This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Add toFOO_utf8_safe() macros
[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 851
a239b1e2
KW
852=for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
853Converts the first UTF-8 encoded character in the sequence starting at C<p> and
854extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
855stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
856that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
857bytes since the uppercase version may be longer than the original character.
858
859The first code point of the uppercased version is returned
21da7284
KW
860(but note, as explained at L<the top of this section|/Character case
861changing>, that there may be more).
1f607577 862
a239b1e2
KW
863The suffix C<_safe> in the function's name indicates that it will not attempt
864to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
865true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
866input character is malformed in some way, the program may croak, or the
867function may return the REPLACEMENT CHARACTER, at the discretion of the
868implementation, and subject to change in future releases.
869
870=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
871This is like C<L</toUPPER_utf8_safe>>, but doesn't have the C<e>
872parameter The function therefore can't check if it is reading
873beyond the end of the string.
1f607577 874
25200305
KW
875=for apidoc Am|U8|toFOLD|U8 ch
876Converts the specified character to foldcase. If the input is anything but an
877ASCII uppercase character, that input character itself is returned. Variant
878C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 879Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 880
d0da05db
KW
881=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
882Converts the code point C<cp> to its foldcase version, and
883stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
884point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
885that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
886bytes since the foldcase version may be longer than the original character.
887
888The first code point of the foldcased version is returned
21da7284
KW
889(but note, as explained at L<the top of this section|/Character case
890changing>, that there may be more).
1f607577 891
a239b1e2
KW
892=for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
893Converts the first UTF-8 encoded character in the sequence starting at C<p> and
894extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
895stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
896that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
897bytes since the foldcase version may be longer than the original character.
898
899The first code point of the foldcased version is returned
21da7284
KW
900(but note, as explained at L<the top of this section|/Character case
901changing>, that there may be more).
1f607577 902
a239b1e2
KW
903The suffix C<_safe> in the function's name indicates that it will not attempt
904to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
905true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
906input character is malformed in some way, the program may croak, or the
907function may return the REPLACEMENT CHARACTER, at the discretion of the
908implementation, and subject to change in future releases.
909
910=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
911This is like C<L</toFOLD_utf8_safe>>, but doesn't have the C<e>
912parameter The function therefore can't check if it is reading
913beyond the end of the string.
1f607577
KW
914
915=for apidoc Am|U8|toLOWER|U8 ch
916Converts the specified character to lowercase. If the input is anything but an
917ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 918C<toLOWER_A> is equivalent.
954c1994 919
1f607577 920=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
921Converts the specified Latin1 character to lowercase. The results are
922undefined if the input doesn't fit in a byte.
1f607577
KW
923
924=for apidoc Am|U8|toLOWER_LC|U8 ch
925Converts the specified character to lowercase using the current locale's rules,
926if possible; otherwise returns the input character itself.
927
d0da05db
KW
928=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
929Converts the code point C<cp> to its lowercase version, and
930stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
931point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
932that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
933bytes since the lowercase version may be longer than the original character.
934
935The first code point of the lowercased version is returned
21da7284
KW
936(but note, as explained at L<the top of this section|/Character case
937changing>, that there may be more).
1f607577 938
a239b1e2
KW
939
940=for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
941Converts the first UTF-8 encoded character in the sequence starting at C<p> and
942extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
943stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
944that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
945bytes since the lowercase version may be longer than the original character.
946
947The first code point of the lowercased version is returned
21da7284
KW
948(but note, as explained at L<the top of this section|/Character case
949changing>, that there may be more).
1f607577 950
a239b1e2
KW
951The suffix C<_safe> in the function's name indicates that it will not attempt
952to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
953true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
954input character is malformed in some way, the program may croak, or the
955function may return the REPLACEMENT CHARACTER, at the discretion of the
956implementation, and subject to change in future releases.
957
958=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
959This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
960parameter The function therefore can't check if it is reading
961beyond the end of the string.
1f607577 962
25200305
KW
963=for apidoc Am|U8|toTITLE|U8 ch
964Converts the specified character to titlecase. If the input is anything but an
965ASCII lowercase character, that input character itself is returned. Variant
b7d90381 966C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 967range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 968not a concept used in locale handling, so there is no functionality for that.)
25200305 969
d0da05db
KW
970=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
971Converts the code point C<cp> to its titlecase version, and
972stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
973point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
974that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
975bytes since the titlecase version may be longer than the original character.
976
977The first code point of the titlecased version is returned
21da7284
KW
978(but note, as explained at L<the top of this section|/Character case
979changing>, that there may be more).
1f607577 980
a239b1e2
KW
981=for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
982Converts the first UTF-8 encoded character in the sequence starting at C<p> and
983extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
984stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
985that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
986bytes since the titlecase version may be longer than the original character.
987
988The first code point of the titlecased version is returned
21da7284
KW
989(but note, as explained at L<the top of this section|/Character case
990changing>, that there may be more).
1f607577 991
a239b1e2
KW
992The suffix C<_safe> in the function's name indicates that it will not attempt
993to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
994true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
995input character is malformed in some way, the program may croak, or the
996function may return the REPLACEMENT CHARACTER, at the discretion of the
997implementation, and subject to change in future releases.
998
999=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
1000This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1001parameter The function therefore can't check if it is reading
1002beyond the end of the string.
1f607577 1003
954c1994 1004=cut
353c9b6f 1005
d0da05db 1006XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1007really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1008and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1009
8a58bdcf 1010Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1011patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1012
954c1994
GS
1013*/
1014
7c062697
KW
1015/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
1016 * is known only in the perl core, and this macro can be called from outside
1017 * that */
1018#ifdef HAS_QUAD
1019# define WIDEST_UTYPE U64TYPE
1020#else
1021# define WIDEST_UTYPE U32
1022#endif
1023
3912bc88
KW
1024/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1025 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1026 * bits of information are lost even on a 64-bit machine, but to get the
1027 * compiler to optimize it out if possible. This is because Configure makes
1028 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1029 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1030 *
1031 * For Coverity, be always true, because otherwise Coverity thinks
1032 * it finds several expressions that are always true, independent
1033 * of operands. Well, they are, but that is kind of the point.
220c71bf 1034 */
7e75d1a1 1035#ifndef __COVERITY__
6c5b02ac
KW
1036 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1037 * pointer) */
1038#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1039 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1040#else
1041#define FITS_IN_8_BITS(c) (1)
1042#endif
cf301eb7 1043
41f43cc2 1044#ifdef EBCDIC
b6340bd0 1045# ifndef _ALL_SOURCE
0852beac
KW
1046 /* The native libc isascii() et.al. functions return the wrong results
1047 * on at least z/OS unless this is defined. */
b6340bd0
KW
1048# error _ALL_SOURCE should probably be defined
1049# endif
41f43cc2 1050#else
0852beac
KW
1051 /* There is a simple definition of ASCII for ASCII platforms. But the
1052 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1053 * other macros below.
3f3c579d
KW
1054 *
1055 * The cast here is used instead of '(c) >= 0', because some compilers emit
1056 * a warning that that test is always true when the parameter is an
1057 * unsigned type. khw supposes that it could be written as
1058 * && ((c) == '\0' || (c) > 0)
1059 * to avoid the message, but the cast will likely avoid extra branches even
1060 * with stupid compilers.
