 * end
f16858ed
KW
948=for apidoc Amh|bool|isSPACE_LC|int ch
949=for apidoc Amh|bool|isSPACE_LC_uvchr|int ch
950=for apidoc Amh|bool|isSPACE_LC_utf8_safe|U8 * s| U8 *end
951
398d098a
KW
952=for apidoc Am|bool|isPSXSPC|char ch
953(short for Posix Space)
779cf272
KW
954Starting in 5.18, this is identical in all its forms to the
955corresponding C<isSPACE()> macros.
398d098a
KW
956The locale forms of this macro are identical to their corresponding
957C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
958non-locale forms differ from their C<isSPACE()> forms only in that the
959C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
960Otherwise they are identical. Thus this macro is analogous to what
961C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 962See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
963variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
964C<isPSXSPC_utf8_safe>, C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>,
965C<isPSXSPC_LC_utf8>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 966
f16858ed
KW
967=for apidoc Amh|bool|isPSXSPC_A|int ch
968=for apidoc Amh|bool|isPSXSPC_L1|int ch
969=for apidoc Amh|bool|isPSXSPC_uvchr|int ch
970=for apidoc Amh|bool|isPSXSPC_utf8_safe|U8 * s|U8 * end
d23c7e08 971=for apidoc Amh|bool|isPSXSPC_utf8|U8 * s|U8 * end
f16858ed
KW
972=for apidoc Amh|bool|isPSXSPC_LC|int ch
973=for apidoc Amh|bool|isPSXSPC_LC_uvchr|int ch
974=for apidoc Amh|bool|isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
975
954c1994 976=for apidoc Am|bool|isUPPER|char ch
2787a470 977Returns a boolean indicating whether the specified character is an
6aff1f14 978uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 979See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
980variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>,
981C<isUPPER_utf8_safe>, C<isUPPER_LC>, C<isUPPER_LC_uvchr>, C<isUPPER_LC_utf8>,
982and C<isUPPER_LC_utf8_safe>.
954c1994 983
f16858ed
KW
984=for apidoc Amh|bool|isUPPER_A|int ch
985=for apidoc Amh|bool|isUPPER_L1|int ch
986=for apidoc Amh|bool|isUPPER_uvchr|int ch
987=for apidoc Amh|bool|isUPPER_utf8_safe|U8 * s|U8 * end
d23c7e08 988=for apidoc Amh|bool|isUPPER_utf8|U8 * s|U8 * end
f16858ed
KW
989=for apidoc Amh|bool|isUPPER_LC|int ch
990=for apidoc Amh|bool|isUPPER_LC_uvchr|int ch
991=for apidoc Amh|bool|isUPPER_LC_utf8_safe|U8 * s| U8 *end
992
243effed 993=for apidoc Am|bool|isPRINT|char ch
8eea39dd 994Returns a boolean indicating whether the specified character is a
6aff1f14 995printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
996See the L<top of this section|/Character classification> for an explanation of
997variants
059703b0
KW
998C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>,
999C<isPRINT_utf8_safe>, C<isPRINT_LC>, C<isPRINT_LC_uvchr>, C<isPRINT_LC_utf8>,
1000and C<isPRINT_LC_utf8_safe>.
243effed 1001
f16858ed
KW
1002=for apidoc Amh|bool|isPRINT_A|int ch
1003=for apidoc Amh|bool|isPRINT_L1|int ch
1004=for apidoc Amh|bool|isPRINT_uvchr|int ch
1005=for apidoc Amh|bool|isPRINT_utf8_safe|U8 * s|U8 * end
d23c7e08 1006=for apidoc Amh|bool|isPRINT_utf8|U8 * s|U8 * end
f16858ed
KW
1007=for apidoc Amh|bool|isPRINT_LC|int ch
1008=for apidoc Amh|bool|isPRINT_LC_uvchr|int ch
1009=for apidoc Amh|bool|isPRINT_LC_utf8_safe|U8 * s| U8 *end
1010
243effed
KW
1011=for apidoc Am|bool|isWORDCHAR|char ch
1012Returns a boolean indicating whether the specified character is a character
1013that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1014in a regular expression. A word character is an alphabetic character, a
1015decimal digit, a connecting punctuation character (such as an underscore), or
1016a "mark" character that attaches to one of those (like some sort of accent).
1017C<isALNUM()> is a synonym provided for backward compatibility, even though a
1018word character includes more than the standard C language meaning of
1019alphanumeric.
dcccc8ff 1020See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
1021variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1022C<isWORDCHAR_utf8>, and C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>,
1023C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and C<isWORDCHAR_LC_utf8_safe>
1024are also as described there, but additionally include the platform's native
1025underscore.
8a58bdcf 1026
f16858ed
KW
1027=for apidoc Amh|bool|isWORDCHAR_A|int ch
1028=for apidoc Amh|bool|isWORDCHAR_L1|int ch
1029=for apidoc Amh|bool|isWORDCHAR_uvchr|int ch
1030=for apidoc Amh|bool|isWORDCHAR_utf8_safe|U8 * s|U8 * end
d23c7e08 1031=for apidoc Amh|bool|isWORDCHAR_utf8|U8 * s|U8 * end
f16858ed
KW
1032=for apidoc Amh|bool|isWORDCHAR_LC|int ch
1033=for apidoc Amh|bool|isWORDCHAR_LC_uvchr|int ch
1034=for apidoc Amh|bool|isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
1035=for apidoc Amh|bool|isALNUM|int ch
1036=for apidoc Amh|bool|isALNUM_A|int ch
1037=for apidoc Amh|bool|isALNUM_LC|int ch
1038=for apidoc Amh|bool|isALNUM_LC_uvchr|int ch
1039
8a58bdcf
KW
1040=for apidoc Am|bool|isXDIGIT|char ch
1041Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
1042digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1043and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
1044See the L<top of this section|/Character classification> for an explanation of
1045variants
059703b0
KW
1046C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>,
1047C<isXDIGIT_LC_uvchr>, C<isXDIGIT_LC_utf8>, and C<isXDIGIT_LC_utf8_safe>.
243effed 1048
f16858ed
KW
1049=for apidoc Amh|bool|isXDIGIT_A|int ch
1050=for apidoc Amh|bool|isXDIGIT_L1|int ch
1051=for apidoc Amh|bool|isXDIGIT_uvchr|int ch
1052=for apidoc Amh|bool|isXDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 1053=for apidoc Amh|bool|isXDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
1054=for apidoc Amh|bool|isXDIGIT_LC|int ch
1055=for apidoc Amh|bool|isXDIGIT_LC_uvchr|int ch
1056=for apidoc Amh|bool|isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1057
3c3ecf18
KW
1058=for apidoc Am|bool|isIDFIRST|char ch
1059Returns a boolean indicating whether the specified character can be the first
1060character of an identifier. This is very close to, but not quite the same as
1061the official Unicode property C<XID_Start>. The difference is that this
1062returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
1063See the L<top of this section|/Character classification> for an explanation of
1064variants
059703b0
KW
1065C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
1066C<isIDFIRST_utf8_safe>, C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>,
1067C<isIDFIRST_LC_utf8>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18 1068
f16858ed
KW
1069=for apidoc Amh|bool|isIDFIRST_A|int ch
1070=for apidoc Amh|bool|isIDFIRST_L1|int ch
1071=for apidoc Amh|bool|isIDFIRST_uvchr|int ch
1072=for apidoc Amh|bool|isIDFIRST_utf8_safe|U8 * s|U8 * end
d23c7e08 1073=for apidoc Amh|bool|isIDFIRST_utf8|U8 * s|U8 * end
f16858ed
KW
1074=for apidoc Amh|bool|isIDFIRST_LC|int ch
1075=for apidoc Amh|bool|isIDFIRST_LC_uvchr|int ch
1076=for apidoc Amh|bool|isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1077
3c3ecf18
KW
1078=for apidoc Am|bool|isIDCONT|char ch
1079Returns a boolean indicating whether the specified character can be the
1080second or succeeding character of an identifier. This is very close to, but
1081not quite the same as the official Unicode property C<XID_Continue>. The
1082difference is that this returns true only if the input character also matches
dcccc8ff 1083L</isWORDCHAR>. See the L<top of this section|/Character classification> for
059703b0
KW
1084an explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
1085C<isIDCONT_utf8>, C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>,
1086C<isIDCONT_LC_utf8>, and C<isIDCONT_LC_utf8_safe>.
