This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regcomp.c, mathoms.c: Convert to use preferred macro
[perl5.git] / handy.h
CommitLineData
a0d0e21e 1/* handy.h
a687059c 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
da5d8dbb 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
a687059c 5 *
6e21c824
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8d063cd8 8 *
8d063cd8
LW
9 */
10
4650c663
KW
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
9d745869
KW
14#ifndef HANDY_H /* Guard against nested #inclusion */
15#define HANDY_H
16
85e6fe83 17#if !defined(__STDC__)
378cc40b
LW
18#ifdef NULL
19#undef NULL
20#endif
a687059c 21# define NULL 0
85e6fe83
LW
22#endif
23
24792b8d
NC
24#ifndef PERL_CORE
25# define Null(type) ((type)NULL)
954c1994
GS
26
27/*
ccfc67b7 28=head1 Handy Values
954c1994 29
3609ea0d 30=for apidoc AmU||Nullch
72d33970
FC
31Null character pointer. (No longer available when C<PERL_CORE> is
32defined.)
2307c6d0 33
954c1994 34=for apidoc AmU||Nullsv
72d33970 35Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
36
37=cut
38*/
39
24792b8d
NC
40# define Nullch Null(char*)
41# define Nullfp Null(PerlIO*)
42# define Nullsv Null(SV*)
43#endif
8d063cd8 44
641d3f0b
PP
45#ifdef TRUE
46#undef TRUE
47#endif
48#ifdef FALSE
49#undef FALSE
50#endif
51#define TRUE (1)
52#define FALSE (0)
53
cf3f0ffb
DM
54/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
55 * (compiler permitting) that casting away const-ness will give a warning;
56 * e.g.:
57 *
58 * const SV *sv = ...;
59 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
60 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
61 */
62
b1bc3f34 63#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
f065df04 64# define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
b1bc3f34
NC
65#else
66# define MUTABLE_PTR(p) ((void *) (p))
67#endif
68
a062e10d 69#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 70#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 71#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 72#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 73#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 74#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 75
f789f6a4 76#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
77# include <stdbool.h>
78# ifndef HAS_BOOL
79# define HAS_BOOL 1
80# endif
81#endif
82
8e84507e 83/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
84 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
85 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 86 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
87 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
88 g++ can be identified by __GNUG__.
89 Andy Dougherty February 2000
5d94fbed 90*/
3609ea0d 91#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 92# ifndef PERL_BOOL_AS_CHAR
5d94fbed 93# ifndef HAS_BOOL
c1d22f6b 94# define HAS_BOOL 1
5d94fbed 95# endif
f789f6a4 96# endif
5d94fbed
AD
97#endif
98
99#ifndef HAS_BOOL
f789f6a4
FC
100# ifdef bool
101# undef bool
102# endif
70d5cb32 103# define bool char
c1d22f6b 104# define HAS_BOOL 1
a687059c 105#endif
0d3e774c 106
236415e2
KW
107/* cast-to-bool. A simple (bool) cast may not do the right thing: if bool is
108 * defined as char for example, then the cast from int is
109 * implementation-defined (bool)!!(cbool) in a ternary triggers a bug in xlc on
110 * AIX */
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__
119#else
9ea40801 120# if (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
be73ebb0 121 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
46c6c7e2
JH
122# define FUNCTION__ ""
123# else
124# define FUNCTION__ __FUNCTION__ /* Common extension. */
125# endif
126#endif
127
27d4fb96
PP
128/* XXX A note on the perl source internal type system. The
129 original intent was that I32 be *exactly* 32 bits.
130
131 Currently, we only guarantee that I32 is *at least* 32 bits.
132 Specifically, if int is 64 bits, then so is I32. (This is the case
133 for the Cray.) This has the advantage of meshing nicely with
134 standard library calls (where we pass an I32 and the library is
135 expecting an int), but the disadvantage that an I32 is not 32 bits.
136 Andy Dougherty August 1996
24fef2a7 137
dc45a647
MB
138 There is no guarantee that there is *any* integral type with
139 exactly 32 bits. It is perfectly legal for a system to have
140 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 141
dc45a647
MB
142 Similarly, there is no guarantee that I16 and U16 have exactly 16
143 bits.
693762b4 144
8e84507e
NIS
145 For dealing with issues that may arise from various 32/64-bit
146 systems, we will ask Configure to check out
8175356b 147
3609ea0d
JH
148 SHORTSIZE == sizeof(short)
149 INTSIZE == sizeof(int)
150 LONGSIZE == sizeof(long)
dc45a647 151 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 152 PTRSIZE == sizeof(void *)
dc45a647
MB
153 DOUBLESIZE == sizeof(double)
154 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 155
27d4fb96
PP
156*/
157
69512466
JH
158#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
159# include <inttypes.h>
dd0eed91
JH
160# ifdef INT32_MIN_BROKEN
161# undef INT32_MIN
162# define INT32_MIN (-2147483647-1)
163# endif
164# ifdef INT64_MIN_BROKEN
165# undef INT64_MIN
166# define INT64_MIN (-9223372036854775807LL-1)
167# endif
69512466
JH
168#endif
169
8175356b
JH
170typedef I8TYPE I8;
171typedef U8TYPE U8;
172typedef I16TYPE I16;
173typedef U16TYPE U16;
174typedef I32TYPE I32;
175typedef U32TYPE U32;
16d89be8
JH
176
177#ifdef HAS_QUAD
8175356b
JH
178typedef I64TYPE I64;
179typedef U64TYPE U64;
16d89be8 180#endif
8175356b 181
9c39b21c
JH
182/* INT64_C/UINT64_C are C99 from <stdint.h> (so they will not be
183 * available in strict C89 mode), but they are nice, so let's define
184 * them if necessary. */
0d3bd0a9 185#if defined(HAS_QUAD)
9c39b21c
JH
186# undef PeRl_INT64_C
187# undef PeRl_UINT64_C
188/* Prefer the native integer types (int and long) over long long
189 * (which is not C89) and Win32-specific __int64. */
190# if QUADKIND == QUAD_IS_INT && INTSIZE == 8
191# define PeRl_INT64_C(c) (c)
192# define PeRl_UINT64_C(c) CAT2(c,U)
193# endif
194# if QUADKIND == QUAD_IS_LONG && LONGSIZE == 8
195# define PeRl_INT64_C(c) CAT2(c,L)
196# define PeRl_UINT64_C(c) CAT2(c,UL)
197# endif
198# if QUADKIND == QUAD_IS_LONG_LONG && defined(HAS_LONG_LONG)
199# define PeRl_INT64_C(c) CAT2(c,LL)
200# define PeRl_UINT64_C(c) CAT2(c,ULL)
201# endif
202# if QUADKIND == QUAD_IS___INT64
203# define PeRl_INT64_C(c) CAT2(c,I64)
204# define PeRl_UINT64_C(c) CAT2(c,UI64)
205# endif
206# ifndef PeRl_INT64_C
702c92eb
JH
207# define PeRl_INT64_C(c) ((I64)(c)) /* last resort */
208# define PeRl_UINT64_C(c) ((U64)(c))
9c39b21c
JH
209# endif
210/* In OS X the INT64_C/UINT64_C are defined with LL/ULL, which will
211 * not fly with C89-pedantic gcc, so let's undefine them first so that
212 * we can redefine them with our native integer preferring versions. */
8eafc6eb 213# if defined(PERL_DARWIN) && defined(PERL_GCC_PEDANTIC)
9c39b21c
JH
214# undef INT64_C
215# undef UINT64_C
216# endif
217# ifndef INT64_C
218# define INT64_C(c) PeRl_INT64_C(c)
219# endif
220# ifndef UINT64_C
221# define UINT64_C(c) PeRl_UINT64_C(c)
222# endif
e8c95190 223#endif
dc750f2e 224
d8668976 225#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 226
5ff3f7a4
GS
227/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
228 Please search CHAR_MAX in perl.h for further details. */
229#define U8_MAX UINT8_MAX
230#define U8_MIN UINT8_MIN
231
5ff3f7a4
GS
232#define I16_MAX INT16_MAX
233#define I16_MIN INT16_MIN
234#define U16_MAX UINT16_MAX
235#define U16_MIN UINT16_MIN
236
5ff3f7a4
GS
237#define I32_MAX INT32_MAX
238#define I32_MIN INT32_MIN
0e983133
GS
239#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
240# define U32_MAX UINT32_MAX
241#else
242# define U32_MAX 4294967295U
243#endif
5ff3f7a4
GS
244#define U32_MIN UINT32_MIN
245
246#else
247
5c9fa16e
KA
248/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
249 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
250#define U8_MAX PERL_UCHAR_MAX
251#define U8_MIN PERL_UCHAR_MIN
79072805 252
27d4fb96
PP
253#define I16_MAX PERL_SHORT_MAX
254#define I16_MIN PERL_SHORT_MIN
255#define U16_MAX PERL_USHORT_MAX
256#define U16_MIN PERL_USHORT_MIN
79072805 257
c4f23d77 258#if LONGSIZE > 4
27d4fb96
PP
259# define I32_MAX PERL_INT_MAX
260# define I32_MIN PERL_INT_MIN
261# define U32_MAX PERL_UINT_MAX
262# define U32_MIN PERL_UINT_MIN
79072805 263#else
27d4fb96
PP
264# define I32_MAX PERL_LONG_MAX
265# define I32_MIN PERL_LONG_MIN
266# define U32_MAX PERL_ULONG_MAX
267# define U32_MIN PERL_ULONG_MIN
79072805
LW
268#endif
269
5ff3f7a4
GS
270#endif
271
58a9a5d5 272/* log(2) is pretty close to 0.30103, just in case anyone is grepping for it */
fc36a67e
PP
273#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
274#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
275#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
276
88794300 277/* Unused by core; should be deprecated */
ff68c719 278#define Ctl(ch) ((ch) & 037)
8d063cd8 279
98fce2a4
KW
280#if defined(PERL_CORE) || defined(PERL_EXT)
281# ifndef MIN
282# define MIN(a,b) ((a) < (b) ? (a) : (b))
283# endif
284# ifndef MAX
285# define MAX(a,b) ((a) > (b) ? (a) : (b))
286# endif
287#endif
288
4650c663
KW
289/* This is a helper macro to avoid preprocessor issues, replaced by nothing
290 * unless under DEBUGGING, where it expands to an assert of its argument,
291 * followed by a comma (hence the comma operator). If we just used a straight
3e94db23
JH
292 * assert(), we would get a comma with nothing before it when not DEBUGGING.
293 *
294 * We also use empty definition under Coverity since the __ASSERT__
295 * checks often check for things that Really Cannot Happen, and Coverity
296 * detects that and gets all excited. */
297
298#if defined(DEBUGGING) && !defined(__COVERITY__)
0f092d08
KW
299# define __ASSERT_(statement) assert(statement),
300#else
301# define __ASSERT_(statement)
302#endif
303
3fe05580
MHM
304/*
305=head1 SV-Body Allocation
306
307=for apidoc Ama|SV*|newSVpvs|const char* s
0c395ea5 308Like C<newSVpvn>, but takes a C<NUL>-terminated literal string instead of a
30a15352 309string/length pair.