1061 *
1062 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1063 * a pointer) */
9c903d59 1064# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1065#endif
1066
38694112
KW
1067/* Take the eight possible bit patterns of the lower 3 bits and you get the
1068 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1069 * can be ignored. If the rest match '0', we have an octal */
1070#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1071
9fb1bf9d 1072#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1073
a500dc72
KW
1074/* Character class numbers. For internal core Perl use only. The ones less
1075 * than 32 are used in PL_charclass[] and the ones up through the one that
1076 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1077 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1078 * their actual definitions are here. If that file has a name not used here,
1079 * it won't compile.
1709d539
KW
1080 *
1081 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
1082 * frequency of their use. This gives a slight edge to exiting a loop earlier
1083 * (in reginclass() in regexec.c) */
1709d539
KW
1084# define _CC_WORDCHAR 0 /* \w and [:word:] */
1085# define _CC_DIGIT 1 /* \d and [:digit:] */
1086# define _CC_ALPHA 2 /* [:alpha:] */
1087# define _CC_LOWER 3 /* [:lower:] */
1088# define _CC_UPPER 4 /* [:upper:] */
1089# define _CC_PUNCT 5 /* [:punct:] */
1090# define _CC_PRINT 6 /* [:print:] */
15861f94 1091# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1092# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1093# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1709d539 1094
b0d691b2 1095#define _FIRST_NON_SWASH_CC 10
1709d539
KW
1096/* The character classes above are implemented with swashes. The second group
1097 * (just below) contains the ones implemented without. These are also sorted
1098 * in rough order of the frequency of their use, except that \v should be last,
1099 * as it isn't a real Posix character class, and some (small) inefficiencies in
1100 * regular expression handling would be introduced by putting it in the middle
1101 * of those that are. Also, cntrl and ascii come after the others as it may be
1102 * useful to group these which have no members that match above Latin1, (or
1103 * above ASCII in the latter case) */
1104
779cf272 1105# define _CC_SPACE 10 /* \s, [:space:] */
34aeb2e9
KW
1106# define _CC_PSXSPC _CC_SPACE /* XXX Temporary, can be removed
1107 when the deprecated isFOO_utf8()
1108 functions are removed */
b0d691b2
KW
1109# define _CC_BLANK 11 /* [:blank:] */
1110# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1111# define _CC_CNTRL 13 /* [:cntrl:] */
1112# define _CC_ASCII 14 /* [:ascii:] */
1113# define _CC_VERTSPACE 15 /* \v */
1709d539 1114
a0947d7b
KW
1115# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1116
1709d539 1117/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1118 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1119# define _CC_IDFIRST 16
1120# define _CC_CHARNAME_CONT 17
1121# define _CC_NONLATIN1_FOLD 18
1122# define _CC_NONLATIN1_SIMPLE_FOLD 19
1123# define _CC_QUOTEMETA 20
1124# define _CC_NON_FINAL_FOLD 21
1125# define _CC_IS_IN_SOME_FOLD 22
1126# define _CC_MNEMONIC_CNTRL 23
073c22b3 1127
34aeb2e9
KW
1128# define _CC_IDCONT 24 /* XXX Temporary, can be removed when the deprecated
1129 isFOO_utf8() functions are removed */
1130
073c22b3
KW
1131/* This next group is only used on EBCDIC platforms, so theoretically could be
1132 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
1133# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1134# define _CC_UTF8_IS_START 29
1135# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1136# define _CC_UTF8_IS_CONTINUATION 31
1137/* Unused: 24-27
f4cdb42c
KW
1138 * If more bits are needed, one could add a second word for non-64bit
1139 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1140 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1141 * is used only for optimization (as of this writing), and differs in the
1142 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1143 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1144 * would just cause /i regexes which match them to run less efficiently.
1145 * Similarly the EBCDIC-only bits are used just for speed, and could be
1146 * replaced by other means */
96ac0975 1147
3a371f2f
KW
1148#if defined(PERL_CORE) || defined(PERL_EXT)
1149/* An enum version of the character class numbers, to help compilers
1150 * optimize */
1151typedef enum {
3a371f2f 1152 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1153 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1154 _CC_ENUM_ASCII = _CC_ASCII,
1155 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1156 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1157 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1158 _CC_ENUM_DIGIT = _CC_DIGIT,
1159 _CC_ENUM_GRAPH = _CC_GRAPH,
1160 _CC_ENUM_LOWER = _CC_LOWER,
1161 _CC_ENUM_PRINT = _CC_PRINT,
1162 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1163 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1164 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1165 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1166 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1167 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1168} _char_class_number;
1169#endif
1170
63c61c3f 1171#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1172#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1173
b7d90381
KW
1174#if defined(PERL_IN_UTF8_C) \
1175 || defined(PERL_IN_REGCOMP_C) \
1176 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1177# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1178 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1179 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1180 #error Need to adjust order of swash_property_names[]
1181# endif
1182
b44daeeb
KW
1183/* This is declared static in each of the few files that this is #defined for
1184 * to keep them from being publicly accessible. Hence there is a small amount
1185 * of wasted space */
1186
922e8cb4
KW
1187static const char* const swash_property_names[] = {
1188 "XPosixWord",
1189 "XPosixDigit",
1190 "XPosixAlpha",
1191 "XPosixLower",
1192 "XPosixUpper",
1193 "XPosixPunct",
1194 "XPosixPrint",
1195 "XPosixAlnum",
b0d691b2
KW
1196 "XPosixGraph",
1197 "Cased"
922e8cb4
KW
1198};
1199#endif
1200
6635f04f 1201START_EXTERN_C
96ac0975
NC
1202# ifdef DOINIT
1203EXTCONST U32 PL_charclass[] = {
1204# include "l1_char_class_tab.h"
1205};
1206
1207# else /* ! DOINIT */
1208EXTCONST U32 PL_charclass[];
1209# endif
6635f04f 1210END_EXTERN_C
96ac0975 1211
265c1f46 1212 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1213# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1214
1215 /* For internal core Perl use only: the base macro for defining macros like
1216 * isALPHA */
ff7ecfc3 1217# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1218 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1219
f4cdb42c
KW
1220 /* The mask for the _A versions of the macros; it just adds in the bit for
1221 * ASCII. */
1222# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1223
4650c663
KW
1224 /* For internal core Perl use only: the base macro for defining macros like
1225 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1226 * the ASCII bit are present */
b7d90381
KW
1227# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1228 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1229 == _CC_mask_A(classnum)))
f4cdb42c 1230
f4cdb42c 1231# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1232# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1233# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1234# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1235# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1236# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1237# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1238# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1239# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1240# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1241# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1242# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1243# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1244 */
d95f8b6a 1245# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1246# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1247# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1248# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1249
1250 /* continuation character for legal NAME in \N{NAME} */
1251# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1252
1253# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1254# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1255# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1256# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1257# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1258# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1259# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1260# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1261# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1262# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1263
0852beac
KW
1264# ifdef EBCDIC
1265# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1266# endif
1267