3c3ecf18 1087
f16858ed
KW
1088=for apidoc Amh|bool|isIDCONT_A|int ch
1089=for apidoc Amh|bool|isIDCONT_L1|int ch
1090=for apidoc Amh|bool|isIDCONT_uvchr|int ch
1091=for apidoc Amh|bool|isIDCONT_utf8_safe|U8 * s|U8 * end
d23c7e08 1092=for apidoc Amh|bool|isIDCONT_utf8|U8 * s|U8 * end
f16858ed
KW
1093=for apidoc Amh|bool|isIDCONT_LC|int ch
1094=for apidoc Amh|bool|isIDCONT_LC_uvchr|int ch
1095=for apidoc Amh|bool|isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1096
243effed 1097=head1 Miscellaneous Functions
8eea39dd 1098
95a59cab 1099=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 1100Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
1101Behaviour is only well defined when isXDIGIT(*str) is true.
1102
e7c1e6c1 1103=head1 Character case changing
21da7284
KW
1104Perl uses "full" Unicode case mappings. This means that converting a single
1105character to another case may result in a sequence of more than one character.
1106For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1107character sequence C<SS>. This presents some complications The lowercase of
1108all characters in the range 0..255 is a single character, and thus
1109C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1110return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1111an API that does allow every possible legal result to be returned.) Likewise
1112no other function that is crippled by not being able to give the correct
1113results for the full range of possible inputs has been implemented here.
e7c1e6c1 1114
d713f9d9 1115=for apidoc Am|U8|toUPPER|int ch
1f607577
KW
1116Converts the specified character to uppercase. If the input is anything but an
1117ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 1118C<toUPPER_A> is equivalent.
954c1994 1119
d0da05db
KW
1120=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1121Converts the code point C<cp> to its uppercase version, and
1122stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1123point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1124that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1125bytes since the uppercase version may be longer than the original character.
1126
1127The first code point of the uppercased version is returned
21da7284
KW
1128(but note, as explained at L<the top of this section|/Character case
1129changing>, that there may be more.)
1f607577 1130
059703b0 1131=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1132Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1133extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
1134stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1135that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1136bytes since the uppercase version may be longer than the original character.
1137
1138The first code point of the uppercased version is returned
21da7284
KW
1139(but note, as explained at L<the top of this section|/Character case
1140changing>, that there may be more).
1f607577 1141
059703b0
KW
1142It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1143S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1144the UTF-8 for the input character is malformed in some way, the program may
1145croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1146of the implementation, and subject to change in future releases.
a239b1e2 1147
059703b0
KW
1148=for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1149Same as L</toUPPER_utf8>.
1f607577 1150
25200305
KW
1151=for apidoc Am|U8|toFOLD|U8 ch
1152Converts the specified character to foldcase. If the input is anything but an
1153ASCII uppercase character, that input character itself is returned. Variant
1154C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 1155Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 1156
d0da05db
KW
1157=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1158Converts the code point C<cp> to its foldcase version, and
1159stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1160point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1161that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1162bytes since the foldcase version may be longer than the original character.
1163
1164The first code point of the foldcased version is returned
21da7284
KW
1165(but note, as explained at L<the top of this section|/Character case
1166changing>, that there may be more).
1f607577 1167
059703b0 1168=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1169Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1170extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
1171stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1172that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1173bytes since the foldcase version may be longer than the original character.
1174
1175The first code point of the foldcased version is returned
21da7284
KW
1176(but note, as explained at L<the top of this section|/Character case
1177changing>, that there may be more).
1f607577 1178
059703b0 1179It will not attempt
a239b1e2
KW
1180to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1181true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1182input character is malformed in some way, the program may croak, or the
1183function may return the REPLACEMENT CHARACTER, at the discretion of the
1184implementation, and subject to change in future releases.
1185
059703b0
KW
1186=for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1187Same as L</toFOLD_utf8>.
1f607577
KW
1188
1189=for apidoc Am|U8|toLOWER|U8 ch
1190Converts the specified character to lowercase. If the input is anything but an
1191ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 1192C<toLOWER_A> is equivalent.
954c1994 1193
1f607577 1194=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
1195Converts the specified Latin1 character to lowercase. The results are
1196undefined if the input doesn't fit in a byte.
1f607577
KW
1197
1198=for apidoc Am|U8|toLOWER_LC|U8 ch
1199Converts the specified character to lowercase using the current locale's rules,
1200if possible; otherwise returns the input character itself.
1201
d0da05db
KW
1202=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1203Converts the code point C<cp> to its lowercase version, and
1204stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1205point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1206that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1207bytes since the lowercase version may be longer than the original character.
1208
1209The first code point of the lowercased version is returned
21da7284
KW
1210(but note, as explained at L<the top of this section|/Character case
1211changing>, that there may be more).
1f607577 1212
bd350c85 1213=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1214Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1215extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
1216stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1217that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1218bytes since the lowercase version may be longer than the original character.
1219
1220The first code point of the lowercased version is returned
21da7284
KW
1221(but note, as explained at L<the top of this section|/Character case
1222changing>, that there may be more).
059703b0
KW
1223It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1224S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1225the UTF-8 for the input character is malformed in some way, the program may
1226croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1227of the implementation, and subject to change in future releases.
1f607577 1228
059703b0
KW
1229=for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1230Same as L</toLOWER_utf8>.
1f607577 1231
25200305
KW
1232=for apidoc Am|U8|toTITLE|U8 ch
1233Converts the specified character to titlecase. If the input is anything but an
1234ASCII lowercase character, that input character itself is returned. Variant
b7d90381 1235C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 1236range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 1237not a concept used in locale handling, so there is no functionality for that.)
25200305 1238
d0da05db
KW
1239=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1240Converts the code point C<cp> to its titlecase version, and
1241stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1242point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1243that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1244bytes since the titlecase version may be longer than the original character.
1245
1246The first code point of the titlecased version is returned
21da7284
KW
1247(but note, as explained at L<the top of this section|/Character case
1248changing>, that there may be more).
1f607577 1249
059703b0 1250=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1251Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1252extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1253stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1254that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1255bytes since the titlecase version may be longer than the original character.
1256
1257The first code point of the titlecased version is returned
21da7284
KW
1258(but note, as explained at L<the top of this section|/Character case
1259changing>, that there may be more).
1f607577 1260
059703b0 1261It will not attempt
a239b1e2
KW
1262to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1263true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1264input character is malformed in some way, the program may croak, or the
1265function may return the REPLACEMENT CHARACTER, at the discretion of the
1266implementation, and subject to change in future releases.
1267
059703b0
KW
1268=for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1269Same as L</toTITLE_utf8>.
1f607577 1270
954c1994 1271=cut
353c9b6f 1272
d0da05db 1273XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1274really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1275and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1276
8a58bdcf 1277Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1278patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1279
954c1994
GS
1280*/
1281
8f5283f4
KW
1282/*
1283 void below because that's the best fit, and works for Devel::PPPort
1284=for apidoc AmnU|void|WIDEST_UTYPE
1285
1286Yields the widest unsigned integer type on the platform, currently either
1287C<U32> or C<64>. This can be used in declarations such as
1288
1289 WIDEST_UTYPE my_uv;
1290
1291or casts
1292
1293 my_uv = (WIDEST_UTYPE) val;
1294
1295=cut
1296
1297*/
de9e2639
KW
1298#ifdef QUADKIND
1299# define WIDEST_UTYPE U64
7c062697
KW
1300#else
1301# define WIDEST_UTYPE U32
1302#endif
1303
3912bc88
KW
1304/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1305 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1306 * bits of information are lost even on a 64-bit machine, but to get the
1307 * compiler to optimize it out if possible. This is because Configure makes
1308 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1309 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1310 *
1311 * For Coverity, be always true, because otherwise Coverity thinks
1312 * it finds several expressions that are always true, independent
1313 * of operands. Well, they are, but that is kind of the point.