3fe05580 310
84bafc02 311=for apidoc Ama|SV*|newSVpvs_flags|const char* s|U32 flags
0c395ea5 312Like C<newSVpvn_flags>, but takes a C<NUL>-terminated literal string instead of
30a15352 313a string/length pair.
84bafc02 314
3fe05580 315=for apidoc Ama|SV*|newSVpvs_share|const char* s
0c395ea5 316Like C<newSVpvn_share>, but takes a C<NUL>-terminated literal string instead of
30a15352 317a string/length pair and omits the hash parameter.
3fe05580 318
9dcc53ea 319=for apidoc Am|void|sv_catpvs_flags|SV* sv|const char* s|I32 flags
0c395ea5 320Like C<sv_catpvn_flags>, but takes a C<NUL>-terminated literal string instead
30a15352 321of a string/length pair.
9dcc53ea
Z
322
323=for apidoc Am|void|sv_catpvs_nomg|SV* sv|const char* s
0c395ea5
KW
324Like C<sv_catpvn_nomg>, but takes a C<NUL>-terminated literal string instead of
325a string/length pair.
9dcc53ea 326
7b4e7469 327=for apidoc Am|void|sv_catpvs|SV* sv|const char* s
0c395ea5
KW
328Like C<sv_catpvn>, but takes a C<NUL>-terminated literal string instead of a
329string/length pair.
3fe05580 330
9dcc53ea 331=for apidoc Am|void|sv_catpvs_mg|SV* sv|const char* s
0c395ea5 332Like C<sv_catpvn_mg>, but takes a C<NUL>-terminated literal string instead of a
9dcc53ea
Z
333string/length pair.
334
7b4e7469 335=for apidoc Am|void|sv_setpvs|SV* sv|const char* s
0c395ea5
KW
336Like C<sv_setpvn>, but takes a C<NUL>-terminated literal string instead of a
337string/length pair.
3fe05580 338
9dcc53ea 339=for apidoc Am|void|sv_setpvs_mg|SV* sv|const char* s
0c395ea5 340Like C<sv_setpvn_mg>, but takes a C<NUL>-terminated literal string instead of a
9dcc53ea
Z
341string/length pair.
342
343=for apidoc Am|SV *|sv_setref_pvs|const char* s
0c395ea5
KW
344Like C<sv_setref_pvn>, but takes a C<NUL>-terminated literal string instead of
345a string/length pair.
9dcc53ea 346
3fe05580
MHM
347=head1 Memory Management
348
349=for apidoc Ama|char*|savepvs|const char* s
0c395ea5 350Like C<savepvn>, but takes a C<NUL>-terminated literal string instead of a
30a15352 351string/length pair.
3fe05580 352
9dcc53ea
Z
353=for apidoc Ama|char*|savesharedpvs|const char* s
354A version of C<savepvs()> which allocates the duplicate string in memory
355which is shared between threads.
356
3fe05580
MHM
357=head1 GV Functions
358
359=for apidoc Am|HV*|gv_stashpvs|const char* name|I32 create
0c395ea5
KW
360Like C<gv_stashpvn>, but takes a C<NUL>-terminated literal string instead of a
361string/length pair.
3fe05580
MHM
362
363=head1 Hash Manipulation Functions
364
365=for apidoc Am|SV**|hv_fetchs|HV* tb|const char* key|I32 lval
0c395ea5
KW
366Like C<hv_fetch>, but takes a C<NUL>-terminated literal string instead of a
367string/length pair.
3fe05580
MHM
368
369=for apidoc Am|SV**|hv_stores|HV* tb|const char* key|NULLOK SV* val
0c395ea5
KW
370Like C<hv_store>, but takes a C<NUL>-terminated literal string instead of a
371string/length pair
3fe05580
MHM
372and omits the hash parameter.
373
510966aa
Z
374=head1 Lexer interface
375
376=for apidoc Amx|void|lex_stuff_pvs|const char *pv|U32 flags
377
0c395ea5
KW
378Like L</lex_stuff_pvn>, but takes a C<NUL>-terminated literal string instead of
379a string/length pair.
510966aa 380
3fe05580
MHM
381=cut
382*/
383
926bb54c
KW
384/* concatenating with "" ensures that only literal strings are accepted as
385 * argument */
26bfd924 386#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
2efa8cc7 387
926bb54c
KW
388/* note that STR_WITH_LEN() can't be used as argument to macros or functions
389 * that under some configurations might be macros, which means that it requires
390 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
ba3a79e7
GA
391 */
392
393/* STR_WITH_LEN() shortcuts */
394#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
395#define newSVpvs_flags(str,flags) \
396 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 397#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
398#define sv_catpvs_flags(sv, str, flags) \
399 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
400#define sv_catpvs_nomg(sv, str) \
401 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
402#define sv_catpvs(sv, str) \
403 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
404#define sv_catpvs_mg(sv, str) \
405 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 406#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
407#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
408#define sv_setref_pvs(rv, classname, str) \
409 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 410#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
411#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
412#define gv_stashpvs(str, create) \
413 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
414#define gv_fetchpvs(namebeg, add, sv_type) \
415 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
416#define gv_fetchpvn(namebeg, len, add, sv_type) \
417 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
418#define sv_catxmlpvs(dsv, str, utf8) \
419 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 420
ba3a79e7 421
510966aa
Z
422#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
423
b96d8cd9
NC
424#define get_cvs(str, flags) \
425 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 426
954c1994 427/*
ccfc67b7
JH
428=head1 Miscellaneous Functions
429
954c1994 430=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
431Test two C<NUL>-terminated strings to see if they are different. Returns true
432or false.
954c1994
GS
433
434=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
435Test two C<NUL>-terminated strings to see if they are equal. Returns true or
436false.
954c1994
GS
437
438=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
439Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
440second, C<s2>. Returns true or false.
954c1994
GS
441
442=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
443Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
444equal to the second, C<s2>. Returns true or false.
954c1994
GS
445
446=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
447Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
448the second, C<s2>. Returns true or false.
954c1994
GS
449
450=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
451Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
452or equal to the second, C<s2>. Returns true or false.
954c1994
GS
453
454=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
455Test two C<NUL>-terminated strings to see if they are different. The C<len>
456parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
457wrapper for C<strncmp>).
458
459=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
460Test two C<NUL>-terminated strings to see if they are equal. The C<len>
461parameter indicates the number of bytes to compare. Returns true or false. (A
462wrapper for C<strncmp>).
954c1994 463
bd18bd40
KW
464=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
465Test two buffers (which may contain embedded C<NUL> characters, to see if they
466are equal. The C<len> parameter indicates the number of bytes to compare.
467Returns zero if equal, or non-zero if non-equal.
468
469=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
470Test two buffers (which may contain embedded C<NUL> characters, to see if they
471are not equal. The C<len> parameter indicates the number of bytes to compare.
472Returns zero if non-equal, or non-zero if equal.
473
954c1994
GS
474=cut
475*/
476
62946e08 477
8d063cd8
LW
478#define strNE(s1,s2) (strcmp(s1,s2))
479#define strEQ(s1,s2) (!strcmp(s1,s2))
480#define strLT(s1,s2) (strcmp(s1,s2) < 0)
481#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
482#define strGT(s1,s2) (strcmp(s1,s2) > 0)
483#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 484
8d063cd8
LW
485#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
486#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
378cc40b 487
62946e08
YO
488#define strNEs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1))
489#define strEQs(s1,s2) (!strncmp(s1,"" s2 "", sizeof(s2)-1))
490
36477c24
PP
491#ifdef HAS_MEMCMP
492# define memNE(s1,s2,l) (memcmp(s1,s2,l))
493# define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
494#else
495# define memNE(s1,s2,l) (bcmp(s1,s2,l))
496# define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
497#endif
498
085b7534 499/* memEQ and memNE where second comparand is a string constant */
568a785a 500#define memEQs(s1, l, s2) \
777fa2cb 501 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
568a785a
NC
502#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
503
777fa2cb
YO
504/* memEQ and memNE where second comparand is a string constant
505 * and we can assume the length of s1 is at least that of the string */
506#define _memEQs(s1, s2) \
507 (memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
508#define _memNEs(s1, s2) (memNE((s1),("" s2 ""),(sizeof(s2)-1)))
509
062b6850
KW
510#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
511#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
512#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
513#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
514
bbce6d69
PP
515/*
516 * Character classes.
517 *
518 * Unfortunately, the introduction of locales means that we
519 * can't trust isupper(), etc. to tell the truth. And when
520 * it comes to /\w+/ with tainting enabled, we *must* be able
521 * to trust our character classes.
522 *
523 * Therefore, the default tests in the text of Perl will be
524 * independent of locale. Any code that wants to depend on
525 * the current locale will use the tests that begin with "lc".
526 */
527
2304df62
AD
528#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
529# ifndef CTYPE256
530# define CTYPE256
531# endif
532#endif
533
954c1994 534/*
ccfc67b7 535
dcccc8ff 536=head1 Character classification
243effed
KW
537This section is about functions (really macros) that classify characters
538into types, such as punctuation versus alphabetic, etc. Most of these are
539analogous to regular expression character classes. (See
540L<perlrecharclass/POSIX Character Classes>.) There are several variants for
541each class. (Not all macros have all variants; each item below lists the
542ones valid for it.) None are affected by C<use bytes>, and only the ones
543with C<LC> in the name are affected by the current locale.
544
545The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
546C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
547represented by that octet is (or on non-ASCII platforms, corresponds to) an
548ASCII character in the named class based on platform, Unicode, and Perl rules.
549If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 550
c98722a4 551Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
552with no suffix C<"_A">. This variant is used to emphasize by its name that
553only ASCII-range characters can return TRUE.
4b9734bf 554
c98722a4 555Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
4b9734bf
KW
556onto the platform. That is, the code points that are ASCII are unaffected,
557since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
558as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
559true when called with the code point 0xDF, which is a word character in both
4650c663 560ASCII and EBCDIC (though it represents different characters in each).
243effed 561
c98722a4 562Variant C<isI<FOO>_uvchr> is like the C<isI<FOO>_L1> variant, but accepts any UV code
243effed
KW
563point as input. If the code point is larger than 255, Unicode rules are used
564to determine if it is in the character class. For example,
d0da05db 565C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 566WITH MACRON in Unicode, and is a word character.
243effed 567
c98722a4 568Variant C<isI<FOO>_utf8_safe> is like C<isI<FOO>_uvchr>, but is used for UTF-8
da8c1a98
KW
569encoded strings. Each call classifies one character, even if the string
570contains many. This variant takes two parameters. The first, C<p>, is a
571pointer to the first byte of the character to be classified. (Recall that it
572may take more than one byte to represent a character in UTF-8 strings.) The
573second parameter, C<e>, points to anywhere in the string beyond the first
574character, up to one byte past the end of the entire string. The suffix
575C<_safe> in the function's name indicates that it will not attempt to read
576beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
577is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
578character is malformed in some way, the program may croak, or the function may
579return FALSE, at the discretion of the implementation, and subject to change in
580future releases.