f12c0118
KW
1268 /* Participates in a single-character fold with a character above 255 */
1269# 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)))
1270
1271 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1272# 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 1273
4eeeb416 1274# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1275# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1276 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1277# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1278 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1279# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1280 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1281#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1282
1283 /* If we don't have perl.h, we are compiling a utility program. Below we
1284 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1285 * to it. Most are coded based on first principles. These are written to
74665a89 1286 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
0852beac
KW
1287# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1288# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1289# define isSPACE_A(c) (isBLANK_A(c) \
1290 || (c) == '\n' \
1291 || (c) == '\r' \
1292 || (c) == '\v' \
0852beac 1293 || (c) == '\f')
74665a89
KW
1294 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1295 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1296 * performance (if optimization isn't on), and allow the same code to be
1297 * used for both platform types */
1298# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1299 && ( (c) <= 'i' \
1300 || ((c) >= 'j' && (c) <= 'r') \
1301 || (c) >= 's'))
1302# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1303 && ( (c) <= 'I' \
1304 || ((c) >= 'J' && (c) <= 'R') \
1305 || (c) >= 'S'))
a4d7a999
KW
1306# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1307# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1308# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1309# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
74665a89
KW
1310# define isXDIGIT_A(c) (isDIGIT_A(c) \
1311 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1312 || ((c) <= 'F' && (c) >= 'A'))
74665a89
KW
1313# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1314 || (c) == '#' || (c) == '$' || (c) == '%' \
1315 || (c) == '&' || (c) == '\'' || (c) == '(' \
1316 || (c) == ')' || (c) == '*' || (c) == '+' \
1317 || (c) == ',' || (c) == '.' || (c) == '/' \
1318 || (c) == ':' || (c) == ';' || (c) == '<' \
1319 || (c) == '=' || (c) == '>' || (c) == '?' \
1320 || (c) == '@' || (c) == '[' || (c) == '\\' \
1321 || (c) == ']' || (c) == '^' || (c) == '_' \
1322 || (c) == '`' || (c) == '{' || (c) == '|' \
1323 || (c) == '}' || (c) == '~')
1324# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1325# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1326
0852beac 1327# ifdef EBCDIC
74665a89
KW
1328 /* The below is accurate for the 3 EBCDIC code pages traditionally
1329 * supported by perl. The only difference between them in the controls
1330 * is the position of \n, and that is represented symbolically below */
1331# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1332 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1333 || (c) == '\t' || (c) == '\v' \
1334 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1335 || (c) == 7 /* U+7F DEL */ \
1336 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1337 /* DLE, DC[1-3] */ \
1338 || (c) == 0x18 /* U+18 CAN */ \
1339 || (c) == 0x19 /* U+19 EOM */ \
1340 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1341 || (c) == 0x26 /* U+17 ETB */ \
1342 || (c) == 0x27 /* U+1B ESC */ \
1343 || (c) == 0x2D /* U+05 ENQ */ \
1344 || (c) == 0x2E /* U+06 ACK */ \
1345 || (c) == 0x32 /* U+16 SYN */ \
1346 || (c) == 0x37 /* U+04 EOT */ \
1347 || (c) == 0x3C /* U+14 DC4 */ \
1348 || (c) == 0x3D /* U+15 NAK */ \
1349 || (c) == 0x3F)/* U+1A SUB */
0852beac 1350# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1351# else /* isASCII is already defined for ASCII platforms, so can use that to
1352 define isCNTRL */
1353# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1354# endif
1355
3ffc8c70 1356 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1357 * during debugging, and it seems best to keep them. We may be called
1358 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1359 * do anything anyway, so make it not a problem */
1360# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1361# define NATIVE_TO_LATIN1(ch) (ch)
1362# endif
3ded5eb0
KW
1363# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1364# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1365# define isBLANK_L1(c) (isBLANK_A(c) \
1366 || (FITS_IN_8_BITS(c) \
1367 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1368# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1369# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1370# define isLOWER_L1(c) (isLOWER_A(c) \
1371 || (FITS_IN_8_BITS(c) \
1372 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1373 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1374 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1375 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1376 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1377# define isPRINT_L1(c) (isPRINT_A(c) \
1378 || (FITS_IN_8_BITS(c) \
1379 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1380# define isPUNCT_L1(c) (isPUNCT_A(c) \
1381 || (FITS_IN_8_BITS(c) \
1382 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1383 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1384 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1385 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1386 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1387 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1388 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1389# define isSPACE_L1(c) (isSPACE_A(c) \
1390 || (FITS_IN_8_BITS(c) \
1391 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1392 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1393# define isUPPER_L1(c) (isUPPER_A(c) \
1394 || (FITS_IN_8_BITS(c) \
1395 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1396 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1397 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1398# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1399# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1400# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1401 || isBLANK_L1(c) \
1402 || (c) == '-' \
1403 || (c) == '(' \
1404 || (c) == ')')
1405 /* The following are not fully accurate in the above-ASCII range. I (khw)
1406 * don't think it's necessary to be so for the purposes where this gets
1407 * compiled */
1408# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1409# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1410
1411 /* And these aren't accurate at all. They are useful only for above
1412 * Latin1, which utilities and bootstrapping don't deal with */
1413# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1414# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1415# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1416
1417 /* Many of the macros later in this file are defined in terms of these. By
1418 * implementing them with a function, which converts the class number into
1419 * a call to the desired macro, all of the later ones work. However, that
1420 * function won't be actually defined when building a utility program (no
1421 * perl.h), and so a compiler error will be generated if one is attempted
1422 * to be used. And the above-Latin1 code points require Unicode tables to
1423 * be present, something unlikely to be the case when bootstrapping */
1424# define _generic_isCC(c, classnum) \
1425 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1426# define _generic_isCC_A(c, classnum) \
1427 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1428#endif /* End of no perl.h H_PERL */
8a58bdcf 1429
e66b99e9
KW
1430#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1431#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1432#define isASCII_A(c) isASCII(c)
1433#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1434#define isBLANK(c) isBLANK_A(c)
1435#define isCNTRL(c) isCNTRL_A(c)
1436#define isDIGIT(c) isDIGIT_A(c)
1437#define isGRAPH(c) isGRAPH_A(c)
1438#define isIDFIRST(c) isIDFIRST_A(c)
1439#define isLOWER(c) isLOWER_A(c)
1440#define isPRINT(c) isPRINT_A(c)
779cf272 1441#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1442#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1443#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1444#define isPUNCT(c) isPUNCT_A(c)
1445#define isSPACE(c) isSPACE_A(c)
1446#define isUPPER(c) isUPPER_A(c)
1447#define isWORDCHAR(c) isWORDCHAR_A(c)
1448#define isXDIGIT(c) isXDIGIT_A(c)
1449
1450/* ASCII casing. These could also be written as
1451 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1452 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1453 which uses table lookup and mask instead of subtraction. (This would
1454 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1455#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1456#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1457
25200305
KW
1458/* In the ASCII range, these are equivalent to what they're here defined to be.