220c71bf 1314 */
7e75d1a1 1315#ifndef __COVERITY__
6c5b02ac
KW
1316 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1317 * pointer) */
1318#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1319 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1320#else
1321#define FITS_IN_8_BITS(c) (1)
1322#endif
cf301eb7 1323
45f4bb73 1324/* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
833b0f46 1325 * Written this way so that after optimization, only one conditional test is
76d3ad4c
KW
1326 * needed. (The NV casts stop any warnings about comparison always being true
1327 * if called with an unsigned. The cast preserves the sign, which is all we
1328 * care about.) */
1329#define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1330 __ASSERT_((NV) (n) >= 0) \
1331 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
833b0f46 1332
94250c4f
KW
1333/* Returns true if c is in the range l..u, where 'l' is non-negative
1334 * Written this way so that after optimization, only one conditional test is
4758c20d 1335 * needed. */
1eaefa6e 1336#define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
4758c20d 1337 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
4758c20d
KW
1338 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1339 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
45f4bb73 1340 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
305fe86e 1341
41f43cc2 1342#ifdef EBCDIC
b6340bd0 1343# ifndef _ALL_SOURCE
0852beac
KW
1344 /* The native libc isascii() et.al. functions return the wrong results
1345 * on at least z/OS unless this is defined. */
b6340bd0
KW
1346# error _ALL_SOURCE should probably be defined
1347# endif
41f43cc2 1348#else
0852beac
KW
1349 /* There is a simple definition of ASCII for ASCII platforms. But the
1350 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1351 * other macros below.
3f3c579d
KW
1352 *
1353 * The cast here is used instead of '(c) >= 0', because some compilers emit
1354 * a warning that that test is always true when the parameter is an
1355 * unsigned type. khw supposes that it could be written as
1356 * && ((c) == '\0' || (c) > 0)
1357 * to avoid the message, but the cast will likely avoid extra branches even
1358 * with stupid compilers.
1359 *
1360 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1361 * a pointer) */
9c903d59 1362# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1363#endif
1364
38694112
KW
1365/* Take the eight possible bit patterns of the lower 3 bits and you get the
1366 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1367 * can be ignored. If the rest match '0', we have an octal */
1368#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1369
9fb1bf9d 1370#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1371
a500dc72
KW
1372/* Character class numbers. For internal core Perl use only. The ones less
1373 * than 32 are used in PL_charclass[] and the ones up through the one that
1374 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1375 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1376 * their actual definitions are here. If that file has a name not used here,
1377 * it won't compile.
1709d539
KW
1378 *
1379 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561 1380 * frequency of their use. This gives a slight edge to exiting a loop earlier
58a3ba2c
KW
1381 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1382 * real Posix character class, and some (small) inefficiencies in regular
1383 * expression handling would be introduced by putting it in the middle of those
1384 * that are. Also, cntrl and ascii come after the others as it may be useful
1385 * to group these which have no members that match above Latin1, (or above
1386 * ASCII in the latter case) */
1387
1709d539
KW
1388# define _CC_WORDCHAR 0 /* \w and [:word:] */
1389# define _CC_DIGIT 1 /* \d and [:digit:] */
1390# define _CC_ALPHA 2 /* [:alpha:] */
1391# define _CC_LOWER 3 /* [:lower:] */
1392# define _CC_UPPER 4 /* [:upper:] */
1393# define _CC_PUNCT 5 /* [:punct:] */
1394# define _CC_PRINT 6 /* [:print:] */
15861f94 1395# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1396# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1397# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
779cf272 1398# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1399# define _CC_BLANK 11 /* [:blank:] */
1400# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1401# define _CC_CNTRL 13 /* [:cntrl:] */
1402# define _CC_ASCII 14 /* [:ascii:] */
1403# define _CC_VERTSPACE 15 /* \v */
1709d539 1404
a0947d7b
KW
1405# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1406
1709d539 1407/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1408 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1409# define _CC_IDFIRST 16
1410# define _CC_CHARNAME_CONT 17
1411# define _CC_NONLATIN1_FOLD 18
1412# define _CC_NONLATIN1_SIMPLE_FOLD 19
1413# define _CC_QUOTEMETA 20
1414# define _CC_NON_FINAL_FOLD 21
1415# define _CC_IS_IN_SOME_FOLD 22
2ae9030c
KW
1416# define _CC_BINDIGIT 23
1417# define _CC_OCTDIGIT 24
1418# define _CC_MNEMONIC_CNTRL 25
073c22b3
KW
1419
1420/* This next group is only used on EBCDIC platforms, so theoretically could be
1421 * shared with something entirely different that's only on ASCII platforms */
abb8abf6 1422# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
0654f0ab 1423/* Unused: 26-30
f4cdb42c
KW
1424 * If more bits are needed, one could add a second word for non-64bit
1425 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1426 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1427 * is used only for optimization (as of this writing), and differs in the
1428 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1429 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1430 * would just cause /i regexes which match them to run less efficiently.
1431 * Similarly the EBCDIC-only bits are used just for speed, and could be
1432 * replaced by other means */
96ac0975 1433
3a371f2f
KW
1434#if defined(PERL_CORE) || defined(PERL_EXT)
1435/* An enum version of the character class numbers, to help compilers
1436 * optimize */
1437typedef enum {
3a371f2f 1438 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1439 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1440 _CC_ENUM_ASCII = _CC_ASCII,
1441 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1442 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1443 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1444 _CC_ENUM_DIGIT = _CC_DIGIT,
1445 _CC_ENUM_GRAPH = _CC_GRAPH,
1446 _CC_ENUM_LOWER = _CC_LOWER,
1447 _CC_ENUM_PRINT = _CC_PRINT,
1448 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1449 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1450 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1451 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1452 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1453 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1454} _char_class_number;
1455#endif
1456
86f72d56 1457#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1458
6635f04f 1459START_EXTERN_C
96ac0975
NC
1460# ifdef DOINIT
1461EXTCONST U32 PL_charclass[] = {
1462# include "l1_char_class_tab.h"
1463};
1464
1465# else /* ! DOINIT */
1466EXTCONST U32 PL_charclass[];
1467# endif
6635f04f 1468END_EXTERN_C
96ac0975 1469
265c1f46 1470 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1471# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1472
1473 /* For internal core Perl use only: the base macro for defining macros like
1474 * isALPHA */
ff7ecfc3 1475# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1476 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1477
f4cdb42c
KW
1478 /* The mask for the _A versions of the macros; it just adds in the bit for
1479 * ASCII. */
1480# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1481
4650c663
KW
1482 /* For internal core Perl use only: the base macro for defining macros like
1483 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1484 * the ASCII bit are present */
b7d90381
KW
1485# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1486 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1487 == _CC_mask_A(classnum)))
f4cdb42c 1488
26c1d9d8
KW
1489/* On ASCII platforms certain classes form a single range. It's faster to
1490 * special case these. isDIGIT is a single range on all platforms */
b877c1ff
KW
1491# ifdef EBCDIC
1492# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1493# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1494# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1495# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1496# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1497# else
26c1d9d8 1498 /* By folding the upper and lowercase, we can use a single range */
b877c1ff 1499# define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
26c1d9d8 1500# define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
b877c1ff
KW
1501# define isLOWER_A(c) inRANGE(c, 'a', 'z')
1502# define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1503# define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1504# endif
15861f94 1505# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1506# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1507# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
b877c1ff 1508# define isDIGIT_A(c) inRANGE(c, '0', '9')
f4cdb42c
KW
1509# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1510# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
f4cdb42c 1511# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1512# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1513 */
d95f8b6a 1514# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1515# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1516# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1517# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1518
1519 /* continuation character for legal NAME in \N{NAME} */
1520# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1521