581
c98722a4 582Variant C<isI<FOO>_utf8> is like C<isI<FOO>_utf8_safe>, but takes just a single
da8c1a98 583parameter, C<p>, which has the same meaning as the corresponding parameter does
c98722a4 584in C<isI<FOO>_utf8_safe>. The function therefore can't check if it is reading
34aeb2e9 585beyond the end of the string. Starting in Perl v5.30, it will take a second
c98722a4 586parameter, becoming a synonym for C<isI<FOO>_utf8_safe>. At that time every
34aeb2e9 587program that uses it will have to be changed to successfully compile. In the
c98722a4 588meantime, the first runtime call to C<isI<FOO>_utf8> from each call point in the
34aeb2e9 589program will raise a deprecation warning, enabled by default. You can convert
c98722a4 590your program now to use C<isI<FOO>_utf8_safe>, and avoid the warnings, and get an
34aeb2e9
KW
591extra measure of protection, or you can wait until v5.30, when you'll be forced
592to add the C<e> parameter.
243effed 593
c98722a4 594Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants, but the
31f05a37
KW
595result is based on the current locale, which is what C<LC> in the name stands
596for. If Perl can determine that the current locale is a UTF-8 locale, it uses
597the published Unicode rules; otherwise, it uses the C library function that
598gives the named classification. For example, C<isDIGIT_LC()> when not in a
599UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
600returned if the input won't fit into an octet. On some platforms where the C
601library function is known to be defective, Perl changes its result to follow
602the POSIX standard's rules.
243effed 603
c98722a4
KW
604Variant C<isI<FOO>_LC_uvchr> is like C<isI<FOO>_LC>, but is defined on any UV. It
605returns the same as C<isI<FOO>_LC> for input code points less than 256, and
243effed
KW
606returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
607
c98722a4 608Variant C<isI<FOO>_LC_utf8_safe> is like C<isI<FOO>_LC_uvchr>, but is used for UTF-8
da8c1a98
KW
609encoded strings. Each call classifies one character, even if the string
610contains many. This variant takes two parameters. The first, C<p>, is a
611pointer to the first byte of the character to be classified. (Recall that it
612may take more than one byte to represent a character in UTF-8 strings.) The
613second parameter, C<e>, points to anywhere in the string beyond the first
614character, up to one byte past the end of the entire string. The suffix
615C<_safe> in the function's name indicates that it will not attempt to read
616beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
617is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
618character is malformed in some way, the program may croak, or the function may
619return FALSE, at the discretion of the implementation, and subject to change in
620future releases.
621
c98722a4 622Variant C<isI<FOO>_LC_utf8> is like C<isI<FOO>_LC_utf8_safe>, but takes just a single
da8c1a98 623parameter, C<p>, which has the same meaning as the corresponding parameter does
c98722a4 624in C<isI<FOO>_LC_utf8_safe>. The function therefore can't check if it is reading
34aeb2e9 625beyond the end of the string. Starting in Perl v5.30, it will take a second
c98722a4 626parameter, becoming a synonym for C<isI<FOO>_LC_utf8_safe>. At that time every
34aeb2e9 627program that uses it will have to be changed to successfully compile. In the
c98722a4 628meantime, the first runtime call to C<isI<FOO>_LC_utf8> from each call point in
34aeb2e9 629the program will raise a deprecation warning, enabled by default. You can
c98722a4 630convert your program now to use C<isI<FOO>_LC_utf8_safe>, and avoid the warnings,
34aeb2e9
KW
631and get an extra measure of protection, or you can wait until v5.30, when
632you'll be forced to add the C<e> parameter.
ccfc67b7 633
954c1994 634=for apidoc Am|bool|isALPHA|char ch
2787a470 635Returns a boolean indicating whether the specified character is an
6aff1f14 636alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
637See the L<top of this section|/Character classification> for an explanation of
638variants
da8c1a98
KW
639C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8_safe>,
640C<isALPHA_LC>, C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8_safe>.
8a58bdcf 641
15861f94
KW
642=for apidoc Am|bool|isALPHANUMERIC|char ch
643Returns a boolean indicating whether the specified character is a either an
6aff1f14 644alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
645See the L<top of this section|/Character classification> for an explanation of
646variants
d0da05db 647C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
da8c1a98
KW
648C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>,
649and C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 650
8a58bdcf
KW
651=for apidoc Am|bool|isASCII|char ch
652Returns a boolean indicating whether the specified character is one of the 128
243effed 653characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 654On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
655character corresponds to an ASCII character. Variants C<isASCII_A()> and
656C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
657See the L<top of this section|/Character classification> for an explanation of
658variants
da8c1a98
KW
659C<isASCII_uvchr>, C<isASCII_utf8_safe>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
660C<isASCII_LC_utf8_safe>. Note, however, that some platforms do not have the C
243effed
KW
661library routine C<isascii()>. In these cases, the variants whose names contain
662C<LC> are the same as the corresponding ones without.
663
d98532ea
KW
664Also note, that because all ASCII characters are UTF-8 invariant (meaning they
665have the exact same representation (always a single byte) whether encoded in
666UTF-8 or not), C<isASCII> will give the correct results when called with any
da8c1a98
KW
667byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8_safe>
668will work properly on any string encoded or not in UTF-8.
d98532ea 669
243effed
KW
670=for apidoc Am|bool|isBLANK|char ch
671Returns a boolean indicating whether the specified character is a
6aff1f14 672character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
673See the L<top of this section|/Character classification> for an explanation of
674variants
da8c1a98
KW
675C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8_safe>,
676C<isBLANK_LC>, C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8_safe>. Note,
677however, that some platforms do not have the C library routine
678C<isblank()>. In these cases, the variants whose names contain C<LC> are
679the same as the corresponding ones without.
243effed
KW
680
681=for apidoc Am|bool|isCNTRL|char ch
682Returns a boolean indicating whether the specified character is a
6aff1f14 683control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
684See the L<top of this section|/Character classification> for an explanation of
685variants
da8c1a98
KW
686C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8_safe>,
687C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8_safe> On EBCDIC
688platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 689
954c1994 690=for apidoc Am|bool|isDIGIT|char ch
2787a470 691Returns a boolean indicating whether the specified character is a
6aff1f14 692digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 693Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
694See the L<top of this section|/Character classification> for an explanation of
695variants
da8c1a98
KW
696C<isDIGIT_uvchr>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
697C<isDIGIT_LC_utf8_safe>.
243effed
KW
698
699=for apidoc Am|bool|isGRAPH|char ch
700Returns a boolean indicating whether the specified character is a
6aff1f14 701graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 702See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
703variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8_safe>,
704C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8_safe>.
954c1994 705
0c82b6df 706=for apidoc Am|bool|isLOWER|char ch
2787a470 707Returns a boolean indicating whether the specified character is a
6aff1f14 708lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
709See the L<top of this section|/Character classification> for an explanation of
710variants
da8c1a98
KW
711C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8_safe>,
712C<isLOWER_LC>, C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8_safe>.
0c82b6df 713
c99e91e9 714=for apidoc Am|bool|isOCTAL|char ch
2787a470 715Returns a boolean indicating whether the specified character is an
6aff1f14 716octal digit, [0-7].
243effed
KW
717The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
718C<isOCTAL>.
719
720=for apidoc Am|bool|isPUNCT|char ch
721Returns a boolean indicating whether the specified character is a
6aff1f14
KW
722punctuation character, analogous to C<m/[[:punct:]]/>.
723Note that the definition of what is punctuation isn't as
243effed
KW
724straightforward as one might desire. See L<perlrecharclass/POSIX Character
725Classes> for details.
dcccc8ff 726See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
727variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8_safe>,
728C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8_safe>.
c99e91e9 729
0c82b6df 730=for apidoc Am|bool|isSPACE|char ch
2787a470 731Returns a boolean indicating whether the specified character is a
6aff1f14 732whitespace character. This is analogous
398d098a 733to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 734this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
735locale forms of this macro (the ones with C<LC> in their names) matched
736precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
737in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
738(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
739See the L<top of this section|/Character classification> for an explanation of
740variants
da8c1a98
KW
741C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8_safe>,
742C<isSPACE_LC>, C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8_safe>.
0c82b6df 743
398d098a
KW
744=for apidoc Am|bool|isPSXSPC|char ch
745(short for Posix Space)
779cf272
KW
746Starting in 5.18, this is identical in all its forms to the
747corresponding C<isSPACE()> macros.
398d098a
KW
748The locale forms of this macro are identical to their corresponding
749C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
750non-locale forms differ from their C<isSPACE()> forms only in that the
751C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
752Otherwise they are identical. Thus this macro is analogous to what
753C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 754See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
755variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8_safe>,
756C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 757
954c1994 758=for apidoc Am|bool|isUPPER|char ch
2787a470 759Returns a boolean indicating whether the specified character is an
6aff1f14 760uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 761See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
762variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8_safe>,
763C<isUPPER_LC>, C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8_safe>.
954c1994 764
243effed 765=for apidoc Am|bool|isPRINT|char ch
8eea39dd 766Returns a boolean indicating whether the specified character is a
6aff1f14 767printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
768See the L<top of this section|/Character classification> for an explanation of
769variants
da8c1a98
KW
770C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8_safe>,
771C<isPRINT_LC>, C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8_safe>.
243effed
KW
772
773=for apidoc Am|bool|isWORDCHAR|char ch
774Returns a boolean indicating whether the specified character is a character
775that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
776in a regular expression. A word character is an alphabetic character, a
777decimal digit, a connecting punctuation character (such as an underscore), or
778a "mark" character that attaches to one of those (like some sort of accent).
779C<isALNUM()> is a synonym provided for backward compatibility, even though a
780word character includes more than the standard C language meaning of
781alphanumeric.
dcccc8ff 782See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
783variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and
784C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and
785C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
786include the platform's native underscore.
8a58bdcf
KW
787
788=for apidoc Am|bool|isXDIGIT|char ch
789Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
790digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
791and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
792See the L<top of this section|/Character classification> for an explanation of
793variants
da8c1a98
KW
794C<isXDIGIT_uvchr>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>,
795and C<isXDIGIT_LC_utf8_safe>.
243effed 796
3c3ecf18
KW
797=for apidoc Am|bool|isIDFIRST|char ch
798Returns a boolean indicating whether the specified character can be the first
799character of an identifier. This is very close to, but not quite the same as
800the official Unicode property C<XID_Start>. The difference is that this
801returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
802See the L<top of this section|/Character classification> for an explanation of
803variants
da8c1a98
KW
804C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8_safe>,
805C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18
KW
806
807=for apidoc Am|bool|isIDCONT|char ch
808Returns a boolean indicating whether the specified character can be the
809second or succeeding character of an identifier. This is very close to, but
810not quite the same as the official Unicode property C<XID_Continue>. The
811difference is that this returns true only if the input character also matches
dcccc8ff
KW
812L</isWORDCHAR>. See the L<top of this section|/Character classification> for
813an
d0da05db 814explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
da8c1a98
KW
815C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
816C<isIDCONT_LC_utf8_safe>.
3c3ecf18 817
243effed 818=head1 Miscellaneous Functions
8eea39dd 819
95a59cab 820=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 821Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
822Behaviour is only well defined when isXDIGIT(*str) is true.
