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