1522# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1523# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1524# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1525# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1526# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1527# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1528# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1529# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1530# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1531# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1532
0852beac
KW
1533# ifdef EBCDIC
1534# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1535# endif
1536
f12c0118
KW
1537 /* Participates in a single-character fold with a character above 255 */
1538# 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)))
1539
1540 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1541# 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 1542
4eeeb416 1543# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1544# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1545 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1546# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1547 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
5e6ebb12
KW
1548
1549/* is c a control character for which we have a mnemonic? */
1550# if defined(PERL_CORE) || defined(PERL_EXT)
1551# define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1552# endif
687c8d01 1553#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1554
1555 /* If we don't have perl.h, we are compiling a utility program. Below we
1556 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1557 * to it. Most are coded based on first principles. These are written to
74665a89 1558 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
182c4ace 1559# define isDIGIT_A(c) inRANGE(c, '0', '9')
0852beac 1560# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1561# define isSPACE_A(c) (isBLANK_A(c) \
1562 || (c) == '\n' \
1563 || (c) == '\r' \
1564 || (c) == '\v' \
0852beac 1565 || (c) == '\f')
74665a89
KW
1566 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1567 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1568 * performance (if optimization isn't on), and allow the same code to be
1569 * used for both platform types */
182c4ace
KW
1570# define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1571 || inRANGE((c), 'j', 'r') \
1572 || inRANGE((c), 's', 'z')
1573# define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1574 || inRANGE((c), 'J', 'R') \
1575 || inRANGE((c), 'S', 'Z')
a4d7a999
KW
1576# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1577# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1578# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1579# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
182c4ace
KW
1580# define isXDIGIT_A(c) ( isDIGIT_A(c) \
1581 || inRANGE((c), 'a', 'f') \
1582 || inRANGE((c), 'A', 'F')
74665a89
KW
1583# define isPUNCT_A(c) ((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) == '[' || (c) == '\\' \
1591 || (c) == ']' || (c) == '^' || (c) == '_' \
1592 || (c) == '`' || (c) == '{' || (c) == '|' \
1593 || (c) == '}' || (c) == '~')
1594# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1595# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1596
0852beac 1597# ifdef EBCDIC
74665a89
KW
1598 /* The below is accurate for the 3 EBCDIC code pages traditionally
1599 * supported by perl. The only difference between them in the controls
1600 * is the position of \n, and that is represented symbolically below */
1601# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1602 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1603 || (c) == '\t' || (c) == '\v' \
182c4ace 1604 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
8ec0a736 1605 || (c) == 7F /* U+7F DEL */ \
182c4ace
KW
1606 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1607 DC[1-3] */ \
74665a89
KW
1608 || (c) == 0x18 /* U+18 CAN */ \
1609 || (c) == 0x19 /* U+19 EOM */ \
182c4ace 1610 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
74665a89
KW
1611 || (c) == 0x26 /* U+17 ETB */ \
1612 || (c) == 0x27 /* U+1B ESC */ \
1613 || (c) == 0x2D /* U+05 ENQ */ \
1614 || (c) == 0x2E /* U+06 ACK */ \
1615 || (c) == 0x32 /* U+16 SYN */ \
1616 || (c) == 0x37 /* U+04 EOT */ \
1617 || (c) == 0x3C /* U+14 DC4 */ \
1618 || (c) == 0x3D /* U+15 NAK */ \
1619 || (c) == 0x3F)/* U+1A SUB */
0852beac 1620# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1621# else /* isASCII is already defined for ASCII platforms, so can use that to
1622 define isCNTRL */
1623# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1624# endif
1625
3ffc8c70 1626 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1627 * during debugging, and it seems best to keep them. We may be called
1628 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1629 * do anything anyway, so make it not a problem */
1630# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1631# define NATIVE_TO_LATIN1(ch) (ch)
1632# endif
3ded5eb0
KW
1633# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1634# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1635# define isBLANK_L1(c) (isBLANK_A(c) \
1636 || (FITS_IN_8_BITS(c) \
1637 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1638# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1639# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1640# define isLOWER_L1(c) (isLOWER_A(c) \
1641 || (FITS_IN_8_BITS(c) \
ae683a5f 1642 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1643 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1644 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1645 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1646 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1647# define isPRINT_L1(c) (isPRINT_A(c) \
1648 || (FITS_IN_8_BITS(c) \
1649 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1650# define isPUNCT_L1(c) (isPUNCT_A(c) \
1651 || (FITS_IN_8_BITS(c) \
ae683a5f 1652 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1653 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1654 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1655 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1656 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1657 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1658 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1659# define isSPACE_L1(c) (isSPACE_A(c) \
1660 || (FITS_IN_8_BITS(c) \
ae683a5f 1661 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1662 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1663# define isUPPER_L1(c) (isUPPER_A(c) \
1664 || (FITS_IN_8_BITS(c) \
182c4ace
KW
1665 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1666 0xC0, 0xDE) \
3ded5eb0
KW
1667 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1668# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1669# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1670# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1671 || isBLANK_L1(c) \
1672 || (c) == '-' \
1673 || (c) == '(' \
1674 || (c) == ')')
1675 /* The following are not fully accurate in the above-ASCII range. I (khw)
1676 * don't think it's necessary to be so for the purposes where this gets
1677 * compiled */
1678# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1679# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1680
1681 /* And these aren't accurate at all. They are useful only for above
1682 * Latin1, which utilities and bootstrapping don't deal with */
1683# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1684# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1685# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1686
1687 /* Many of the macros later in this file are defined in terms of these. By
1688 * implementing them with a function, which converts the class number into
1689 * a call to the desired macro, all of the later ones work. However, that
1690 * function won't be actually defined when building a utility program (no
1691 * perl.h), and so a compiler error will be generated if one is attempted
1692 * to be used. And the above-Latin1 code points require Unicode tables to
1693 * be present, something unlikely to be the case when bootstrapping */
1694# define _generic_isCC(c, classnum) \
1695 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1696# define _generic_isCC_A(c, classnum) \
1697 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1698#endif /* End of no perl.h H_PERL */
8a58bdcf 1699
e66b99e9
KW
1700#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1701#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1702#define isASCII_A(c) isASCII(c)
1703#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1704#define isBLANK(c) isBLANK_A(c)
1705#define isCNTRL(c) isCNTRL_A(c)
1706#define isDIGIT(c) isDIGIT_A(c)
1707#define isGRAPH(c) isGRAPH_A(c)
1708#define isIDFIRST(c) isIDFIRST_A(c)
1709#define isLOWER(c) isLOWER_A(c)
1710#define isPRINT(c) isPRINT_A(c)
779cf272 1711#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1712#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1713#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1714#define isPUNCT(c) isPUNCT_A(c)
1715#define isSPACE(c) isSPACE_A(c)
1716#define isUPPER(c) isUPPER_A(c)
1717#define isWORDCHAR(c) isWORDCHAR_A(c)
1718#define isXDIGIT(c) isXDIGIT_A(c)
1719
1720/* ASCII casing. These could also be written as
1721 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1722 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1723 which uses table lookup and mask instead of subtraction. (This would
c5e9991e
KW
1724 work because the _MOD does not apply in the ASCII range).
1725
1726 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1727 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1728 there non-ASCII invariants, and all of them are controls.) */
68067e4e
DM
1729#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1730#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1731
25200305
KW
1732/* In the ASCII range, these are equivalent to what they're here defined to be.