823
e7c1e6c1 824=head1 Character case changing
21da7284
KW
825Perl uses "full" Unicode case mappings. This means that converting a single
826character to another case may result in a sequence of more than one character.
827For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
828character sequence C<SS>. This presents some complications The lowercase of
829all characters in the range 0..255 is a single character, and thus
830C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
831return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
832an API that does allow every possible legal result to be returned.) Likewise
833no other function that is crippled by not being able to give the correct
834results for the full range of possible inputs has been implemented here.
e7c1e6c1 835
1f607577
KW
836=for apidoc Am|U8|toUPPER|U8 ch
837Converts the specified character to uppercase. If the input is anything but an
838ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 839C<toUPPER_A> is equivalent.
954c1994 840
d0da05db
KW
841=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
842Converts the code point C<cp> to its uppercase version, and
843stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
844point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
845that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
846bytes since the uppercase version may be longer than the original character.
847
848The first code point of the uppercased version is returned
21da7284
KW
849(but note, as explained at L<the top of this section|/Character case
850changing>, that there may be more.)
1f607577
KW
851
852=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
853Converts the UTF-8 encoded character at C<p> to its uppercase version, and
854stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
855that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
856bytes since the uppercase version may be longer than the original character.
857
858The first code point of the uppercased version is returned
21da7284
KW
859(but note, as explained at L<the top of this section|/Character case
860changing>, that there may be more).
1f607577
KW
861
862The input character at C<p> is assumed to be well-formed.
863
25200305
KW
864=for apidoc Am|U8|toFOLD|U8 ch
865Converts the specified character to foldcase. If the input is anything but an
866ASCII uppercase character, that input character itself is returned. Variant
867C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 868Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 869
d0da05db
KW
870=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
871Converts the code point C<cp> to its foldcase version, and
872stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
873point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
874that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
875bytes since the foldcase version may be longer than the original character.
876
877The first code point of the foldcased version is returned
21da7284
KW
878(but note, as explained at L<the top of this section|/Character case
879changing>, that there may be more).
1f607577
KW
880
881=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
882Converts the UTF-8 encoded character at C<p> to its foldcase version, and
883stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
884that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
885bytes since the foldcase version may be longer than the original character.
886
887The first code point of the foldcased version is returned
21da7284
KW
888(but note, as explained at L<the top of this section|/Character case
889changing>, that there may be more).
1f607577
KW
890
891The input character at C<p> is assumed to be well-formed.
892
893=for apidoc Am|U8|toLOWER|U8 ch
894Converts the specified character to lowercase. If the input is anything but an
895ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 896C<toLOWER_A> is equivalent.
954c1994 897
1f607577 898=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
899Converts the specified Latin1 character to lowercase. The results are
900undefined if the input doesn't fit in a byte.
1f607577
KW
901
902=for apidoc Am|U8|toLOWER_LC|U8 ch
903Converts the specified character to lowercase using the current locale's rules,
904if possible; otherwise returns the input character itself.
905
d0da05db
KW
906=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
907Converts the code point C<cp> to its lowercase version, and
908stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
909point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
910that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
911bytes since the lowercase version may be longer than the original character.
912
913The first code point of the lowercased version is returned
21da7284
KW
914(but note, as explained at L<the top of this section|/Character case
915changing>, that there may be more).
1f607577
KW
916
917=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
918Converts the UTF-8 encoded character at C<p> to its lowercase version, and
919stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
920that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
921bytes since the lowercase version may be longer than the original character.
922
923The first code point of the lowercased version is returned
21da7284
KW
924(but note, as explained at L<the top of this section|/Character case
925changing>, that there may be more).
1f607577
KW
926
927The input character at C<p> is assumed to be well-formed.
928
25200305
KW
929=for apidoc Am|U8|toTITLE|U8 ch
930Converts the specified character to titlecase. If the input is anything but an
931ASCII lowercase character, that input character itself is returned. Variant
b7d90381 932C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 933range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 934not a concept used in locale handling, so there is no functionality for that.)
25200305 935
d0da05db
KW
936=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
937Converts the code point C<cp> to its titlecase version, and
938stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
939point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
940that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
941bytes since the titlecase version may be longer than the original character.
942
943The first code point of the titlecased version is returned
21da7284
KW
944(but note, as explained at L<the top of this section|/Character case
945changing>, that there may be more).
1f607577
KW
946
947=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
948Converts the UTF-8 encoded character at C<p> to its titlecase version, and
949stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
950that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
951bytes since the titlecase version may be longer than the original character.
952
953The first code point of the titlecased version is returned
21da7284
KW
954(but note, as explained at L<the top of this section|/Character case
955changing>, that there may be more).
1f607577
KW
956
957The input character at C<p> is assumed to be well-formed.
958
954c1994 959=cut
353c9b6f 960
d0da05db 961XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
962really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
963and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 964
8a58bdcf 965Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
966patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
967
954c1994
GS
968*/
969
7c062697
KW
970/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
971 * is known only in the perl core, and this macro can be called from outside
972 * that */
973#ifdef HAS_QUAD
974# define WIDEST_UTYPE U64TYPE
975#else
976# define WIDEST_UTYPE U32
977#endif
978
3912bc88
KW
979/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
980 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
981 * bits of information are lost even on a 64-bit machine, but to get the
982 * compiler to optimize it out if possible. This is because Configure makes
983 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
984 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
985 *
986 * For Coverity, be always true, because otherwise Coverity thinks
987 * it finds several expressions that are always true, independent
988 * of operands. Well, they are, but that is kind of the point.
220c71bf 989 */
7e75d1a1 990#ifndef __COVERITY__
6c5b02ac
KW
991 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
992 * pointer) */
993#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 994 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
995#else
996#define FITS_IN_8_BITS(c) (1)
997#endif
cf301eb7 998
41f43cc2 999#ifdef EBCDIC
b6340bd0 1000# ifndef _ALL_SOURCE
0852beac
KW
1001 /* The native libc isascii() et.al. functions return the wrong results
1002 * on at least z/OS unless this is defined. */
b6340bd0
KW
1003# error _ALL_SOURCE should probably be defined
1004# endif
41f43cc2 1005#else
0852beac
KW
1006 /* There is a simple definition of ASCII for ASCII platforms. But the
1007 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1008 * other macros below.
3f3c579d
KW
1009 *
1010 * The cast here is used instead of '(c) >= 0', because some compilers emit
1011 * a warning that that test is always true when the parameter is an
1012 * unsigned type. khw supposes that it could be written as
1013 * && ((c) == '\0' || (c) > 0)
1014 * to avoid the message, but the cast will likely avoid extra branches even
1015 * with stupid compilers.
1016 *
1017 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1018 * a pointer) */
9c903d59 1019# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1020#endif
1021
38694112
KW
1022/* Take the eight possible bit patterns of the lower 3 bits and you get the
1023 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1024 * can be ignored. If the rest match '0', we have an octal */
1025#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1026
9fb1bf9d 1027#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1028
a500dc72
KW
1029/* Character class numbers. For internal core Perl use only. The ones less
1030 * than 32 are used in PL_charclass[] and the ones up through the one that
1031 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1032 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1033 * their actual definitions are here. If that file has a name not used here,
1034 * it won't compile.
1709d539
KW
1035 *
1036 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
1037 * frequency of their use. This gives a slight edge to exiting a loop earlier
1038 * (in reginclass() in regexec.c) */
1709d539
KW
1039# define _CC_WORDCHAR 0 /* \w and [:word:] */
1040# define _CC_DIGIT 1 /* \d and [:digit:] */
1041# define _CC_ALPHA 2 /* [:alpha:] */
1042# define _CC_LOWER 3 /* [:lower:] */
1043# define _CC_UPPER 4 /* [:upper:] */
1044# define _CC_PUNCT 5 /* [:punct:] */
1045# define _CC_PRINT 6 /* [:print:] */
15861f94 1046# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1047# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1048# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1709d539 1049
b0d691b2 1050#define _FIRST_NON_SWASH_CC 10
1709d539
KW
1051/* The character classes above are implemented with swashes. The second group
1052 * (just below) contains the ones implemented without. These are also sorted
1053 * in rough order of the frequency of their use, except that \v should be last,
1054 * as it isn't a real Posix character class, and some (small) inefficiencies in
1055 * regular expression handling would be introduced by putting it in the middle
1056 * of those that are. Also, cntrl and ascii come after the others as it may be
1057 * useful to group these which have no members that match above Latin1, (or
1058 * above ASCII in the latter case) */
1059
779cf272 1060# define _CC_SPACE 10 /* \s, [:space:] */
34aeb2e9
KW
1061# define _CC_PSXSPC _CC_SPACE /* XXX Temporary, can be removed
1062 when the deprecated isFOO_utf8()
1063 functions are removed */
b0d691b2
KW
1064# define _CC_BLANK 11 /* [:blank:] */
1065# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1066# define _CC_CNTRL 13 /* [:cntrl:] */
1067# define _CC_ASCII 14 /* [:ascii:] */
1068# define _CC_VERTSPACE 15 /* \v */
1709d539 1069
a0947d7b
KW
1070# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1071
1709d539 1072/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1073 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1074# define _CC_IDFIRST 16
1075# define _CC_CHARNAME_CONT 17
1076# define _CC_NONLATIN1_FOLD 18
1077# define _CC_NONLATIN1_SIMPLE_FOLD 19
1078# define _CC_QUOTEMETA 20
1079# define _CC_NON_FINAL_FOLD 21
1080# define _CC_IS_IN_SOME_FOLD 22
1081# define _CC_MNEMONIC_CNTRL 23
073c22b3 1082
34aeb2e9
KW
1083# define _CC_IDCONT 24 /* XXX Temporary, can be removed when the deprecated
1084 isFOO_utf8() functions are removed */
1085
073c22b3
KW
1086/* This next group is only used on EBCDIC platforms, so theoretically could be
1087 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
1088# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1089# define _CC_UTF8_IS_START 29
1090# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1091# define _CC_UTF8_IS_CONTINUATION 31
1092/* Unused: 24-27
f4cdb42c
KW
1093 * If more bits are needed, one could add a second word for non-64bit
1094 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1095 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1096 * is used only for optimization (as of this writing), and differs in the
1097 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1098 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1099 * would just cause /i regexes which match them to run less efficiently.