1459 * But by creating these definitions, other code doesn't have to be aware of
1460 * this detail */
1461#define toFOLD(c) toLOWER(c)
25200305
KW
1462#define toTITLE(c) toUPPER(c)
1463
c753c8d3
KW
1464#define toLOWER_A(c) toLOWER(c)
1465#define toUPPER_A(c) toUPPER(c)
25200305
KW
1466#define toFOLD_A(c) toFOLD(c)
1467#define toTITLE_A(c) toTITLE(c)
1a0901db 1468
4650c663 1469/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1470#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1471 ? (c) \
f4cd282c 1472 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1473#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1474
1a0901db 1475/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1476 * all mapped to one of them, and these need special handling; returns the
1477 * input itself if is out-of-range */
b2bf251f 1478#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1479 ? (c) \
f4cd282c 1480 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1481#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1482
beab9ebe
KW
1483/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1484
1485/* For internal core Perl use only: the base macro for defining macros like
1486 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1487 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1488 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1489 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1490 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1491 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1492 * code points under locale rules, use the macros based on _generic_LC_uvchr
1493 * instead of this. */
beab9ebe
KW
1494#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1495 (! FITS_IN_8_BITS(c) \
1496 ? 0 \
31f05a37
KW
1497 : IN_UTF8_CTYPE_LOCALE \
1498 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1499 : cBOOL(non_utf8))
1500
1501/* For internal core Perl use only: a helper macro for defining macros like
1502 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1503 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1504 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1505 * _generic_LC_base for more info */
1506#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1507 _generic_LC_base(c,utf8_locale_classnum, \
1508 non_utf8_func( (_LC_CAST) (c)))
1509
1510/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1511 * 'c' is the platform's native underscore character */
1512#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1513 _generic_LC_base(c, utf8_locale_classnum, \
1514 (non_utf8_func( (_LC_CAST) (c)) \
1515 || (char)(c) == '_'))
1516
1517/* These next three are also for internal core Perl use only: case-change
1518 * helper macros */
1519#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1520 ? (c) \
31f05a37
KW
1521 : (IN_UTF8_CTYPE_LOCALE) \
1522 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1523 : (cast)function((cast)(c)))
beab9ebe 1524
31f05a37
KW
1525/* Note that the result can be larger than a byte in a UTF-8 locale. It
1526 * returns a single value, so can't adequately return the upper case of LATIN
1527 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1528 * values "SS"); instead it asserts against that under DEBUGGING, and
1529 * otherwise returns its input */
beab9ebe
KW
1530#define _generic_toUPPER_LC(c, function, cast) \
1531 (! FITS_IN_8_BITS(c) \
1532 ? (c) \
31f05a37 1533 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1534 ? (cast)function((cast)(c)) \
31f05a37
KW
1535 : ((((U8)(c)) == MICRO_SIGN) \
1536 ? GREEK_CAPITAL_LETTER_MU \
1537 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1538 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1539 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1540 ? (__ASSERT_(0) (c)) \
1541 : PL_mod_latin1_uc[ (U8) (c) ])))))
1542
1543/* Note that the result can be larger than a byte in a UTF-8 locale. It
1544 * returns a single value, so can't adequately return the fold case of LATIN
1545 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1546 * values "ss"); instead it asserts against that under DEBUGGING, and
1547 * otherwise returns its input */
beab9ebe 1548#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1549 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1550 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1551 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1552 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1553 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1554
84061b6a 1555/* Use the libc versions for these if available. */
f05550c0 1556#if defined(HAS_ISASCII)
84061b6a
KW
1557# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1558#else
1559# define isASCII_LC(c) isASCII(c)
1560#endif
1561
f05550c0 1562#if defined(HAS_ISBLANK)
84061b6a 1563# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1564#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1565# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1566#endif
1567
f05550c0 1568#define _LC_CAST U8
bbce6d69 1569
f05550c0 1570#ifdef WIN32
375f5f06
KW
1571 /* The Windows functions don't bother to follow the POSIX standard, which
1572 * for example says that something can't both be a printable and a control.
1573 * But Windows treats the \t control as a printable, and does such things
1574 * as making superscripts into both digits and punctuation. This tames
1575 * these flaws by assuming that the definitions of both controls and space
1576 * are correct, and then making sure that other definitions don't have
1577 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1578 * Not all possible weirdnesses are checked for, just the ones that were
1579 * detected on actual Microsoft code pages */
1580
b7d90381
KW
1581# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1582# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1583
1584# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1585 && isALPHANUMERIC_LC(c))
1586# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1587 ! isPUNCT_LC(c))
1588# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1589 isALPHANUMERIC_LC(c))
1590# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1591# define isIDFIRST_LC(c) (((c) == '_') \
1592 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1593# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1594# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1595# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1596# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1597# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1598# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1599 && isALPHANUMERIC_LC(c))
f05550c0
BF
1600
1601# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1602# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1603# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1604
1605#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1606 /* For most other platforms */
beab9ebe 1607
f05550c0
BF
1608# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1609# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1610# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1611# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1612# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1613# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1614# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1615# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1616# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1617# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1618# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1619# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1620# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1621
1622
1623# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1624# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1625# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1626
1627#else /* The final fallback position */
1628
b7d90381
KW
1629# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1630# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1631# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1632# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1633# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1634# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1635# define isLOWER_LC(c) (isascii(c) && islower(c))
1636# define isPRINT_LC(c) (isascii(c) && isprint(c))
1637# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1638# define isSPACE_LC(c) (isascii(c) && isspace(c))
1639# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1640# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1641# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1642
1643# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1644# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1645# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1646
f05550c0 1647#endif
55204971 1648
eba68aa0
KW
1649#define isIDCONT(c) isWORDCHAR(c)
1650#define isIDCONT_A(c) isWORDCHAR_A(c)
1651#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1652#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1653#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1654
4650c663 1655/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1656 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1657 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1658 * classes where there is a macro or function 'above_latin1' that takes the
1659 * single argument 'c' and returns the desired value. These exist for those
1660 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1661 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1662 * for classes where that overhead is faster than a direct lookup.