1733 * But by creating these definitions, other code doesn't have to be aware of
c5e9991e
KW
1734 * this detail. Actually this works for all UTF-8 invariants, not just the
1735 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
25200305 1736#define toFOLD(c) toLOWER(c)
25200305
KW
1737#define toTITLE(c) toUPPER(c)
1738
c753c8d3
KW
1739#define toLOWER_A(c) toLOWER(c)
1740#define toUPPER_A(c) toUPPER(c)
25200305
KW
1741#define toFOLD_A(c) toFOLD(c)
1742#define toTITLE_A(c) toTITLE(c)
1a0901db 1743
4650c663 1744/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1745#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1746 ? (c) \
f4cd282c 1747 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1748#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1749
1a0901db 1750/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1751 * all mapped to one of them, and these need special handling; returns the
1752 * input itself if is out-of-range */
b2bf251f 1753#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1754 ? (c) \
f4cd282c 1755 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1756#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1757
beab9ebe
KW
1758/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1759
1760/* For internal core Perl use only: the base macro for defining macros like
1761 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1762 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1763 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1764 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1765 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1766 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1767 * code points under locale rules, use the macros based on _generic_LC_uvchr
1768 * instead of this. */
beab9ebe
KW
1769#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1770 (! FITS_IN_8_BITS(c) \
1771 ? 0 \
31f05a37
KW
1772 : IN_UTF8_CTYPE_LOCALE \
1773 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1774 : cBOOL(non_utf8))
1775
1776/* For internal core Perl use only: a helper macro for defining macros like
1777 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1778 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1779 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1780 * _generic_LC_base for more info */
1781#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1782 _generic_LC_base(c,utf8_locale_classnum, \
1783 non_utf8_func( (_LC_CAST) (c)))
1784
1785/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1786 * 'c' is the platform's native underscore character */
1787#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1788 _generic_LC_base(c, utf8_locale_classnum, \
1789 (non_utf8_func( (_LC_CAST) (c)) \
1790 || (char)(c) == '_'))
1791
1792/* These next three are also for internal core Perl use only: case-change
247985d4
KW
1793 * helper macros. The reason for using the PL_latin arrays is in case the
1794 * system function is defective; it ensures uniform results that conform to the
b257a28c 1795 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
beab9ebe
KW
1796#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1797 ? (c) \
31f05a37
KW
1798 : (IN_UTF8_CTYPE_LOCALE) \
1799 ? PL_latin1_lc[ (U8) (c) ] \
5a10328c 1800 : (cast)function((cast)(c)))
beab9ebe 1801
31f05a37
KW
1802/* Note that the result can be larger than a byte in a UTF-8 locale. It
1803 * returns a single value, so can't adequately return the upper case of LATIN
1804 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1805 * values "SS"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1806 * otherwise returns its input. It does not handle the anomalies in UTF-8
1807 * Turkic locales. */
beab9ebe
KW
1808#define _generic_toUPPER_LC(c, function, cast) \
1809 (! FITS_IN_8_BITS(c) \
1810 ? (c) \
31f05a37 1811 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1812 ? (cast)function((cast)(c)) \
31f05a37
KW
1813 : ((((U8)(c)) == MICRO_SIGN) \
1814 ? GREEK_CAPITAL_LETTER_MU \
1815 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1816 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1817 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1818 ? (__ASSERT_(0) (c)) \
1819 : PL_mod_latin1_uc[ (U8) (c) ])))))
1820
1821/* Note that the result can be larger than a byte in a UTF-8 locale. It
1822 * returns a single value, so can't adequately return the fold case of LATIN
1823 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1824 * values "ss"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1825 * otherwise returns its input. It does not handle the anomalies in UTF-8
1826 * Turkic locales */
beab9ebe 1827#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1828 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1829 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1830 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1831 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1832 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1833
84061b6a 1834/* Use the libc versions for these if available. */
f05550c0 1835#if defined(HAS_ISASCII)
84061b6a
KW
1836# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1837#else
1838# define isASCII_LC(c) isASCII(c)
1839#endif
1840
f05550c0 1841#if defined(HAS_ISBLANK)
84061b6a 1842# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1843#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1844# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1845#endif
1846
f05550c0 1847#define _LC_CAST U8
bbce6d69 1848
f05550c0 1849#ifdef WIN32
375f5f06
KW
1850 /* The Windows functions don't bother to follow the POSIX standard, which
1851 * for example says that something can't both be a printable and a control.
1852 * But Windows treats the \t control as a printable, and does such things
1853 * as making superscripts into both digits and punctuation. This tames
1854 * these flaws by assuming that the definitions of both controls and space
1855 * are correct, and then making sure that other definitions don't have
1856 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1857 * Not all possible weirdnesses are checked for, just the ones that were
1858 * detected on actual Microsoft code pages */
1859
b7d90381
KW
1860# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1861# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1862
1863# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1864 && isALPHANUMERIC_LC(c))
1865# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1866 ! isPUNCT_LC(c))
1867# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1868 isALPHANUMERIC_LC(c))
1869# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1870# define isIDFIRST_LC(c) (((c) == '_') \
1871 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1872# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1873# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1874# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1875# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1876# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1877# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1878 && isALPHANUMERIC_LC(c))
f05550c0
BF
1879
1880# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1881# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1882# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1883
1884#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1885 /* For most other platforms */
beab9ebe 1886
f05550c0
BF
1887# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1888# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1889# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1890# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1891# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1892# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1893# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1894# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1895# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1896# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1897# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1898# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1899# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1900
1901
1902# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1903# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1904# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1905
1906#else /* The final fallback position */
1907
b7d90381
KW
1908# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1909# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1910# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1911# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1912# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1913# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1914# define isLOWER_LC(c) (isascii(c) && islower(c))
1915# define isPRINT_LC(c) (isascii(c) && isprint(c))
1916# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1917# define isSPACE_LC(c) (isascii(c) && isspace(c))
1918# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1919# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1920# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1921
1922# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1923# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1924# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1925
f05550c0 1926#endif
55204971 1927
eba68aa0
KW
1928#define isIDCONT(c) isWORDCHAR(c)
1929#define isIDCONT_A(c) isWORDCHAR_A(c)
1930#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1931#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1932#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1933
4650c663 1934/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1935 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1936 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1937 * classes where there is a macro or function 'above_latin1' that takes the
1938 * single argument 'c' and returns the desired value. These exist for those
2366ba44
KW
1939 * classes which have simple definitions, avoiding the overhead of an inversion
1940 * list binary search. _generic_invlist_uvchr() can be used
4650c663 1941 * for classes where that overhead is faster than a direct lookup.
d0da05db 1942 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1943 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1944 * have duplicate checks here, so could create versions of the macros that
1945 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1946
1947/* Note that all ignore 'use bytes' */
1e222e4f
KW
1948#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1949 ? _generic_isCC(c, classnum) \
cd500f2f 1950 : above_latin1(c))
2366ba44 1951#define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
1e222e4f 1952 ? _generic_isCC(c, classnum) \
922e8cb4 1953 : _is_uni_FOO(classnum, c))
2366ba44
KW
1954#define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
1955#define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
d0da05db
KW
1956#define isASCII_uvchr(c) isASCII(c)
1957#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1958#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2366ba44
KW
1959#define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
1960#define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1961#define isIDCONT_uvchr(c) \
1962 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1963#define isIDFIRST_uvchr(c) \
1964 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2366ba44
KW
1965#define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
1966#define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
d0da05db 1967
2366ba44 1968#define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
d0da05db
KW
1969#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1970#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1971
2366ba44 1972#define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
d0da05db 1973#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2366ba44 1974#define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
d0da05db
KW
1975#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1976
1977#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1978#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1979#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1980#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1981
1982/* For backwards compatibility, even though '_uni' should mean official Unicode
1983 * code points, in Perl it means native for those below 256 */
1984#define isALPHA_uni(c) isALPHA_uvchr(c)
1985#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1986#define isASCII_uni(c) isASCII_uvchr(c)
1987#define isBLANK_uni(c) isBLANK_uvchr(c)
1988#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1989#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1990#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1991#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1992#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1993#define isLOWER_uni(c) isLOWER_uvchr(c)
1994#define isPRINT_uni(c) isPRINT_uvchr(c)
1995#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1996#define isSPACE_uni(c) isSPACE_uvchr(c)
1997#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1998#define isUPPER_uni(c) isUPPER_uvchr(c)
1999#define isVERTWS_uni(c) isVERTWS_uvchr(c)
2000#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2001#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2002#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2003#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2004#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2005#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 2006
4650c663
KW
2007/* For internal core Perl use only: the base macros for defining macros like
2008 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 2009 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 2010 * simple class definitions; the other for more complex. These are like
d0da05db 2011 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
2012#define _generic_LC_uvchr(latin1, above_latin1, c) \
2013 (c < 256 ? latin1(c) : above_latin1(c))
2366ba44 2014#define _generic_LC_invlist_uvchr(latin1, classnum, c) \
cd500f2f
KW
2015 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2016
2366ba44
KW
2017#define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2018#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
cd500f2f 2019 _CC_ALPHANUMERIC, c)
b7d90381
KW
2020#define isASCII_LC_uvchr(c) isASCII_LC(c)
2021#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2022 is_HORIZWS_cp_high, c)
feeab5a9 2023#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2366ba44
KW
2024#define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2025#define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 2026#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 2027 _is_uni_perl_idcont, c)
b7d90381 2028#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f 2029 _is_uni_perl_idstart, c)
2366ba44
KW
2030#define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2031#define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 2032#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2366ba44 2033#define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 2034#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 2035 is_XPERLSPACE_cp_high, c)
2366ba44
KW
2036#define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2037#define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
cd500f2f 2038 _CC_WORDCHAR, c)
b7d90381
KW
2039#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2040 is_XDIGIT_cp_high, c)
e712593e 2041
b7d90381 2042#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 2043
da8c1a98
KW
2044/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2045 * they don't otherwise go out of their way to look for malformed UTF-8. If
2046 * they can return accurate results without knowing if the input is otherwise
2047 * malformed, they do so. For example isASCII is accurate in spite of any
2048 * non-length malformations because it looks only at a single byte. Likewise
2049 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2050 * variant ones return FALSE. But, if the input has to be well-formed in order
2051 * for the results to be accurate, the macros will test and if malformed will
2052 * call a routine to die
2053 *
2054 * Except for toke.c, the macros do assume that e > p, asserting that on
2055 * DEBUGGING builds. Much code that calls these depends on this being true,
2056 * for other reasons. toke.c is treated specially as using the regular
2057 * assertion breaks it in many ways. All strings that these operate on there
2058 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2059 * bunch of code in toke.c assumes that this is true, so the assertion allows
2060 * for that */
2061#ifdef PERL_IN_TOKE_C
2062# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2063#else
2064# define _utf8_safe_assert(p,e) ((e) > (p))
2065#endif
2066
2067#define _generic_utf8_safe(classnum, p, e, above_latin1) \
c81b3562
KW
2068 ((! _utf8_safe_assert(p, e)) \
2069 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2070 : (UTF8_IS_INVARIANT(*(p))) \
da8c1a98
KW
2071 ? _generic_isCC(*(p), classnum) \
2072 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2073 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2074 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2075 classnum) \
2076 : (_force_out_malformed_utf8_message( \
2077 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2078 : above_latin1))
b7d90381
KW
2079/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2080 * 'above_latin1' can be a macro */
da8c1a98
KW
2081#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2082 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2366ba44 2083#define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2084 _generic_utf8_safe(classnum, p, e, \
2085 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2086 ? (_force_out_malformed_utf8_message( \
2087 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2088 : above_latin1(p)))
2366ba44
KW
2089/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2090 * 'above_latin1' parameter */
2091#define _generic_invlist_utf8_safe(classnum, p, e) \
2092 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
922e8cb4 2093
cc8ab7c0 2094/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
2095 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2096 * class is TRUE for. Hence it can skip the tests for this range.