1100 * Similarly the EBCDIC-only bits are used just for speed, and could be
1101 * replaced by other means */
96ac0975 1102
3a371f2f
KW
1103#if defined(PERL_CORE) || defined(PERL_EXT)
1104/* An enum version of the character class numbers, to help compilers
1105 * optimize */
1106typedef enum {
3a371f2f 1107 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1108 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1109 _CC_ENUM_ASCII = _CC_ASCII,
1110 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1111 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1112 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1113 _CC_ENUM_DIGIT = _CC_DIGIT,
1114 _CC_ENUM_GRAPH = _CC_GRAPH,
1115 _CC_ENUM_LOWER = _CC_LOWER,
1116 _CC_ENUM_PRINT = _CC_PRINT,
1117 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1118 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1119 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1120 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1121 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1122 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1123} _char_class_number;
1124#endif
1125
63c61c3f 1126#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1127#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1128
b7d90381
KW
1129#if defined(PERL_IN_UTF8_C) \
1130 || defined(PERL_IN_REGCOMP_C) \
1131 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1132# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1133 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1134 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1135 #error Need to adjust order of swash_property_names[]
1136# endif
1137
b44daeeb
KW
1138/* This is declared static in each of the few files that this is #defined for
1139 * to keep them from being publicly accessible. Hence there is a small amount
1140 * of wasted space */
1141
922e8cb4
KW
1142static const char* const swash_property_names[] = {
1143 "XPosixWord",
1144 "XPosixDigit",
1145 "XPosixAlpha",
1146 "XPosixLower",
1147 "XPosixUpper",
1148 "XPosixPunct",
1149 "XPosixPrint",
1150 "XPosixAlnum",
b0d691b2
KW
1151 "XPosixGraph",
1152 "Cased"
922e8cb4
KW
1153};
1154#endif
1155
6635f04f 1156START_EXTERN_C
96ac0975
NC
1157# ifdef DOINIT
1158EXTCONST U32 PL_charclass[] = {
1159# include "l1_char_class_tab.h"
1160};
1161
1162# else /* ! DOINIT */
1163EXTCONST U32 PL_charclass[];
1164# endif
6635f04f 1165END_EXTERN_C
96ac0975 1166
265c1f46 1167 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1168# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1169
1170 /* For internal core Perl use only: the base macro for defining macros like
1171 * isALPHA */
ff7ecfc3 1172# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1173 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1174
f4cdb42c
KW
1175 /* The mask for the _A versions of the macros; it just adds in the bit for
1176 * ASCII. */
1177# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1178
4650c663
KW
1179 /* For internal core Perl use only: the base macro for defining macros like
1180 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1181 * the ASCII bit are present */
b7d90381
KW
1182# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1183 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1184 == _CC_mask_A(classnum)))
f4cdb42c 1185
f4cdb42c 1186# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1187# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1188# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1189# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1190# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1191# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1192# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1193# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1194# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1195# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1196# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1197# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1198# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1199 */
d95f8b6a 1200# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1201# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1202# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1203# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1204
1205 /* continuation character for legal NAME in \N{NAME} */
1206# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1207
1208# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1209# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1210# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1211# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1212# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1213# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1214# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1215# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1216# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1217# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1218
0852beac
KW
1219# ifdef EBCDIC
1220# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1221# endif
1222
f12c0118
KW
1223 /* Participates in a single-character fold with a character above 255 */
1224# 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)))
1225
1226 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1227# 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 1228
4eeeb416 1229# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1230# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1231 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1232# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1233 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1234# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1235 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1236#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1237
1238 /* If we don't have perl.h, we are compiling a utility program. Below we
1239 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1240 * to it. Most are coded based on first principles. These are written to
74665a89 1241 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
0852beac
KW
1242# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1243# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1244# define isSPACE_A(c) (isBLANK_A(c) \
1245 || (c) == '\n' \
1246 || (c) == '\r' \
1247 || (c) == '\v' \
0852beac 1248 || (c) == '\f')
74665a89
KW
1249 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1250 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1251 * performance (if optimization isn't on), and allow the same code to be
1252 * used for both platform types */
1253# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1254 && ( (c) <= 'i' \
1255 || ((c) >= 'j' && (c) <= 'r') \
1256 || (c) >= 's'))
1257# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1258 && ( (c) <= 'I' \
1259 || ((c) >= 'J' && (c) <= 'R') \
1260 || (c) >= 'S'))
a4d7a999
KW
1261# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1262# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1263# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1264# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
74665a89
KW
1265# define isXDIGIT_A(c) (isDIGIT_A(c) \
1266 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1267 || ((c) <= 'F' && (c) >= 'A'))
74665a89
KW
1268# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1269 || (c) == '#' || (c) == '$' || (c) == '%' \
1270 || (c) == '&' || (c) == '\'' || (c) == '(' \
1271 || (c) == ')' || (c) == '*' || (c) == '+' \
1272 || (c) == ',' || (c) == '.' || (c) == '/' \
1273 || (c) == ':' || (c) == ';' || (c) == '<' \
1274 || (c) == '=' || (c) == '>' || (c) == '?' \
1275 || (c) == '@' || (c) == '[' || (c) == '\\' \
1276 || (c) == ']' || (c) == '^' || (c) == '_' \
1277 || (c) == '`' || (c) == '{' || (c) == '|' \
1278 || (c) == '}' || (c) == '~')
1279# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1280# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1281
0852beac 1282# ifdef EBCDIC
74665a89
KW
1283 /* The below is accurate for the 3 EBCDIC code pages traditionally
1284 * supported by perl. The only difference between them in the controls
1285 * is the position of \n, and that is represented symbolically below */
1286# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1287 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1288 || (c) == '\t' || (c) == '\v' \
1289 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1290 || (c) == 7 /* U+7F DEL */ \
1291 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1292 /* DLE, DC[1-3] */ \
1293 || (c) == 0x18 /* U+18 CAN */ \
1294 || (c) == 0x19 /* U+19 EOM */ \
1295 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1296 || (c) == 0x26 /* U+17 ETB */ \
1297 || (c) == 0x27 /* U+1B ESC */ \
1298 || (c) == 0x2D /* U+05 ENQ */ \
1299 || (c) == 0x2E /* U+06 ACK */ \
1300 || (c) == 0x32 /* U+16 SYN */ \
1301 || (c) == 0x37 /* U+04 EOT */ \
1302 || (c) == 0x3C /* U+14 DC4 */ \
1303 || (c) == 0x3D /* U+15 NAK */ \
1304 || (c) == 0x3F)/* U+1A SUB */
0852beac 1305# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1306# else /* isASCII is already defined for ASCII platforms, so can use that to
1307 define isCNTRL */
1308# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1309# endif
1310
3ffc8c70 1311 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1312 * during debugging, and it seems best to keep them. We may be called
1313 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1314 * do anything anyway, so make it not a problem */
1315# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1316# define NATIVE_TO_LATIN1(ch) (ch)
1317# endif
3ded5eb0
KW
1318# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1319# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1320# define isBLANK_L1(c) (isBLANK_A(c) \
1321 || (FITS_IN_8_BITS(c) \
1322 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1323# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1324# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1325# define isLOWER_L1(c) (isLOWER_A(c) \
1326 || (FITS_IN_8_BITS(c) \
1327 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1328 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1329 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1330 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1331 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1332# define isPRINT_L1(c) (isPRINT_A(c) \
1333 || (FITS_IN_8_BITS(c) \
1334 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1335# define isPUNCT_L1(c) (isPUNCT_A(c) \
1336 || (FITS_IN_8_BITS(c) \
1337 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1338 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1339 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1340 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1341 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1342 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1343 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1344# define isSPACE_L1(c) (isSPACE_A(c) \
1345 || (FITS_IN_8_BITS(c) \
1346 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1347 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1348# define isUPPER_L1(c) (isUPPER_A(c) \
1349 || (FITS_IN_8_BITS(c) \
1350 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1351 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1352 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1353# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1354# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1355# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1356 || isBLANK_L1(c) \
1357 || (c) == '-' \
1358 || (c) == '(' \
1359 || (c) == ')')
1360 /* The following are not fully accurate in the above-ASCII range. I (khw)
1361 * don't think it's necessary to be so for the purposes where this gets
1362 * compiled */
1363# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1364# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1365
1366 /* And these aren't accurate at all. They are useful only for above
1367 * Latin1, which utilities and bootstrapping don't deal with */
1368# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1369# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1370# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1371
1372 /* Many of the macros later in this file are defined in terms of these. By
1373 * implementing them with a function, which converts the class number into
1374 * a call to the desired macro, all of the later ones work. However, that
1375 * function won't be actually defined when building a utility program (no
1376 * perl.h), and so a compiler error will be generated if one is attempted
1377 * to be used. And the above-Latin1 code points require Unicode tables to
1378 * be present, something unlikely to be the case when bootstrapping */
1379# define _generic_isCC(c, classnum) \
1380 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1381# define _generic_isCC_A(c, classnum) \
1382 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1383#endif /* End of no perl.h H_PERL */
8a58bdcf 1384
e66b99e9
KW
1385#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1386#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1387#define isASCII_A(c) isASCII(c)
1388#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1389#define isBLANK(c) isBLANK_A(c)
1390#define isCNTRL(c) isCNTRL_A(c)
1391#define isDIGIT(c) isDIGIT_A(c)
1392#define isGRAPH(c) isGRAPH_A(c)
1393#define isIDFIRST(c) isIDFIRST_A(c)
1394#define isLOWER(c) isLOWER_A(c)
1395#define isPRINT(c) isPRINT_A(c)
779cf272 1396#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1397#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1398#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1399#define isPUNCT(c) isPUNCT_A(c)
1400#define isSPACE(c) isSPACE_A(c)
1401#define isUPPER(c) isUPPER_A(c)
1402#define isWORDCHAR(c) isWORDCHAR_A(c)
1403#define isXDIGIT(c) isXDIGIT_A(c)
1404
1405/* ASCII casing. These could also be written as
1406 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1407 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1408 which uses table lookup and mask instead of subtraction. (This would
1409 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1410#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1411#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1412
25200305
KW
1413/* In the ASCII range, these are equivalent to what they're here defined to be.