d0da05db 1663 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1664 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1665 * have duplicate checks here, so could create versions of the macros that
1666 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1667
1668/* Note that all ignore 'use bytes' */
1e222e4f
KW
1669#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1670 ? _generic_isCC(c, classnum) \
cd500f2f 1671 : above_latin1(c))
1e222e4f
KW
1672#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
1673 ? _generic_isCC(c, classnum) \
922e8cb4 1674 : _is_uni_FOO(classnum, c))
d0da05db
KW
1675#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1676#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1677#define isASCII_uvchr(c) isASCII(c)
1678#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1679#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1680#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1681#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1682#define isIDCONT_uvchr(c) \
1683 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1684#define isIDFIRST_uvchr(c) \
1685 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
d0da05db
KW
1686#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1687#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1688
1689#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1690#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1691#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1692
1693#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1694#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1695#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1696#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1697
1698#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1699#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1700#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1701#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1702
1703/* For backwards compatibility, even though '_uni' should mean official Unicode
1704 * code points, in Perl it means native for those below 256 */
1705#define isALPHA_uni(c) isALPHA_uvchr(c)
1706#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1707#define isASCII_uni(c) isASCII_uvchr(c)
1708#define isBLANK_uni(c) isBLANK_uvchr(c)
1709#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1710#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1711#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1712#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1713#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1714#define isLOWER_uni(c) isLOWER_uvchr(c)
1715#define isPRINT_uni(c) isPRINT_uvchr(c)
1716#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1717#define isSPACE_uni(c) isSPACE_uvchr(c)
1718#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1719#define isUPPER_uni(c) isUPPER_uvchr(c)
1720#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1721#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1722#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1723#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1724#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1725#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1726#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1727
4650c663
KW
1728/* For internal core Perl use only: the base macros for defining macros like
1729 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1730 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1731 * simple class definitions; the other for more complex. These are like
d0da05db 1732 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1733#define _generic_LC_uvchr(latin1, above_latin1, c) \
1734 (c < 256 ? latin1(c) : above_latin1(c))
1735#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1736 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1737
1738#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1739#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1740 _CC_ALPHANUMERIC, c)
b7d90381
KW
1741#define isASCII_LC_uvchr(c) isASCII_LC(c)
1742#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1743 is_HORIZWS_cp_high, c)
feeab5a9 1744#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1745#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1746#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1747#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1748 _is_uni_perl_idcont, c)
b7d90381 1749#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1750 _is_uni_perl_idstart, c)
1751#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1752#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1753#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1754#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1755#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1756 is_XPERLSPACE_cp_high, c)
cd500f2f 1757#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1758#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1759 _CC_WORDCHAR, c)
b7d90381
KW
1760#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1761 is_XDIGIT_cp_high, c)
e712593e 1762
b7d90381 1763#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1764
4650c663
KW
1765/* For internal core Perl use only: the base macros for defining macros like
1766 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1767 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1768 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1769 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1770 * same representation whether utf8 or not. Note that it assumes that the utf8
1771 * has been validated, and ignores 'use bytes' */
34aeb2e9
KW
1772#define _base_generic_utf8(enum_name, name, p, use_locale ) \
1773 _is_utf8_FOO(CAT2(_CC_, enum_name), \
1774 (const U8 *) p, \
1775 "is" STRINGIFY(name) "_utf8", \
1776 "is" STRINGIFY(name) "_utf8_safe", \
1777 1, use_locale, __FILE__,__LINE__)
1778
1779#define _generic_utf8(name, p) _base_generic_utf8(name, name, p, 0)
da8c1a98
KW
1780
1781/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
1782 * they don't otherwise go out of their way to look for malformed UTF-8. If
1783 * they can return accurate results without knowing if the input is otherwise
1784 * malformed, they do so. For example isASCII is accurate in spite of any
1785 * non-length malformations because it looks only at a single byte. Likewise
1786 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
1787 * variant ones return FALSE. But, if the input has to be well-formed in order
1788 * for the results to be accurate, the macros will test and if malformed will
1789 * call a routine to die
1790 *
1791 * Except for toke.c, the macros do assume that e > p, asserting that on
1792 * DEBUGGING builds. Much code that calls these depends on this being true,
1793 * for other reasons. toke.c is treated specially as using the regular
1794 * assertion breaks it in many ways. All strings that these operate on there
1795 * are supposed to have an extra NUL character at the end, so that *e = \0. A
1796 * bunch of code in toke.c assumes that this is true, so the assertion allows
1797 * for that */
1798#ifdef PERL_IN_TOKE_C
1799# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
1800#else
1801# define _utf8_safe_assert(p,e) ((e) > (p))
1802#endif
1803
1804#define _generic_utf8_safe(classnum, p, e, above_latin1) \
1805 (__ASSERT_(_utf8_safe_assert(p, e)) \
1806 (UTF8_IS_INVARIANT(*(p))) \
1807 ? _generic_isCC(*(p), classnum) \
1808 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1809 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1810 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
1811 classnum) \
1812 : (_force_out_malformed_utf8_message( \
1813 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1814 : above_latin1))
b7d90381
KW
1815/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1816 * 'above_latin1' can be a macro */
da8c1a98
KW
1817#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
1818 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
1819#define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1820 _generic_utf8_safe(classnum, p, e, \
1821 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1822 ? (_force_out_malformed_utf8_message( \
1823 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1824 : above_latin1(p)))
4650c663 1825/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1826 * 'above_latin1' parameter */
da8c1a98
KW
1827#define _generic_swash_utf8_safe(classnum, p, e) \
1828_generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e))
922e8cb4 1829
cc8ab7c0 1830/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1831 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1832 * class is TRUE for. Hence it can skip the tests for this range.
1833 * 'above_latin1' should include its arguments */
da8c1a98
KW
1834#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
1835 (__ASSERT_(_utf8_safe_assert(p, e)) \
1836 (UTF8_IS_INVARIANT(*(p))) \
1837 ? _generic_isCC(*(p), classnum) \
1838 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1839 ? 0 /* Note that doesn't check validity for latin1 */ \
1840 : above_latin1)
1841
84238efa
KW
1842/* NOTE that some of these macros have very similar ones in regcharclass.h.
1843 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1844 * there, differing in name only by an underscore from the one here
1845 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1846 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1847 * points; the regcharclass.h ones are implemented as a series of
1848 * "if-else-if-else ..." */
1849
34aeb2e9
KW