2097 * 'above_latin1' should include its arguments */
da8c1a98
KW
2098#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2099 (__ASSERT_(_utf8_safe_assert(p, e)) \
2100 (UTF8_IS_INVARIANT(*(p))) \
2101 ? _generic_isCC(*(p), classnum) \
2102 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2103 ? 0 /* Note that doesn't check validity for latin1 */ \
2104 : above_latin1)
2105
84238efa 2106
059703b0
KW
2107#define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2108#define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2109#define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2110#define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2111#define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2112#define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2113#define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2114#define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2115#define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2116#define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2117#define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2118#define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2119#define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2120#define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2121#define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2122#define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2123#define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2124#define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
e8fa43e2 2125
2366ba44 2126#define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
da8c1a98 2127#define isALPHANUMERIC_utf8_safe(p, e) \
2366ba44 2128 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
da8c1a98
KW
2129#define isASCII_utf8_safe(p, e) \
2130 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2131 * works */ \
2132 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2133#define isBLANK_utf8_safe(p, e) \
2366ba44 2134 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
da8c1a98 2135
e8fa43e2
KW
2136#ifdef EBCDIC
2137 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2138 * more efficient macro instead of the more general one */
da8c1a98 2139# define isCNTRL_utf8_safe(p, e) \
56d02b8c 2140 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 2141#else
da8c1a98 2142# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
2143#endif
2144
da8c1a98
KW
2145#define isDIGIT_utf8_safe(p, e) \
2146 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2366ba44
KW
2147 _is_utf8_FOO(_CC_DIGIT, p, e))
2148#define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
da8c1a98 2149#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
dd1a3ba7 2150 _is_utf8_perl_idcont, p, e)
e5dcd934 2151
c11ff943
KW
2152/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2153 * IDFIRST is an alnum. See
8034715d 2154 * https://github.com/Perl/perl5/issues/10275 for more detail than you
f91dcd13
KW
2155 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2156 * This used to be not the XID version, but we decided to go with the more
2157 * modern Unicode definition */
da8c1a98
KW
2158#define isIDFIRST_utf8_safe(p, e) \
2159 _generic_func_utf8_safe(_CC_IDFIRST, \
dd1a3ba7 2160 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
da8c1a98 2161
2366ba44
KW
2162#define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2163#define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
da8c1a98 2164#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2366ba44 2165#define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
da8c1a98 2166#define isSPACE_utf8_safe(p, e) \
2366ba44
KW
2167 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2168#define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
da8c1a98 2169#define isVERTWS_utf8_safe(p, e) \
2366ba44 2170 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
da8c1a98 2171#define isWORDCHAR_utf8_safe(p, e) \
2366ba44 2172 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
da8c1a98
KW
2173#define isXDIGIT_utf8_safe(p, e) \
2174 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2175 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2176 ? (_force_out_malformed_utf8_message( \
2177 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2178 : is_XDIGIT_high(p)))
a0ed51b3 2179
059703b0
KW
2180#define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2181#define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2182#define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2183#define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2e8adce6 2184
567b353c 2185/* For internal core use only, subject to change */
059703b0
KW
2186#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2187#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2188#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2189#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
a1a5ec35
KW
2190
2191#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2192#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2193#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2194#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 2195
059703b0
KW
2196#define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2197#define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2198#define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2199#define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2200#define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2201#define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2202#define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2203#define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2204#define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2205#define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2206#define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2207#define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2208#define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2209#define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2210#define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2211#define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2212#define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
34aeb2e9 2213
da8c1a98
KW
2214/* For internal core Perl use only: the base macros for defining macros like
2215 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2216 * point in 'p' is within the 0-255 range, it uses locale rules from the
2217 * passed-in 'macro' parameter */
2218#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2219 (__ASSERT_(_utf8_safe_assert(p, e)) \
2220 (UTF8_IS_INVARIANT(*(p))) \
2221 ? macro(*(p)) \
2222 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2223 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2224 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2225 : (_force_out_malformed_utf8_message( \
2226 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2227 : above_latin1))
2228
2366ba44
KW
2229#define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2230 _generic_LC_utf8_safe(macro, p, e, \
2231 _is_utf8_FOO(classnum, p, e))
da8c1a98
KW
2232
2233#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2234 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2235
2366ba44 2236#define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2237 _generic_LC_utf8_safe(classnum, p, e, \
2238 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2239 ? (_force_out_malformed_utf8_message( \
2240 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2241 : above_latin1(p)))
2242
2243#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2366ba44 2244 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
da8c1a98
KW
2245 _CC_ALPHANUMERIC, p, e)
2246#define isALPHA_LC_utf8_safe(p, e) \
2366ba44 2247 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
da8c1a98
KW
2248#define isASCII_LC_utf8_safe(p, e) \
2249 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2250#define isBLANK_LC_utf8_safe(p, e) \
2366ba44 2251 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
da8c1a98
KW
2252#define isCNTRL_LC_utf8_safe(p, e) \
2253 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2254#define isDIGIT_LC_utf8_safe(p, e) \
2366ba44 2255 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
da8c1a98 2256#define isGRAPH_LC_utf8_safe(p, e) \
2366ba44 2257 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
da8c1a98
KW
2258#define isIDCONT_LC_utf8_safe(p, e) \
2259 _generic_LC_func_utf8_safe(isIDCONT_LC, \
dd1a3ba7 2260 _is_utf8_perl_idcont, p, e)
da8c1a98
KW
2261#define isIDFIRST_LC_utf8_safe(p, e) \
2262 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
dd1a3ba7 2263 _is_utf8_perl_idstart, p, e)
da8c1a98 2264#define isLOWER_LC_utf8_safe(p, e) \
2366ba44 2265 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
da8c1a98 2266#define isPRINT_LC_utf8_safe(p, e) \
2366ba44 2267 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
da8c1a98
KW
2268#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2269#define isPUNCT_LC_utf8_safe(p, e) \
2366ba44 2270 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
da8c1a98 2271#define isSPACE_LC_utf8_safe(p, e) \
2366ba44 2272 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
da8c1a98 2273#define isUPPER_LC_utf8_safe(p, e) \
2366ba44 2274 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
da8c1a98 2275#define isWORDCHAR_LC_utf8_safe(p, e) \
2366ba44 2276 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
da8c1a98 2277#define isXDIGIT_LC_utf8_safe(p, e) \
2366ba44 2278 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2279
fbc19f27
KW
2280/* Macros for backwards compatibility and for completeness when the ASCII and
2281 * Latin1 values are identical */
b7d90381
KW
2282#define isALPHAU(c) isALPHA_L1(c)
2283#define isDIGIT_L1(c) isDIGIT_A(c)
2284#define isOCTAL(c) isOCTAL_A(c)
2285#define isOCTAL_L1(c) isOCTAL_A(c)
2286#define isXDIGIT_L1(c) isXDIGIT_A(c)
2287#define isALNUM(c) isWORDCHAR(c)
a377c856 2288#define isALNUM_A(c) isALNUM(c)
b7d90381
KW
2289#define isALNUMU(c) isWORDCHAR_L1(c)
2290#define isALNUM_LC(c) isWORDCHAR_LC(c)
2291#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2292#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
059703b0 2293#define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