1414 * But by creating these definitions, other code doesn't have to be aware of
1415 * this detail */
1416#define toFOLD(c) toLOWER(c)
25200305
KW
1417#define toTITLE(c) toUPPER(c)
1418
c753c8d3
KW
1419#define toLOWER_A(c) toLOWER(c)
1420#define toUPPER_A(c) toUPPER(c)
25200305
KW
1421#define toFOLD_A(c) toFOLD(c)
1422#define toTITLE_A(c) toTITLE(c)
1a0901db 1423
4650c663 1424/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1425#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1426 ? (c) \
f4cd282c 1427 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1428#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1429
1a0901db 1430/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1431 * all mapped to one of them, and these need special handling; returns the
1432 * input itself if is out-of-range */
b2bf251f 1433#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1434 ? (c) \
f4cd282c 1435 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1436#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1437
beab9ebe
KW
1438/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1439
1440/* For internal core Perl use only: the base macro for defining macros like
1441 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1442 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1443 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1444 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1445 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1446 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1447 * code points under locale rules, use the macros based on _generic_LC_uvchr
1448 * instead of this. */
beab9ebe
KW
1449#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1450 (! FITS_IN_8_BITS(c) \
1451 ? 0 \
31f05a37
KW
1452 : IN_UTF8_CTYPE_LOCALE \
1453 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1454 : cBOOL(non_utf8))
1455
1456/* For internal core Perl use only: a helper macro for defining macros like
1457 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1458 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1459 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1460 * _generic_LC_base for more info */
1461#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1462 _generic_LC_base(c,utf8_locale_classnum, \
1463 non_utf8_func( (_LC_CAST) (c)))
1464
1465/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1466 * 'c' is the platform's native underscore character */
1467#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1468 _generic_LC_base(c, utf8_locale_classnum, \
1469 (non_utf8_func( (_LC_CAST) (c)) \
1470 || (char)(c) == '_'))
1471
1472/* These next three are also for internal core Perl use only: case-change
1473 * helper macros */
1474#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1475 ? (c) \
31f05a37
KW
1476 : (IN_UTF8_CTYPE_LOCALE) \
1477 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1478 : (cast)function((cast)(c)))
beab9ebe 1479
31f05a37
KW
1480/* Note that the result can be larger than a byte in a UTF-8 locale. It
1481 * returns a single value, so can't adequately return the upper case of LATIN
1482 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1483 * values "SS"); instead it asserts against that under DEBUGGING, and
1484 * otherwise returns its input */
beab9ebe
KW
1485#define _generic_toUPPER_LC(c, function, cast) \
1486 (! FITS_IN_8_BITS(c) \
1487 ? (c) \
31f05a37 1488 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1489 ? (cast)function((cast)(c)) \
31f05a37
KW
1490 : ((((U8)(c)) == MICRO_SIGN) \
1491 ? GREEK_CAPITAL_LETTER_MU \
1492 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1493 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1494 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1495 ? (__ASSERT_(0) (c)) \
1496 : PL_mod_latin1_uc[ (U8) (c) ])))))
1497
1498/* Note that the result can be larger than a byte in a UTF-8 locale. It
1499 * returns a single value, so can't adequately return the fold case of LATIN
1500 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1501 * values "ss"); instead it asserts against that under DEBUGGING, and
1502 * otherwise returns its input */
beab9ebe 1503#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1504 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1505 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1506 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1507 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1508 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1509
84061b6a 1510/* Use the libc versions for these if available. */
f05550c0 1511#if defined(HAS_ISASCII)
84061b6a
KW
1512# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1513#else
1514# define isASCII_LC(c) isASCII(c)
1515#endif
1516
f05550c0 1517#if defined(HAS_ISBLANK)
84061b6a 1518# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1519#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1520# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1521#endif
1522
f05550c0 1523#define _LC_CAST U8
bbce6d69 1524
f05550c0 1525#ifdef WIN32
375f5f06
KW
1526 /* The Windows functions don't bother to follow the POSIX standard, which
1527 * for example says that something can't both be a printable and a control.
1528 * But Windows treats the \t control as a printable, and does such things
1529 * as making superscripts into both digits and punctuation. This tames
1530 * these flaws by assuming that the definitions of both controls and space
1531 * are correct, and then making sure that other definitions don't have
1532 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1533 * Not all possible weirdnesses are checked for, just the ones that were
1534 * detected on actual Microsoft code pages */
1535
b7d90381
KW
1536# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1537# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1538
1539# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1540 && isALPHANUMERIC_LC(c))
1541# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1542 ! isPUNCT_LC(c))
1543# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1544 isALPHANUMERIC_LC(c))
1545# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1546# define isIDFIRST_LC(c) (((c) == '_') \
1547 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1548# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1549# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1550# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1551# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1552# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1553# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1554 && isALPHANUMERIC_LC(c))
f05550c0
BF
1555
1556# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1557# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1558# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1559
1560#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1561 /* For most other platforms */
beab9ebe 1562
f05550c0
BF
1563# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1564# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1565# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1566# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1567# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1568# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1569# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1570# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1571# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1572# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1573# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1574# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1575# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1576
1577
1578# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1579# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1580# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1581
1582#else /* The final fallback position */
1583
b7d90381
KW
1584# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1585# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1586# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1587# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1588# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1589# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1590# define isLOWER_LC(c) (isascii(c) && islower(c))
1591# define isPRINT_LC(c) (isascii(c) && isprint(c))
1592# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1593# define isSPACE_LC(c) (isascii(c) && isspace(c))
1594# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1595# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1596# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1597
1598# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1599# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1600# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1601
f05550c0 1602#endif
55204971 1603
eba68aa0
KW
1604#define isIDCONT(c) isWORDCHAR(c)
1605#define isIDCONT_A(c) isWORDCHAR_A(c)
1606#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1607#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1608#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1609
4650c663 1610/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1611 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1612 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1613 * classes where there is a macro or function 'above_latin1' that takes the
1614 * single argument 'c' and returns the desired value. These exist for those
1615 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1616 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1617 * for classes where that overhead is faster than a direct lookup.
d0da05db 1618 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1619 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1620 * have duplicate checks here, so could create versions of the macros that
1621 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1622
1623/* Note that all ignore 'use bytes' */
1e222e4f
KW
1624#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1625 ? _generic_isCC(c, classnum) \
cd500f2f 1626 : above_latin1(c))
1e222e4f
KW
1627#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
1628 ? _generic_isCC(c, classnum) \
922e8cb4 1629 : _is_uni_FOO(classnum, c))
d0da05db
KW
1630#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1631#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1632#define isASCII_uvchr(c) isASCII(c)
1633#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1634#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1635#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1636#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1637#define isIDCONT_uvchr(c) \
1638 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1639#define isIDFIRST_uvchr(c) \
1640 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
d0da05db
KW
1641#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1642#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1643
1644#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1645#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1646#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1647
1648#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1649#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1650#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1651#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1652
1653#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1654#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1655#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1656#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1657
1658/* For backwards compatibility, even though '_uni' should mean official Unicode
1659 * code points, in Perl it means native for those below 256 */
1660#define isALPHA_uni(c) isALPHA_uvchr(c)
1661#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1662#define isASCII_uni(c) isASCII_uvchr(c)
1663#define isBLANK_uni(c) isBLANK_uvchr(c)
1664#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1665#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1666#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1667#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1668#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1669#define isLOWER_uni(c) isLOWER_uvchr(c)
1670#define isPRINT_uni(c) isPRINT_uvchr(c)
1671#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1672#define isSPACE_uni(c) isSPACE_uvchr(c)
1673#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1674#define isUPPER_uni(c) isUPPER_uvchr(c)
1675#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1676#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1677#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1678#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1679#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1680#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1681#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1682
4650c663
KW
1683/* For internal core Perl use only: the base macros for defining macros like
1684 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1685 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1686 * simple class definitions; the other for more complex. These are like
d0da05db 1687 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1688#define _generic_LC_uvchr(latin1, above_latin1, c) \
1689 (c < 256 ? latin1(c) : above_latin1(c))
1690#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1691 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1692
1693#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1694#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1695 _CC_ALPHANUMERIC, c)
b7d90381
KW
1696#define isASCII_LC_uvchr(c) isASCII_LC(c)
1697#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1698 is_HORIZWS_cp_high, c)
feeab5a9 1699#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1700#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1701#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1702#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1703 _is_uni_perl_idcont, c)
b7d90381 1704#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1705 _is_uni_perl_idstart, c)
1706#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1707#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1708#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1709#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1710#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1711 is_XPERLSPACE_cp_high, c)
cd500f2f 1712#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1713#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1714 _CC_WORDCHAR, c)
b7d90381
KW
1715#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1716 is_XDIGIT_cp_high, c)
e712593e 1717
b7d90381 1718#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1719
4650c663
KW
1720/* For internal core Perl use only: the base macros for defining macros like
1721 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1722 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1723 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1724 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1725 * same representation whether utf8 or not. Note that it assumes that the utf8
1726 * has been validated, and ignores 'use bytes' */
34aeb2e9
KW
1727#define _base_generic_utf8(enum_name, name, p, use_locale ) \
1728 _is_utf8_FOO(CAT2(_CC_, enum_name), \
1729 (const U8 *) p, \
1730 "is" STRINGIFY(name) "_utf8", \
1731 "is" STRINGIFY(name) "_utf8_safe", \
1732 1, use_locale, __FILE__,__LINE__)
1733
1734#define _generic_utf8(name, p) _base_generic_utf8(name, name, p, 0)
da8c1a98
KW
1735
1736/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
1737 * they don't otherwise go out of their way to look for malformed UTF-8. If
1738 * they can return accurate results without knowing if the input is otherwise
1739 * malformed, they do so. For example isASCII is accurate in spite of any
1740 * non-length malformations because it looks only at a single byte. Likewise
1741 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
1742 * variant ones return FALSE. But, if the input has to be well-formed in order
1743 * for the results to be accurate, the macros will test and if malformed will
1744 * call a routine to die
1745 *
1746 * Except for toke.c, the macros do assume that e > p, asserting that on
1747 * DEBUGGING builds. Much code that calls these depends on this being true,
1748 * for other reasons. toke.c is treated specially as using the regular
1749 * assertion breaks it in many ways. All strings that these operate on there
1750 * are supposed to have an extra NUL character at the end, so that *e = \0. A
1751 * bunch of code in toke.c assumes that this is true, so the assertion allows
1752 * for that */
1753#ifdef PERL_IN_TOKE_C
1754# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
1755#else
1756# define _utf8_safe_assert(p,e) ((e) > (p))
1757#endif
1758
1759#define _generic_utf8_safe(classnum, p, e, above_latin1) \
1760 (__ASSERT_(_utf8_safe_assert(p, e)) \
1761 (UTF8_IS_INVARIANT(*(p))) \
1762 ? _generic_isCC(*(p), classnum) \
1763 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1764 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1765 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
1766 classnum) \
1767 : (_force_out_malformed_utf8_message( \
1768 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1769 : above_latin1))
b7d90381
KW
1770/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1771 * 'above_latin1' can be a macro */
da8c1a98
KW
1772#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
1773 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
1774#define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1775 _generic_utf8_safe(classnum, p, e, \
1776 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1777 ? (_force_out_malformed_utf8_message( \
1778 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1779 : above_latin1(p)))
4650c663 1780/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1781 * 'above_latin1' parameter */
da8c1a98
KW
1782#define _generic_swash_utf8_safe(classnum, p, e) \
1783_generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e))
922e8cb4 1784
cc8ab7c0 1785/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1786 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1787 * class is TRUE for. Hence it can skip the tests for this range.
1788 * 'above_latin1' should include its arguments */
da8c1a98
KW
1789#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
1790 (__ASSERT_(_utf8_safe_assert(p, e)) \
1791 (UTF8_IS_INVARIANT(*(p))) \
1792 ? _generic_isCC(*(p), classnum) \
1793 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1794 ? 0 /* Note that doesn't check validity for latin1 */ \
1795 : above_latin1)
1796
84238efa
KW
1797/* NOTE that some of these macros have very similar ones in regcharclass.h.