1850#define isALPHA_utf8(p) _generic_utf8(ALPHA, p)
1851#define isALPHANUMERIC_utf8(p) _generic_utf8(ALPHANUMERIC, p)
1852#define isASCII_utf8(p) _generic_utf8(ASCII, p)
1853#define isBLANK_utf8(p) _generic_utf8(BLANK, p)
1854#define isCNTRL_utf8(p) _generic_utf8(CNTRL, p)
1855#define isDIGIT_utf8(p) _generic_utf8(DIGIT, p)
1856#define isGRAPH_utf8(p) _generic_utf8(GRAPH, p)
1857#define isIDCONT_utf8(p) _generic_utf8(IDCONT, p)
1858#define isIDFIRST_utf8(p) _generic_utf8(IDFIRST, p)
1859#define isLOWER_utf8(p) _generic_utf8(LOWER, p)
1860#define isPRINT_utf8(p) _generic_utf8(PRINT, p)
1861#define isPSXSPC_utf8(p) _generic_utf8(PSXSPC, p)
1862#define isPUNCT_utf8(p) _generic_utf8(PUNCT, p)
1863#define isSPACE_utf8(p) _generic_utf8(SPACE, p)
1864#define isUPPER_utf8(p) _generic_utf8(UPPER, p)
1865#define isVERTWS_utf8(p) _generic_utf8(VERTSPACE, p)
1866#define isWORDCHAR_utf8(p) _generic_utf8(WORDCHAR, p)
1867#define isXDIGIT_utf8(p) _generic_utf8(XDIGIT, p)
e8fa43e2 1868
da8c1a98
KW
1869#define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e)
1870#define isALPHANUMERIC_utf8_safe(p, e) \
1871 _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e)
1872#define isASCII_utf8_safe(p, e) \
1873 /* Because ASCII is invariant under utf8, the non-utf8 macro \
1874 * works */ \
1875 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
1876#define isBLANK_utf8_safe(p, e) \
1877 _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
1878
e8fa43e2
KW
1879#ifdef EBCDIC
1880 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1881 * more efficient macro instead of the more general one */
da8c1a98
KW
1882# define isCNTRL_utf8_safe(p, e) \
1883 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p))
e8fa43e2 1884#else
da8c1a98 1885# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
1886#endif
1887
da8c1a98
KW
1888#define isDIGIT_utf8_safe(p, e) \
1889 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
1890 _is_utf8_FOO_with_len(_CC_DIGIT, p, e))
1891#define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e)
1892#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
1893 _is_utf8_perl_idcont_with_len, p, e)
e5dcd934 1894
c11ff943
KW
1895/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1896 * IDFIRST is an alnum. See
c8362b00 1897 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1898 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1899 * This used to be not the XID version, but we decided to go with the more
1900 * modern Unicode definition */
da8c1a98
KW
1901#define isIDFIRST_utf8_safe(p, e) \
1902 _generic_func_utf8_safe(_CC_IDFIRST, \
1903 _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e))
1904
1905#define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e)
1906#define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e)
1907#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
1908#define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e)
1909#define isSPACE_utf8_safe(p, e) \
1910 _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
1911#define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e)
1912#define isVERTWS_utf8_safe(p, e) \
1913 _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
1914#define isWORDCHAR_utf8_safe(p, e) \
1915 _generic_swash_utf8_safe(_CC_WORDCHAR, p, e)
1916#define isXDIGIT_utf8_safe(p, e) \
1917 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
1918 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1919 ? (_force_out_malformed_utf8_message( \
1920 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1921 : is_XDIGIT_high(p)))
a0ed51b3 1922
7d2b61dd 1923#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1924#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1925#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1926#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1927
567b353c 1928/* For internal core use only, subject to change */
a239b1e2
KW
1929#define _toFOLD_utf8_flags(p,s,l,f) _to_utf8_fold_flags (p,NULL,s,l,f)
1930#define _toLOWER_utf8_flags(p,s,l,f) _to_utf8_lower_flags(p,NULL,s,l,f)
1931#define _toTITLE_utf8_flags(p,s,l,f) _to_utf8_title_flags(p,NULL,s,l,f)
1932#define _toUPPER_utf8_flags(p,s,l,f) _to_utf8_upper_flags(p,NULL,s,l,f)
1933
1934#define toFOLD_utf8_safe(p,e,s,l) _to_utf8_fold_flags(p,e,s,l, FOLD_FLAGS_FULL)
1935#define toLOWER_utf8_safe(p,e,s,l) _to_utf8_lower_flags(p,e,s,l, 0)
1936#define toTITLE_utf8_safe(p,e,s,l) _to_utf8_title_flags(p,e,s,l, 0)
1937#define toUPPER_utf8_safe(p,e,s,l) _to_utf8_upper_flags(p,e,s,l, 0)
567b353c 1938
4650c663
KW
1939/* For internal core Perl use only: the base macros for defining macros like
1940 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1941 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1942 * 'macro' parameter */
34aeb2e9
KW
1943#define _generic_LC_utf8(name, p) _base_generic_utf8(name, name, p, 1)
1944
1945#define isALPHA_LC_utf8(p) _generic_LC_utf8(ALPHA, p)
1946#define isALPHANUMERIC_LC_utf8(p) _generic_LC_utf8(ALPHANUMERIC, p)
1947#define isASCII_LC_utf8(p) _generic_LC_utf8(ASCII, p)
1948#define isBLANK_LC_utf8(p) _generic_LC_utf8(BLANK, p)
1949#define isCNTRL_LC_utf8(p) _generic_LC_utf8(CNTRL, p)
1950#define isDIGIT_LC_utf8(p) _generic_LC_utf8(DIGIT, p)
1951#define isGRAPH_LC_utf8(p) _generic_LC_utf8(GRAPH, p)
1952#define isIDCONT_LC_utf8(p) _generic_LC_utf8(IDCONT, p)
1953#define isIDFIRST_LC_utf8(p) _generic_LC_utf8(IDFIRST, p)
1954#define isLOWER_LC_utf8(p) _generic_LC_utf8(LOWER, p)
1955#define isPRINT_LC_utf8(p) _generic_LC_utf8(PRINT, p)
1956#define isPSXSPC_LC_utf8(p) _generic_LC_utf8(PSXSPC, p)
1957#define isPUNCT_LC_utf8(p) _generic_LC_utf8(PUNCT, p)
1958#define isSPACE_LC_utf8(p) _generic_LC_utf8(SPACE, p)
1959#define isUPPER_LC_utf8(p) _generic_LC_utf8(UPPER, p)
1960#define isWORDCHAR_LC_utf8(p) _generic_LC_utf8(WORDCHAR, p)
1961#define isXDIGIT_LC_utf8(p) _generic_LC_utf8(XDIGIT, p)
1962
da8c1a98
KW
1963/* For internal core Perl use only: the base macros for defining macros like
1964 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
1965 * point in 'p' is within the 0-255 range, it uses locale rules from the
1966 * passed-in 'macro' parameter */
1967#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
1968 (__ASSERT_(_utf8_safe_assert(p, e)) \
1969 (UTF8_IS_INVARIANT(*(p))) \
1970 ? macro(*(p)) \
1971 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1972 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1973 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
1974 : (_force_out_malformed_utf8_message( \
1975 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1976 : above_latin1))
1977
1978#define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \
1979 _generic_LC_utf8_safe(macro, p, e, \
1980 _is_utf8_FOO_with_len(classnum, p, e))
1981
1982#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
1983 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
1984
1985#define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1986 _generic_LC_utf8_safe(classnum, p, e, \
1987 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1988 ? (_force_out_malformed_utf8_message( \
1989 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1990 : above_latin1(p)))
1991
1992#define isALPHANUMERIC_LC_utf8_safe(p, e) \
1993 _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \
1994 _CC_ALPHANUMERIC, p, e)
1995#define isALPHA_LC_utf8_safe(p, e) \
1996 _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
1997#define isASCII_LC_utf8_safe(p, e) \
1998 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
1999#define isBLANK_LC_utf8_safe(p, e) \
2000 _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2001#define isCNTRL_LC_utf8_safe(p, e) \
2002 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2003#define isDIGIT_LC_utf8_safe(p, e) \
2004 _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2005#define isGRAPH_LC_utf8_safe(p, e) \
2006 _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2007#define isIDCONT_LC_utf8_safe(p, e) \
2008 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2009 _is_utf8_perl_idcont_with_len, p, e)
2010#define isIDFIRST_LC_utf8_safe(p, e) \
2011 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2012 _is_utf8_perl_idstart_with_len, p, e)
2013#define isLOWER_LC_utf8_safe(p, e) \
2014 _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2015#define isPRINT_LC_utf8_safe(p, e) \
2016 _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2017#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2018#define isPUNCT_LC_utf8_safe(p, e) \
2019 _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2020#define isSPACE_LC_utf8_safe(p, e) \
2021 _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2022#define isUPPER_LC_utf8_safe(p, e) \
2023 _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2024#define isWORDCHAR_LC_utf8_safe(p, e) \
2025 _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2026#define isXDIGIT_LC_utf8_safe(p, e) \
2027 _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2028
fbc19f27
KW
2029/* Macros for backwards compatibility and for completeness when the ASCII and
2030 * Latin1 values are identical */
b7d90381
KW
2031#define isALPHAU(c) isALPHA_L1(c)
2032#define isDIGIT_L1(c) isDIGIT_A(c)
2033#define isOCTAL(c) isOCTAL_A(c)
2034#define isOCTAL_L1(c) isOCTAL_A(c)
2035#define isXDIGIT_L1(c) isXDIGIT_A(c)
2036#define isALNUM(c) isWORDCHAR(c)
2037#define isALNUMU(c) isWORDCHAR_L1(c)
2038#define isALNUM_LC(c) isWORDCHAR_LC(c)
2039#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2040#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
2041#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