4c1d9526 2294#define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
059703b0 2295#define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
4c1d9526 2296#define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
b7d90381
KW
2297#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2298#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2299#define isALNUMC(c) isALPHANUMERIC(c)
2300#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2301#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2302#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
059703b0 2303#define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
4c1d9526
KW
2304#define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2305#define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
fbc19f27 2306
2bd1cbf6
KW
2307/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2308 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2309 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2310 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2311 * the outlier from the block that contains the other controls, just like
2312 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2313 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2314 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2315 * etc. */
2bd1cbf6 2316#ifndef EBCDIC
75763b3a 2317# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2318#else
75763b3a
KW
2319# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2320 ((isPRINT_A(c)) \
2321 ? (UNLIKELY((c) == '?') \
2322 ? QUESTION_MARK_CTRL \
2323 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2324 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2325 ? '?' \
2326 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2327#endif
bbce6d69 2328
dea28490
JJ
2329/* Line numbers are unsigned, 32 bits. */
2330typedef U32 line_t;
e5dcd934 2331#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2332
91152fc1
DG
2333/* Helpful alias for version prescan */
2334#define is_LAX_VERSION(a,b) \
2335 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2336
2337#define is_STRICT_VERSION(a,b) \
2338 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2339
2340#define BADVERSION(a,b,c) \
2341 if (b) { \
2342 *b = c; \
2343 } \
2344 return a;
8c52afec 2345
1ce77b7d
KW
2346/* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2347 * a-f) to its numeric value without using any branches. The input is
2348 * validated only by an assert() in DEBUGGING builds.
2349 *
2350 * It works by right shifting and isolating the bit that is 0 for the digits,
2351 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2352 * position, and then to the eights position. Both are added together to form
2353 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2354 * final four bits of the input to form the correct value. */
2355#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2356 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2357 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2358 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2359
2360/* The argument is a string pointer, which is advanced. */
2361#define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
95a59cab 2362
cb27eebd
KW
2363/* Converts a character known to represent an octal digit (0-7) to its numeric
2364 * value. The input is validated only by an assert() in DEBUGGING builds. In
2365 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2366#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2367
305b8651
KW
2368/* Efficiently returns a boolean as to if two native characters are equivalent
2369 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2370 * the ALPHA in the name is to remind you of that. This is asserted() in
2371 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2372 * works (on valid input) for both non- and UTF-8-encoded bytes.
2373 *
2374 * When one of the inputs is a compile-time constant and gets folded by the
2375 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2376 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2377 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2378 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2379 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2380 * just a single 0, in the bit position where the upper- and lowercase differ.
2381 * */
305b8651
KW
2382#define isALPHA_FOLD_EQ(c1, c2) \
2383 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2384 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2385#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2386
8e84507e 2387/*
ccfc67b7
JH
2388=head1 Memory Management
2389
a02a5408 2390=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2391The XSUB-writer's interface to the C C<malloc> function.
2392
596f7718 2393Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2394
c5008215
JC
2395In 5.9.3, Newx() and friends replace the older New() API, and drops
2396the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2397themselves. This aid has been superseded by a new build option,
d10b4965 2398PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2399there for use in XS modules supporting older perls.
2400
a02a5408 2401=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2402The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2403cast. See also C<L</Newx>>.
954c1994 2404
596f7718 2405Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2406
a02a5408 2407=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2408The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2409memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2410
596f7718 2411Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2412
954c1994
GS
2413=for apidoc Am|void|Renew|void* ptr|int nitems|type
2414The XSUB-writer's interface to the C C<realloc> function.
2415
596f7718 2416Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2417
954c1994
GS
2418=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2419The XSUB-writer's interface to the C C<realloc> function, with
2420cast.
2421
596f7718 2422Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2423
49b8b560 2424=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2425The XSUB-writer's interface to the C C<free> function.
2426
596f7718 2427This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2428
954c1994
GS
2429=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2430The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2431source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2432C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2433
e90e2364 2434=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2435Like C<Move> but returns C<dest>. Useful
72d33970 2436for encouraging compilers to tail-call
e90e2364
NC
2437optimise.
2438
954c1994
GS
2439=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2440The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2441source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2442C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2443
e90e2364
NC
2444=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2445
796b6530 2446Like C<Copy> but returns C<dest>. Useful
72d33970 2447for encouraging compilers to tail-call
e90e2364
NC
2448optimise.
2449
954c1994
GS
2450=for apidoc Am|void|Zero|void* dest|int nitems|type
2451
2452The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2453destination, C<nitems> is the number of items, and C<type> is the type.
2454
e90e2364
NC
2455=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2456
72d33970
FC
2457Like C<Zero> but returns dest. Useful
2458for encouraging compilers to tail-call
e90e2364
NC
2459optimise.
2460
da5d8dbb 2461=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2462This is an architecture-independent macro to copy one structure to another.
954c1994 2463
7e337ee0
JH
2464=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2465
2466Fill up memory with a byte pattern (a byte repeated over and over
2467again) that hopefully catches attempts to access uninitialized memory.
2468
2469=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2470
2471PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2472
1c12ffb4 2473=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2474
2475PoisonWith(0xEF) for catching access to freed memory.
2476
9965345d
JH
2477=for apidoc Am|void|Poison|void* dest|int nitems|type
2478
7e337ee0 2479PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2480
2481=cut */
954c1994 2482
561b68a9
SH
2483/* Maintained for backwards-compatibility only. Use newSV() instead. */
2484#ifndef PERL_CORE
ff06c60c 2485#define NEWSV(x,len) newSV(len)
561b68a9 2486#endif
ff06c60c 2487
b7112dce 2488#define MEM_SIZE_MAX ((MEM_SIZE)-1)
19a94d75 2489
a500027b 2490#define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
e6bdf523 2491
27d5b266 2492#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2493
2494/* This expression will be constant-folded at compile time. It checks
2495 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2496 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2497 * As well as avoiding the need for a run-time check in some cases, it's
2498 * designed to avoid compiler warnings like:
2499 * comparison is always false due to limited range of data type
73e8ff00
DM
2500 * It's mathematically equivalent to
2501 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2502 */
2503
2504# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
445198b9
LM
2505 ( sizeof(MEM_SIZE) < sizeof(n) \
2506 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 2507
88f9f128 2508/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2509 * compiler warnings. We *want* to write the expression as
2510 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2511 * (for some compile-time constant C), but even when the LHS
2512 * constant-folds to false at compile-time, g++ insists on emitting
2513 * warnings about the RHS (e.g. "comparison is always false"), so instead
2514 * we write it as
e6bdf523 2515 *
d98e5cde 2516 * (cond ? n : X) > C
88f9f128 2517 *
d98e5cde
DM
2518 * where X is a constant with X > C always false. Choosing a value for X
2519 * is tricky. If 0, some compilers will complain about 0 > C always being
2520 * false; if 1, Coverity complains when n happens to be the constant value
2521 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2522 * for X and hope that nothing else whines.