1798 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1799 * there, differing in name only by an underscore from the one here
1800 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1801 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1802 * points; the regcharclass.h ones are implemented as a series of
1803 * "if-else-if-else ..." */
1804
34aeb2e9
KW
1805#define isALPHA_utf8(p) _generic_utf8(ALPHA, p)
1806#define isALPHANUMERIC_utf8(p) _generic_utf8(ALPHANUMERIC, p)
1807#define isASCII_utf8(p) _generic_utf8(ASCII, p)
1808#define isBLANK_utf8(p) _generic_utf8(BLANK, p)
1809#define isCNTRL_utf8(p) _generic_utf8(CNTRL, p)
1810#define isDIGIT_utf8(p) _generic_utf8(DIGIT, p)
1811#define isGRAPH_utf8(p) _generic_utf8(GRAPH, p)
1812#define isIDCONT_utf8(p) _generic_utf8(IDCONT, p)
1813#define isIDFIRST_utf8(p) _generic_utf8(IDFIRST, p)
1814#define isLOWER_utf8(p) _generic_utf8(LOWER, p)
1815#define isPRINT_utf8(p) _generic_utf8(PRINT, p)
1816#define isPSXSPC_utf8(p) _generic_utf8(PSXSPC, p)
1817#define isPUNCT_utf8(p) _generic_utf8(PUNCT, p)
1818#define isSPACE_utf8(p) _generic_utf8(SPACE, p)
1819#define isUPPER_utf8(p) _generic_utf8(UPPER, p)
1820#define isVERTWS_utf8(p) _generic_utf8(VERTSPACE, p)
1821#define isWORDCHAR_utf8(p) _generic_utf8(WORDCHAR, p)
1822#define isXDIGIT_utf8(p) _generic_utf8(XDIGIT, p)
e8fa43e2 1823
da8c1a98
KW
1824#define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e)
1825#define isALPHANUMERIC_utf8_safe(p, e) \
1826 _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e)
1827#define isASCII_utf8_safe(p, e) \
1828 /* Because ASCII is invariant under utf8, the non-utf8 macro \
1829 * works */ \
1830 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
1831#define isBLANK_utf8_safe(p, e) \
1832 _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
1833
e8fa43e2
KW
1834#ifdef EBCDIC
1835 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1836 * more efficient macro instead of the more general one */
da8c1a98
KW
1837# define isCNTRL_utf8_safe(p, e) \
1838 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p))
e8fa43e2 1839#else
da8c1a98 1840# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
1841#endif
1842
da8c1a98
KW
1843#define isDIGIT_utf8_safe(p, e) \
1844 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
1845 _is_utf8_FOO_with_len(_CC_DIGIT, p, e))
1846#define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e)
1847#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
1848 _is_utf8_perl_idcont_with_len, p, e)
e5dcd934 1849
c11ff943
KW
1850/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1851 * IDFIRST is an alnum. See
c8362b00 1852 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1853 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1854 * This used to be not the XID version, but we decided to go with the more
1855 * modern Unicode definition */
da8c1a98
KW
1856#define isIDFIRST_utf8_safe(p, e) \
1857 _generic_func_utf8_safe(_CC_IDFIRST, \
1858 _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e))
1859
1860#define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e)
1861#define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e)
1862#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
1863#define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e)
1864#define isSPACE_utf8_safe(p, e) \
1865 _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
1866#define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e)
1867#define isVERTWS_utf8_safe(p, e) \
1868 _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
1869#define isWORDCHAR_utf8_safe(p, e) \
1870 _generic_swash_utf8_safe(_CC_WORDCHAR, p, e)
1871#define isXDIGIT_utf8_safe(p, e) \
1872 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
1873 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1874 ? (_force_out_malformed_utf8_message( \
1875 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1876 : is_XDIGIT_high(p)))
a0ed51b3 1877
7d2b61dd 1878#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1879#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1880#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1881#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1882
4650c663
KW
1883/* For internal core Perl use only: the base macros for defining macros like
1884 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1885 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1886 * 'macro' parameter */
34aeb2e9
KW
1887#define _generic_LC_utf8(name, p) _base_generic_utf8(name, name, p, 1)
1888
1889#define isALPHA_LC_utf8(p) _generic_LC_utf8(ALPHA, p)
1890#define isALPHANUMERIC_LC_utf8(p) _generic_LC_utf8(ALPHANUMERIC, p)
1891#define isASCII_LC_utf8(p) _generic_LC_utf8(ASCII, p)
1892#define isBLANK_LC_utf8(p) _generic_LC_utf8(BLANK, p)
1893#define isCNTRL_LC_utf8(p) _generic_LC_utf8(CNTRL, p)
1894#define isDIGIT_LC_utf8(p) _generic_LC_utf8(DIGIT, p)
1895#define isGRAPH_LC_utf8(p) _generic_LC_utf8(GRAPH, p)
1896#define isIDCONT_LC_utf8(p) _generic_LC_utf8(IDCONT, p)
1897#define isIDFIRST_LC_utf8(p) _generic_LC_utf8(IDFIRST, p)
1898#define isLOWER_LC_utf8(p) _generic_LC_utf8(LOWER, p)
1899#define isPRINT_LC_utf8(p) _generic_LC_utf8(PRINT, p)
1900#define isPSXSPC_LC_utf8(p) _generic_LC_utf8(PSXSPC, p)
1901#define isPUNCT_LC_utf8(p) _generic_LC_utf8(PUNCT, p)
1902#define isSPACE_LC_utf8(p) _generic_LC_utf8(SPACE, p)
1903#define isUPPER_LC_utf8(p) _generic_LC_utf8(UPPER, p)
1904#define isWORDCHAR_LC_utf8(p) _generic_LC_utf8(WORDCHAR, p)
1905#define isXDIGIT_LC_utf8(p) _generic_LC_utf8(XDIGIT, p)
1906
da8c1a98
KW
1907/* For internal core Perl use only: the base macros for defining macros like
1908 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
1909 * point in 'p' is within the 0-255 range, it uses locale rules from the
1910 * passed-in 'macro' parameter */
1911#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
1912 (__ASSERT_(_utf8_safe_assert(p, e)) \
1913 (UTF8_IS_INVARIANT(*(p))) \
1914 ? macro(*(p)) \
1915 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1916 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1917 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
1918 : (_force_out_malformed_utf8_message( \
1919 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1920 : above_latin1))
1921
1922#define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \
1923 _generic_LC_utf8_safe(macro, p, e, \
1924 _is_utf8_FOO_with_len(classnum, p, e))
1925
1926#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
1927 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
1928
1929#define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1930 _generic_LC_utf8_safe(classnum, p, e, \
1931 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1932 ? (_force_out_malformed_utf8_message( \
1933 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1934 : above_latin1(p)))
1935
1936#define isALPHANUMERIC_LC_utf8_safe(p, e) \
1937 _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \
1938 _CC_ALPHANUMERIC, p, e)
1939#define isALPHA_LC_utf8_safe(p, e) \
1940 _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
1941#define isASCII_LC_utf8_safe(p, e) \
1942 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
1943#define isBLANK_LC_utf8_safe(p, e) \
1944 _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
1945#define isCNTRL_LC_utf8_safe(p, e) \
1946 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
1947#define isDIGIT_LC_utf8_safe(p, e) \
1948 _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
1949#define isGRAPH_LC_utf8_safe(p, e) \
1950 _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
1951#define isIDCONT_LC_utf8_safe(p, e) \
1952 _generic_LC_func_utf8_safe(isIDCONT_LC, \
1953 _is_utf8_perl_idcont_with_len, p, e)
1954#define isIDFIRST_LC_utf8_safe(p, e) \
1955 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
1956 _is_utf8_perl_idstart_with_len, p, e)
1957#define isLOWER_LC_utf8_safe(p, e) \
1958 _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
1959#define isPRINT_LC_utf8_safe(p, e) \
1960 _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
1961#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
1962#define isPUNCT_LC_utf8_safe(p, e) \
1963 _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
1964#define isSPACE_LC_utf8_safe(p, e) \
1965 _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
1966#define isUPPER_LC_utf8_safe(p, e) \
1967 _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
1968#define isWORDCHAR_LC_utf8_safe(p, e) \
1969 _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
1970#define isXDIGIT_LC_utf8_safe(p, e) \
1971 _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 1972
fbc19f27
KW
1973/* Macros for backwards compatibility and for completeness when the ASCII and
1974 * Latin1 values are identical */
b7d90381
KW
1975#define isALPHAU(c) isALPHA_L1(c)
1976#define isDIGIT_L1(c) isDIGIT_A(c)
1977#define isOCTAL(c) isOCTAL_A(c)
1978#define isOCTAL_L1(c) isOCTAL_A(c)
1979#define isXDIGIT_L1(c) isXDIGIT_A(c)
1980#define isALNUM(c) isWORDCHAR(c)
1981#define isALNUMU(c) isWORDCHAR_L1(c)
1982#define isALNUM_LC(c) isWORDCHAR_LC(c)
1983#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 1984#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
1985#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1986#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1987#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1988#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1989#define isALNUMC(c) isALPHANUMERIC(c)
1990#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1991#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 1992#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 1993#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 1994#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1995
2bd1cbf6
KW
1996/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1997 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1998 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1999 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2000 * the outlier from the block that contains the other controls, just like
2001 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2002 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2003 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2004 * etc. */
2bd1cbf6 2005#ifndef EBCDIC
75763b3a 2006# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2007#else
75763b3a
KW
2008# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2009 ((isPRINT_A(c)) \
2010 ? (UNLIKELY((c) == '?') \
2011 ? QUESTION_MARK_CTRL \
2012 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2013 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2014 ? '?' \
2015 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2016#endif
bbce6d69 2017
dea28490
JJ
2018/* Line numbers are unsigned, 32 bits. */
2019typedef U32 line_t;
e5dcd934 2020#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2021
91152fc1
DG
2022/* Helpful alias for version prescan */
2023#define is_LAX_VERSION(a,b) \
2024 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2025
2026#define is_STRICT_VERSION(a,b) \
2027 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2028
2029#define BADVERSION(a,b,c) \
2030 if (b) { \
2031 *b = c; \
2032 } \
2033 return a;
8c52afec 2034
cb27eebd
KW
2035/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2036 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2037 * which is advanced. The input is validated only by an assert() in DEBUGGING
2038 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2039 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2040#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2041 ? (c) \
2042 : ((c) + 9))))
902008b8
KW
2043#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2044 ? (*(s)++) \
2045 : (*(s)++ + 9))))
95a59cab 2046
cb27eebd
KW
2047/* Converts a character known to represent an octal digit (0-7) to its numeric
2048 * value. The input is validated only by an assert() in DEBUGGING builds. In
2049 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2050#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2051
305b8651
KW
2052/* Efficiently returns a boolean as to if two native characters are equivalent
2053 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2054 * the ALPHA in the name is to remind you of that. This is asserted() in
2055 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2056 * works (on valid input) for both non- and UTF-8-encoded bytes.
2057 *
2058 * When one of the inputs is a compile-time constant and gets folded by the
2059 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2060 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2061 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2062 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2063 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2064 * just a single 0, in the bit position where the upper- and lowercase differ.
2065 * */
305b8651
KW
2066#define isALPHA_FOLD_EQ(c1, c2) \
2067 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2068 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2069#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2070
8e84507e 2071/*
ccfc67b7
JH
2072=head1 Memory Management
2073
a02a5408 2074=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2075The XSUB-writer's interface to the C C<malloc> function.
2076
596f7718 2077Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2078
c5008215
JC
2079In 5.9.3, Newx() and friends replace the older New() API, and drops
2080the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2081themselves. This aid has been superseded by a new build option,
d10b4965 2082PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2083there for use in XS modules supporting older perls.
2084
a02a5408 2085=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2086The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2087cast. See also C<L</Newx>>.