2042#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
2043#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2044#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2045#define isALNUMC(c) isALPHANUMERIC(c)
2046#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2047#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2048#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 2049#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 2050#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 2051
2bd1cbf6
KW
2052/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2053 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2054 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2055 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2056 * the outlier from the block that contains the other controls, just like
2057 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2058 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2059 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2060 * etc. */
2bd1cbf6 2061#ifndef EBCDIC
75763b3a 2062# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2063#else
75763b3a
KW
2064# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2065 ((isPRINT_A(c)) \
2066 ? (UNLIKELY((c) == '?') \
2067 ? QUESTION_MARK_CTRL \
2068 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2069 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2070 ? '?' \
2071 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2072#endif
bbce6d69 2073
dea28490
JJ
2074/* Line numbers are unsigned, 32 bits. */
2075typedef U32 line_t;
e5dcd934 2076#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2077
91152fc1
DG
2078/* Helpful alias for version prescan */
2079#define is_LAX_VERSION(a,b) \
2080 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2081
2082#define is_STRICT_VERSION(a,b) \
2083 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2084
2085#define BADVERSION(a,b,c) \
2086 if (b) { \
2087 *b = c; \
2088 } \
2089 return a;
8c52afec 2090
cb27eebd
KW
2091/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2092 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2093 * which is advanced. The input is validated only by an assert() in DEBUGGING
2094 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2095 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2096#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2097 ? (c) \
2098 : ((c) + 9))))
902008b8
KW
2099#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2100 ? (*(s)++) \
2101 : (*(s)++ + 9))))
95a59cab 2102
cb27eebd
KW
2103/* Converts a character known to represent an octal digit (0-7) to its numeric
2104 * value. The input is validated only by an assert() in DEBUGGING builds. In
2105 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2106#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2107
305b8651
KW
2108/* Efficiently returns a boolean as to if two native characters are equivalent
2109 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2110 * the ALPHA in the name is to remind you of that. This is asserted() in
2111 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2112 * works (on valid input) for both non- and UTF-8-encoded bytes.
2113 *
2114 * When one of the inputs is a compile-time constant and gets folded by the
2115 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2116 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2117 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2118 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2119 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2120 * just a single 0, in the bit position where the upper- and lowercase differ.
2121 * */
305b8651
KW
2122#define isALPHA_FOLD_EQ(c1, c2) \
2123 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2124 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2125#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2126
8e84507e 2127/*
ccfc67b7
JH
2128=head1 Memory Management
2129
a02a5408 2130=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2131The XSUB-writer's interface to the C C<malloc> function.
2132
596f7718 2133Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2134
c5008215
JC
2135In 5.9.3, Newx() and friends replace the older New() API, and drops
2136the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2137themselves. This aid has been superseded by a new build option,
d10b4965 2138PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2139there for use in XS modules supporting older perls.
2140
a02a5408 2141=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2142The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2143cast. See also C<L</Newx>>.
954c1994 2144
596f7718 2145Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2146
a02a5408 2147=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2148The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2149memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2150
596f7718 2151Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2152
954c1994
GS
2153=for apidoc Am|void|Renew|void* ptr|int nitems|type
2154The XSUB-writer's interface to the C C<realloc> function.
2155
596f7718 2156Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2157
954c1994
GS
2158=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2159The XSUB-writer's interface to the C C<realloc> function, with
2160cast.
2161
596f7718 2162Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2163
49b8b560 2164=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2165The XSUB-writer's interface to the C C<free> function.
2166
596f7718 2167This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2168
954c1994
GS
2169=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2170The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2171source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2172C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2173
e90e2364 2174=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2175Like C<Move> but returns C<dest>. Useful
72d33970 2176for encouraging compilers to tail-call
e90e2364
NC
2177optimise.
2178
954c1994
GS
2179=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2180The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2181source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2182C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2183
e90e2364
NC
2184=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2185
796b6530 2186Like C<Copy> but returns C<dest>. Useful
72d33970 2187for encouraging compilers to tail-call
e90e2364
NC
2188optimise.
2189
954c1994
GS
2190=for apidoc Am|void|Zero|void* dest|int nitems|type
2191
2192The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2193destination, C<nitems> is the number of items, and C<type> is the type.
2194
e90e2364
NC
2195=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2196
72d33970
FC
2197Like C<Zero> but returns dest. Useful
2198for encouraging compilers to tail-call
e90e2364
NC
2199optimise.
2200
da5d8dbb 2201=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2202This is an architecture-independent macro to copy one structure to another.
954c1994 2203
7e337ee0
JH
2204=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2205
2206Fill up memory with a byte pattern (a byte repeated over and over
2207again) that hopefully catches attempts to access uninitialized memory.
2208
2209=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2210
2211PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2212
1c12ffb4 2213=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2214
2215PoisonWith(0xEF) for catching access to freed memory.
2216
9965345d
JH
2217=for apidoc Am|void|Poison|void* dest|int nitems|type
2218
7e337ee0 2219PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2220
2221=cut */
954c1994 2222
561b68a9
SH
2223/* Maintained for backwards-compatibility only. Use newSV() instead. */
2224#ifndef PERL_CORE
ff06c60c 2225#define NEWSV(x,len) newSV(len)
561b68a9 2226#endif
ff06c60c 2227
19a94d75
JH
2228#define MEM_SIZE_MAX ((MEM_SIZE)~0)
2229
e6bdf523 2230
27d5b266 2231#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2232
2233/* This expression will be constant-folded at compile time. It checks
2234 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2235 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2236 * As well as avoiding the need for a run-time check in some cases, it's
2237 * designed to avoid compiler warnings like:
2238 * comparison is always false due to limited range of data type
73e8ff00
DM
2239 * It's mathematically equivalent to
2240 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2241 */
2242
2243# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
004073ba 2244 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
e6bdf523 2245
88f9f128 2246/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2247 * compiler warnings. We *want* to write the expression as
2248 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2249 * (for some compile-time constant C), but even when the LHS
2250 * constant-folds to false at compile-time, g++ insists on emitting
2251 * warnings about the RHS (e.g. "comparison is always false"), so instead
2252 * we write it as
e6bdf523 2253 *
d98e5cde 2254 * (cond ? n : X) > C
88f9f128 2255 *
d98e5cde
DM
2256 * where X is a constant with X > C always false. Choosing a value for X
2257 * is tricky. If 0, some compilers will complain about 0 > C always being
2258 * false; if 1, Coverity complains when n happens to be the constant value
2259 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2260 * for X and hope that nothing else whines.