e6bdf523
DM
2523 */
2524
2525# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2526 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2527 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2528
2529# define MEM_WRAP_CHECK(n,t) \
2530 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2531 && (croak_memory_wrap(),0))
2532
2533# define MEM_WRAP_CHECK_1(n,t,a) \
2534 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2535 && (Perl_croak_nocontext("%s",(a)),0))
2536
814eedc8
DD
2537/* "a" arg must be a string literal */
2538# define MEM_WRAP_CHECK_s(n,t,a) \
2539 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2540 && (Perl_croak_nocontext("" a ""),0))
2541
8b44ba4c 2542#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2543
a500027b 2544#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
2545#else
2546
410319be
NC
2547#define MEM_WRAP_CHECK(n,t)
2548#define MEM_WRAP_CHECK_1(n,t,a)
814eedc8 2549#define MEM_WRAP_CHECK_s(n,t,a)
8b44ba4c
NC
2550#define MEM_WRAP_CHECK_(n,t)
2551
a500027b 2552#define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
27d5b266 2553
1936d2a7 2554#endif
8b44ba4c 2555
fe4f188c 2556#ifdef PERL_MEM_LOG
46c6c7e2 2557/*
9f653bb5 2558 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2559 * go through functions, which are handy for debugging breakpoints, but
2560 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2561 * line number, and C function name if available) passed in. This info can
2562 * then be used for logging the calls, for which one gets a sample
73d1d973 2563 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2564 *
46c6c7e2 2565 * Known problems:
94e892a6 2566 * - not all memory allocs get logged, only those
46c6c7e2 2567 * that go through Newx() and derivatives (while all
94e892a6 2568 * Safefrees do get logged)
46c6c7e2
JH
2569 * - __FILE__ and __LINE__ do not work everywhere
2570 * - __func__ or __FUNCTION__ even less so
2571 * - I think more goes on after the perlio frees but
2572 * the thing is that STDERR gets closed (as do all
2573 * the file descriptors)
2574 * - no deeper calling stack than the caller of the Newx()
2575 * or the kind, but do I look like a C reflection/introspection
2576 * utility to you?
2577 * - the function prototypes for the logging functions
2578 * probably should maybe be somewhere else than handy.h
2579 * - one could consider inlining (macrofying) the logging
2580 * for speed, but I am too lazy
2581 * - one could imagine recording the allocations in a hash,
2582 * (keyed by the allocation address?), and maintain that
2583 * through reallocs and frees, but how to do that without
2584 * any News() happening...?
73d1d973 2585 * - lots of -Ddefines to get useful/controllable output
b953482e 2586 * - lots of ENV reads
46c6c7e2
JH
2587 */
2588
0b0ab801 2589# ifdef PERL_CORE
73d1d973 2590# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2591enum mem_log_type {
2592 MLT_ALLOC,
2593 MLT_REALLOC,
d7a2c63c
MHM
2594 MLT_FREE,
2595 MLT_NEW_SV,
2596 MLT_DEL_SV
0b0ab801
MHM
2597};
2598# endif
12754f92 2599# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2600void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2601void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2602# endif
0b0ab801
MHM
2603# endif
2604
fe4f188c
JH
2605#endif
2606
2607#ifdef PERL_MEM_LOG
d1401ee9
MHM
2608#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2609#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2610#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2611#endif
2612
2613#ifndef MEM_LOG_ALLOC
2614#define MEM_LOG_ALLOC(n,t,a) (a)
2615#endif
2616#ifndef MEM_LOG_REALLOC
2617#define MEM_LOG_REALLOC(n,t,v,a) (a)
2618#endif
2619#ifndef MEM_LOG_FREE
2620#define MEM_LOG_FREE(a) (a)
2621#endif
2622
d1401ee9
MHM
2623#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2624#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2625#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2626
2627#ifndef PERL_CORE
a02a5408
JC
2628/* pre 5.9.x compatibility */
2629#define New(x,v,n,t) Newx(v,n,t)
2630#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2631#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2632#endif
a02a5408 2633
ff68c719 2634#define Renew(v,n,t) \
d1401ee9 2635 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2636#define Renewc(v,n,t,c) \
d1401ee9 2637 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2638
2639#ifdef PERL_POISON
2640#define Safefree(d) \
06c0cc96 2641 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2642#else
fe4f188c 2643#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2644#endif
55497cff 2645
dbb57106
YO
2646/* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2647 * as it handles cases like constant string arguments without throwing warnings *
2648 * the cast is required, as is the inequality check, to avoid warnings */
45908e4d 2649#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2650
45908e4d
YO
2651
2652#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)))
2653#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)))
2654#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2655
bdd1531d 2656/* Like above, but returns a pointer to 'd' */
45908e4d
YO
2657#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)))
2658#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 2659#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2660
7e337ee0
JH
2661#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2662#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2663#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2664#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2665
caa674f3
DD
2666#ifdef PERL_POISON
2667# define PERL_POISON_EXPR(x) x
2668#else
2669# define PERL_POISON_EXPR(x)
2670#endif
2671
ff68c719 2672#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2673
1b7e2294
KW
2674/*
2675=head1 Handy Values
2676
2677=for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2678
2679Returns the number of elements in the input C array (so you want your
2680zero-based indices to be less than but not equal to).
2681
2682=for apidoc Am|void *|C_ARRAY_END|void *a
2683
2684Returns a pointer to one element past the final element of the input C array.
2685
2686=cut
2687
2688C_ARRAY_END is one past the last: half-open/half-closed range, not
2689last-inclusive range.
2690*/
622913ab 2691#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2692#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2693
2cc61e15
DD
2694#ifdef NEED_VA_COPY
2695# ifdef va_copy
2696# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2697# elif defined(__va_copy)
2698# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2699# else
07798b17 2700# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2701# endif
2702#endif
2703
472d47bc
SB
2704/* convenience debug macros */
2705#ifdef USE_ITHREADS
2706#define pTHX_FORMAT "Perl interpreter: 0x%p"
2707#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2708#define pTHX_VALUE_ (void *)my_perl,
2709#define pTHX_VALUE (void *)my_perl
2710#define pTHX__VALUE_ ,(void *)my_perl,
2711#define pTHX__VALUE ,(void *)my_perl
472d47bc 2712#else
3609ea0d 2713#define pTHX_FORMAT
472d47bc 2714#define pTHX__FORMAT
3609ea0d 2715#define pTHX_VALUE_
472d47bc 2716#define pTHX_VALUE
3609ea0d 2717#define pTHX__VALUE_
472d47bc
SB
2718#define pTHX__VALUE
2719#endif /* USE_ITHREADS */
3609ea0d 2720
2acdbac1
NC
2721/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2722 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2723 CPAN::Unpack show any users outside the core. */
2724#ifdef PERL_CORE
dc6e8de0
A
2725# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2726 "Use of " s " is deprecated")
c9680906
A
2727# define deprecate_disappears_in(when,message) \
2728 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2729 message ", and will disappear in Perl " when)
ac641426
A
2730# define deprecate_fatal_in(when,message) \
2731 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2732 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2733#endif
2734
dfff4baf
BF
2735/* Internal macros to deal with gids and uids */
2736#ifdef PERL_CORE
2737
2738# if Uid_t_size > IVSIZE
2739# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2740# define SvUID(sv) SvNV(sv)
07798b17
AC
2741# elif Uid_t_sign <= 0
2742# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2743# define SvUID(sv) SvIV(sv)
dfff4baf 2744# else
07798b17
AC
2745# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2746# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2747# endif /* Uid_t_size */
2748
2749# if Gid_t_size > IVSIZE
2750# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2751# define SvGID(sv) SvNV(sv)
07798b17
AC
2752# elif Gid_t_sign <= 0
2753# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2754# define SvGID(sv) SvIV(sv)
dfff4baf 2755# else
07798b17
AC
2756# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2757# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2758# endif /* Gid_t_size */
2759
2760#endif
2761
6a5bc5ac 2762#endif /* PERL_HANDY_H_ */
9d745869 2763
e9a8c099 2764/*
14d04a33 2765 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2766 */