954c1994 2088
596f7718 2089Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2090
a02a5408 2091=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2092The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2093memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2094
596f7718 2095Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2096
954c1994
GS
2097=for apidoc Am|void|Renew|void* ptr|int nitems|type
2098The XSUB-writer's interface to the C C<realloc> function.
2099
596f7718 2100Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2101
954c1994
GS
2102=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2103The XSUB-writer's interface to the C C<realloc> function, with
2104cast.
2105
596f7718 2106Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2107
49b8b560 2108=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2109The XSUB-writer's interface to the C C<free> function.
2110
596f7718 2111This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2112
954c1994
GS
2113=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2114The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2115source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2116C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2117
e90e2364 2118=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2119Like C<Move> but returns C<dest>. Useful
72d33970 2120for encouraging compilers to tail-call
e90e2364
NC
2121optimise.
2122
954c1994
GS
2123=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2124The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2125source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2126C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2127
e90e2364
NC
2128=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2129
796b6530 2130Like C<Copy> but returns C<dest>. Useful
72d33970 2131for encouraging compilers to tail-call
e90e2364
NC
2132optimise.
2133
954c1994
GS
2134=for apidoc Am|void|Zero|void* dest|int nitems|type
2135
2136The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2137destination, C<nitems> is the number of items, and C<type> is the type.
2138
e90e2364
NC
2139=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2140
72d33970
FC
2141Like C<Zero> but returns dest. Useful
2142for encouraging compilers to tail-call
e90e2364
NC
2143optimise.
2144
da5d8dbb 2145=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2146This is an architecture-independent macro to copy one structure to another.
954c1994 2147
7e337ee0
JH
2148=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2149
2150Fill up memory with a byte pattern (a byte repeated over and over
2151again) that hopefully catches attempts to access uninitialized memory.
2152
2153=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2154
2155PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2156
1c12ffb4 2157=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2158
2159PoisonWith(0xEF) for catching access to freed memory.
2160
9965345d
JH
2161=for apidoc Am|void|Poison|void* dest|int nitems|type
2162
7e337ee0 2163PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2164
2165=cut */
954c1994 2166
561b68a9
SH
2167/* Maintained for backwards-compatibility only. Use newSV() instead. */
2168#ifndef PERL_CORE
ff06c60c 2169#define NEWSV(x,len) newSV(len)
561b68a9 2170#endif
ff06c60c 2171
19a94d75
JH
2172#define MEM_SIZE_MAX ((MEM_SIZE)~0)
2173
e6bdf523 2174
27d5b266 2175#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2176
2177/* This expression will be constant-folded at compile time. It checks
2178 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2179 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2180 * As well as avoiding the need for a run-time check in some cases, it's
2181 * designed to avoid compiler warnings like:
2182 * comparison is always false due to limited range of data type
73e8ff00
DM
2183 * It's mathematically equivalent to
2184 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2185 */
2186
2187# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
004073ba 2188 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
e6bdf523 2189
88f9f128 2190/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2191 * compiler warnings. We *want* to write the expression as
2192 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2193 * (for some compile-time constant C), but even when the LHS
2194 * constant-folds to false at compile-time, g++ insists on emitting
2195 * warnings about the RHS (e.g. "comparison is always false"), so instead
2196 * we write it as
e6bdf523 2197 *
d98e5cde 2198 * (cond ? n : X) > C
88f9f128 2199 *
d98e5cde
DM
2200 * where X is a constant with X > C always false. Choosing a value for X
2201 * is tricky. If 0, some compilers will complain about 0 > C always being
2202 * false; if 1, Coverity complains when n happens to be the constant value
2203 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2204 * for X and hope that nothing else whines.
e6bdf523
DM
2205 */
2206
2207# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2208 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2209 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2210
2211# define MEM_WRAP_CHECK(n,t) \
2212 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2213 && (croak_memory_wrap(),0))
2214
2215# define MEM_WRAP_CHECK_1(n,t,a) \
2216 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2217 && (Perl_croak_nocontext("%s",(a)),0))
2218
8b44ba4c 2219#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2220
d1decf2b 2221#define PERL_STRLEN_ROUNDUP(n) ((void)(((n) > MEM_SIZE_MAX - 2 * PERL_STRLEN_ROUNDUP_QUANTUM) ? (croak_memory_wrap(),0):0),((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266
JH
2222#else
2223
410319be
NC
2224#define MEM_WRAP_CHECK(n,t)
2225#define MEM_WRAP_CHECK_1(n,t,a)
2226#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
2227#define MEM_WRAP_CHECK_(n,t)
2228
6fff79ce 2229#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 2230
1936d2a7 2231#endif
8b44ba4c 2232
fe4f188c 2233#ifdef PERL_MEM_LOG
46c6c7e2 2234/*
9f653bb5 2235 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2236 * go through functions, which are handy for debugging breakpoints, but
2237 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2238 * line number, and C function name if available) passed in. This info can
2239 * then be used for logging the calls, for which one gets a sample
73d1d973 2240 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2241 *
46c6c7e2 2242 * Known problems:
94e892a6 2243 * - not all memory allocs get logged, only those
46c6c7e2 2244 * that go through Newx() and derivatives (while all
94e892a6 2245 * Safefrees do get logged)
46c6c7e2
JH
2246 * - __FILE__ and __LINE__ do not work everywhere
2247 * - __func__ or __FUNCTION__ even less so
2248 * - I think more goes on after the perlio frees but
2249 * the thing is that STDERR gets closed (as do all
2250 * the file descriptors)
2251 * - no deeper calling stack than the caller of the Newx()
2252 * or the kind, but do I look like a C reflection/introspection
2253 * utility to you?
2254 * - the function prototypes for the logging functions
2255 * probably should maybe be somewhere else than handy.h
2256 * - one could consider inlining (macrofying) the logging
2257 * for speed, but I am too lazy
2258 * - one could imagine recording the allocations in a hash,
2259 * (keyed by the allocation address?), and maintain that
2260 * through reallocs and frees, but how to do that without
2261 * any News() happening...?
73d1d973 2262 * - lots of -Ddefines to get useful/controllable output
b953482e 2263 * - lots of ENV reads
46c6c7e2
JH
2264 */
2265
0b0ab801 2266# ifdef PERL_CORE
73d1d973 2267# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2268enum mem_log_type {
2269 MLT_ALLOC,
2270 MLT_REALLOC,
d7a2c63c
MHM
2271 MLT_FREE,
2272 MLT_NEW_SV,
2273 MLT_DEL_SV
0b0ab801
MHM
2274};
2275# endif
12754f92 2276# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2277void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2278void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2279# endif
0b0ab801
MHM
2280# endif
2281
fe4f188c
JH
2282#endif
2283
2284#ifdef PERL_MEM_LOG
d1401ee9
MHM
2285#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2286#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2287#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2288#endif
2289
2290#ifndef MEM_LOG_ALLOC
2291#define MEM_LOG_ALLOC(n,t,a) (a)
2292#endif
2293#ifndef MEM_LOG_REALLOC
2294#define MEM_LOG_REALLOC(n,t,v,a) (a)
2295#endif
2296#ifndef MEM_LOG_FREE
2297#define MEM_LOG_FREE(a) (a)
2298#endif
2299
d1401ee9
MHM
2300#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2301#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2302#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2303
2304#ifndef PERL_CORE
a02a5408
JC
2305/* pre 5.9.x compatibility */
2306#define New(x,v,n,t) Newx(v,n,t)
2307#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2308#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2309#endif
a02a5408 2310
ff68c719 2311#define Renew(v,n,t) \
d1401ee9 2312 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2313#define Renewc(v,n,t,c) \
d1401ee9 2314 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2315
2316#ifdef PERL_POISON
2317#define Safefree(d) \
06c0cc96 2318 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2319#else
fe4f188c 2320#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2321#endif
55497cff 2322
8b44ba4c
NC
2323#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2324#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2325#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2326
8b44ba4c
NC
2327#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2328#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2329#ifdef HAS_MEMSET
8b44ba4c 2330#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2331#else
8b44ba4c
NC
2332/* Using bzero(), which returns void. */
2333#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2334#endif
2335
7e337ee0
JH
2336#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2337#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2338#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2339#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2340
caa674f3
DD
2341#ifdef PERL_POISON
2342# define PERL_POISON_EXPR(x) x
2343#else
2344# define PERL_POISON_EXPR(x)
2345#endif
2346
2304df62 2347#ifdef USE_STRUCT_COPY
ff68c719 2348#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2349#else
2350#define StructCopy(s,d,t) Copy(s,d,1,t)
2351#endif
2cc61e15 2352
c3caa5c3
JH
2353/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2354 * want your zero-based indices to be less than but not equal to).
2355 *
2356 * C_ARRAY_END is one past the last: half-open/half-closed range,
2357 * not last-inclusive range. */
622913ab 2358#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2359#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2360
2cc61e15
DD
2361#ifdef NEED_VA_COPY
2362# ifdef va_copy
2363# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2364# else
a1866d1b
JH
2365# if defined(__va_copy)
2366# define Perl_va_copy(s, d) __va_copy(d, s)
2367# else
2368# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2369# endif
2cc61e15
DD
2370# endif
2371#endif
2372
472d47bc
SB
2373/* convenience debug macros */
2374#ifdef USE_ITHREADS
2375#define pTHX_FORMAT "Perl interpreter: 0x%p"
2376#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2377#define pTHX_VALUE_ (void *)my_perl,
2378#define pTHX_VALUE (void *)my_perl
2379#define pTHX__VALUE_ ,(void *)my_perl,
2380#define pTHX__VALUE ,(void *)my_perl
472d47bc 2381#else
3609ea0d 2382#define pTHX_FORMAT
472d47bc 2383#define pTHX__FORMAT
3609ea0d 2384#define pTHX_VALUE_
472d47bc 2385#define pTHX_VALUE
3609ea0d 2386#define pTHX__VALUE_
472d47bc
SB
2387#define pTHX__VALUE
2388#endif /* USE_ITHREADS */
3609ea0d 2389
2acdbac1
NC
2390/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2391 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2392 CPAN::Unpack show any users outside the core. */
2393#ifdef PERL_CORE
b7d90381
KW
2394# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2395 "Use of " s " is deprecated")
2acdbac1
NC
2396#endif
2397
dfff4baf
BF
2398/* Internal macros to deal with gids and uids */
2399#ifdef PERL_CORE
2400
2401# if Uid_t_size > IVSIZE
2402# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2403# define SvUID(sv) SvNV(sv)
2404# else
2405# if Uid_t_sign <= 0
2406# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2407# define SvUID(sv) SvIV(sv)
2408# else
2409# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2410# define SvUID(sv) SvUV(sv)
2411# endif
2412# endif /* Uid_t_size */
2413
2414# if Gid_t_size > IVSIZE
2415# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2416# define SvGID(sv) SvNV(sv)
2417# else
2418# if Gid_t_sign <= 0
2419# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2420# define SvGID(sv) SvIV(sv)
2421# else
2422# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2423# define SvGID(sv) SvUV(sv)
2424# endif
2425# endif /* Gid_t_size */
2426
2427#endif
2428
9d745869
KW
2429#endif /* HANDY_H */
2430
e9a8c099 2431/*
14d04a33 2432 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2433 */