e6bdf523
DM
2261 */
2262
2263# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2264 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2265 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2266
2267# define MEM_WRAP_CHECK(n,t) \
2268 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2269 && (croak_memory_wrap(),0))
2270
2271# define MEM_WRAP_CHECK_1(n,t,a) \
2272 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2273 && (Perl_croak_nocontext("%s",(a)),0))
2274
8b44ba4c 2275#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2276
d1decf2b 2277#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
2278#else
2279
410319be
NC
2280#define MEM_WRAP_CHECK(n,t)
2281#define MEM_WRAP_CHECK_1(n,t,a)
2282#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
2283#define MEM_WRAP_CHECK_(n,t)
2284
6fff79ce 2285#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 2286
1936d2a7 2287#endif
8b44ba4c 2288
fe4f188c 2289#ifdef PERL_MEM_LOG
46c6c7e2 2290/*
9f653bb5 2291 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2292 * go through functions, which are handy for debugging breakpoints, but
2293 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2294 * line number, and C function name if available) passed in. This info can
2295 * then be used for logging the calls, for which one gets a sample
73d1d973 2296 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2297 *
46c6c7e2 2298 * Known problems:
94e892a6 2299 * - not all memory allocs get logged, only those
46c6c7e2 2300 * that go through Newx() and derivatives (while all
94e892a6 2301 * Safefrees do get logged)
46c6c7e2
JH
2302 * - __FILE__ and __LINE__ do not work everywhere
2303 * - __func__ or __FUNCTION__ even less so
2304 * - I think more goes on after the perlio frees but
2305 * the thing is that STDERR gets closed (as do all
2306 * the file descriptors)
2307 * - no deeper calling stack than the caller of the Newx()
2308 * or the kind, but do I look like a C reflection/introspection
2309 * utility to you?
2310 * - the function prototypes for the logging functions
2311 * probably should maybe be somewhere else than handy.h
2312 * - one could consider inlining (macrofying) the logging
2313 * for speed, but I am too lazy
2314 * - one could imagine recording the allocations in a hash,
2315 * (keyed by the allocation address?), and maintain that
2316 * through reallocs and frees, but how to do that without
2317 * any News() happening...?
73d1d973 2318 * - lots of -Ddefines to get useful/controllable output
b953482e 2319 * - lots of ENV reads
46c6c7e2
JH
2320 */
2321
0b0ab801 2322# ifdef PERL_CORE
73d1d973 2323# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2324enum mem_log_type {
2325 MLT_ALLOC,
2326 MLT_REALLOC,
d7a2c63c
MHM
2327 MLT_FREE,
2328 MLT_NEW_SV,
2329 MLT_DEL_SV
0b0ab801
MHM
2330};
2331# endif
12754f92 2332# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2333void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2334void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2335# endif
0b0ab801
MHM
2336# endif
2337
fe4f188c
JH
2338#endif
2339
2340#ifdef PERL_MEM_LOG
d1401ee9
MHM
2341#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2342#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2343#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2344#endif
2345
2346#ifndef MEM_LOG_ALLOC
2347#define MEM_LOG_ALLOC(n,t,a) (a)
2348#endif
2349#ifndef MEM_LOG_REALLOC
2350#define MEM_LOG_REALLOC(n,t,v,a) (a)
2351#endif
2352#ifndef MEM_LOG_FREE
2353#define MEM_LOG_FREE(a) (a)
2354#endif
2355
d1401ee9
MHM
2356#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2357#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2358#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2359
2360#ifndef PERL_CORE
a02a5408
JC
2361/* pre 5.9.x compatibility */
2362#define New(x,v,n,t) Newx(v,n,t)
2363#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2364#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2365#endif
a02a5408 2366
ff68c719 2367#define Renew(v,n,t) \
d1401ee9 2368 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2369#define Renewc(v,n,t,c) \
d1401ee9 2370 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2371
2372#ifdef PERL_POISON
2373#define Safefree(d) \
06c0cc96 2374 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2375#else
fe4f188c 2376#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2377#endif
55497cff 2378
8b44ba4c
NC
2379#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2380#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2381#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2382
8b44ba4c
NC
2383#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2384#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2385#ifdef HAS_MEMSET
8b44ba4c 2386#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2387#else
8b44ba4c
NC
2388/* Using bzero(), which returns void. */
2389#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2390#endif
2391
7e337ee0
JH
2392#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2393#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2394#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2395#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2396
caa674f3
DD
2397#ifdef PERL_POISON
2398# define PERL_POISON_EXPR(x) x
2399#else
2400# define PERL_POISON_EXPR(x)
2401#endif
2402
2304df62 2403#ifdef USE_STRUCT_COPY
ff68c719 2404#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2405#else
2406#define StructCopy(s,d,t) Copy(s,d,1,t)
2407#endif
2cc61e15 2408
c3caa5c3
JH
2409/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2410 * want your zero-based indices to be less than but not equal to).
2411 *
2412 * C_ARRAY_END is one past the last: half-open/half-closed range,
2413 * not last-inclusive range. */
622913ab 2414#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2415#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2416
2cc61e15
DD
2417#ifdef NEED_VA_COPY
2418# ifdef va_copy
2419# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2420# else
a1866d1b
JH
2421# if defined(__va_copy)
2422# define Perl_va_copy(s, d) __va_copy(d, s)
2423# else
2424# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2425# endif
2cc61e15
DD
2426# endif
2427#endif
2428
472d47bc
SB
2429/* convenience debug macros */
2430#ifdef USE_ITHREADS
2431#define pTHX_FORMAT "Perl interpreter: 0x%p"
2432#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2433#define pTHX_VALUE_ (void *)my_perl,
2434#define pTHX_VALUE (void *)my_perl
2435#define pTHX__VALUE_ ,(void *)my_perl,
2436#define pTHX__VALUE ,(void *)my_perl
472d47bc 2437#else
3609ea0d 2438#define pTHX_FORMAT
472d47bc 2439#define pTHX__FORMAT
3609ea0d 2440#define pTHX_VALUE_
472d47bc 2441#define pTHX_VALUE
3609ea0d 2442#define pTHX__VALUE_
472d47bc
SB
2443#define pTHX__VALUE
2444#endif /* USE_ITHREADS */
3609ea0d 2445
2acdbac1
NC
2446/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2447 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2448 CPAN::Unpack show any users outside the core. */
2449#ifdef PERL_CORE
b7d90381
KW
2450# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2451 "Use of " s " is deprecated")
2acdbac1
NC
2452#endif
2453
dfff4baf
BF
2454/* Internal macros to deal with gids and uids */
2455#ifdef PERL_CORE
2456
2457# if Uid_t_size > IVSIZE
2458# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2459# define SvUID(sv) SvNV(sv)
2460# else
2461# if Uid_t_sign <= 0
2462# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2463# define SvUID(sv) SvIV(sv)
2464# else
2465# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2466# define SvUID(sv) SvUV(sv)
2467# endif
2468# endif /* Uid_t_size */
2469
2470# if Gid_t_size > IVSIZE
2471# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2472# define SvGID(sv) SvNV(sv)
2473# else
2474# if Gid_t_sign <= 0
2475# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2476# define SvGID(sv) SvIV(sv)
2477# else
2478# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2479# define SvGID(sv) SvUV(sv)
2480# endif
2481# endif /* Gid_t_size */
2482
2483#endif
2484
9d745869
KW
2485#endif /* HANDY_H */
2486
e9a8c099 2487/*
14d04a33 2488 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2489 */