This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regexec.c: Make isFOO_lc() non-static
[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
6aff1f14 551Variant C<isFOO_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
KW
554
555Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
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
d0da05db 562Variant C<isFOO_uvchr> is like the C<isFOO_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
da8c1a98
KW
568Variant C<isFOO_utf8_safe> is like C<isFOO_uvchr>, but is used for UTF-8
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
582Variant C<isFOO_utf8> is like C<isFOO_utf8_safe>, but takes just a single
583parameter, C<p>, which has the same meaning as the corresponding parameter does
584in C<isFOO_utf8_safe>. The function therefore can't check if it is reading
585beyond the end of the string.
243effed 586
31f05a37
KW
587Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
588result is based on the current locale, which is what C<LC> in the name stands
589for. If Perl can determine that the current locale is a UTF-8 locale, it uses
590the published Unicode rules; otherwise, it uses the C library function that
591gives the named classification. For example, C<isDIGIT_LC()> when not in a
592UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
593returned if the input won't fit into an octet. On some platforms where the C
594library function is known to be defective, Perl changes its result to follow
595the POSIX standard's rules.
243effed
KW
596
597Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
598returns the same as C<isFOO_LC> for input code points less than 256, and
599returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
600
da8c1a98
KW
601Variant C<isFOO_LC_utf8_safe> is like C<isFOO_LC_uvchr>, but is used for UTF-8
602encoded strings. Each call classifies one character, even if the string
603contains many. This variant takes two parameters. The first, C<p>, is a
604pointer to the first byte of the character to be classified. (Recall that it
605may take more than one byte to represent a character in UTF-8 strings.) The
606second parameter, C<e>, points to anywhere in the string beyond the first
607character, up to one byte past the end of the entire string. The suffix
608C<_safe> in the function's name indicates that it will not attempt to read
609beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
610is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
611character is malformed in some way, the program may croak, or the function may
612return FALSE, at the discretion of the implementation, and subject to change in
613future releases.
614
615Variant C<isFOO_LC_utf8> is like C<isFOO_LC_utf8_safe>, but takes just a single
616parameter, C<p>, which has the same meaning as the corresponding parameter does
617in C<isFOO_LC_utf8_safe>. The function therefore can't check if it is reading
618beyond the end of the string.
ccfc67b7 619
954c1994 620=for apidoc Am|bool|isALPHA|char ch
2787a470 621Returns a boolean indicating whether the specified character is an
6aff1f14 622alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
623See the L<top of this section|/Character classification> for an explanation of
624variants
da8c1a98
KW
625C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8_safe>,
626C<isALPHA_LC>, C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8_safe>.
8a58bdcf 627
15861f94
KW
628=for apidoc Am|bool|isALPHANUMERIC|char ch
629Returns a boolean indicating whether the specified character is a either an
6aff1f14 630alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
631See the L<top of this section|/Character classification> for an explanation of
632variants
d0da05db 633C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
da8c1a98
KW
634C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>,
635and C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 636
8a58bdcf
KW
637=for apidoc Am|bool|isASCII|char ch
638Returns a boolean indicating whether the specified character is one of the 128
243effed 639characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 640On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
641character corresponds to an ASCII character. Variants C<isASCII_A()> and
642C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
643See the L<top of this section|/Character classification> for an explanation of
644variants
da8c1a98
KW
645C<isASCII_uvchr>, C<isASCII_utf8_safe>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
646C<isASCII_LC_utf8_safe>. Note, however, that some platforms do not have the C
243effed
KW
647library routine C<isascii()>. In these cases, the variants whose names contain
648C<LC> are the same as the corresponding ones without.
649
d98532ea
KW
650Also note, that because all ASCII characters are UTF-8 invariant (meaning they
651have the exact same representation (always a single byte) whether encoded in
652UTF-8 or not), C<isASCII> will give the correct results when called with any
da8c1a98
KW
653byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8_safe>
654will work properly on any string encoded or not in UTF-8.
d98532ea 655
243effed
KW
656=for apidoc Am|bool|isBLANK|char ch
657Returns a boolean indicating whether the specified character is a
6aff1f14 658character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
659See the L<top of this section|/Character classification> for an explanation of
660variants
da8c1a98
KW
661C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8_safe>,
662C<isBLANK_LC>, C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8_safe>. Note,
663however, that some platforms do not have the C library routine
664C<isblank()>. In these cases, the variants whose names contain C<LC> are
665the same as the corresponding ones without.
243effed
KW
666
667=for apidoc Am|bool|isCNTRL|char ch
668Returns a boolean indicating whether the specified character is a
6aff1f14 669control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
670See the L<top of this section|/Character classification> for an explanation of
671variants
da8c1a98
KW
672C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8_safe>,
673C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8_safe> On EBCDIC
674platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 675
954c1994 676=for apidoc Am|bool|isDIGIT|char ch
2787a470 677Returns a boolean indicating whether the specified character is a
6aff1f14 678digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 679Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
680See the L<top of this section|/Character classification> for an explanation of
681variants
da8c1a98
KW
682C<isDIGIT_uvchr>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
683C<isDIGIT_LC_utf8_safe>.
243effed
KW
684
685=for apidoc Am|bool|isGRAPH|char ch
686Returns a boolean indicating whether the specified character is a
6aff1f14 687graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 688See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
689variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8_safe>,
690C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8_safe>.
954c1994 691
0c82b6df 692=for apidoc Am|bool|isLOWER|char ch
2787a470 693Returns a boolean indicating whether the specified character is a
6aff1f14 694lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
695See the L<top of this section|/Character classification> for an explanation of
696variants
da8c1a98
KW
697C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8_safe>,
698C<isLOWER_LC>, C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8_safe>.
0c82b6df 699
c99e91e9 700=for apidoc Am|bool|isOCTAL|char ch
2787a470 701Returns a boolean indicating whether the specified character is an
6aff1f14 702octal digit, [0-7].
243effed
KW
703The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
704C<isOCTAL>.
705
706=for apidoc Am|bool|isPUNCT|char ch
707Returns a boolean indicating whether the specified character is a
6aff1f14
KW
708punctuation character, analogous to C<m/[[:punct:]]/>.
709Note that the definition of what is punctuation isn't as
243effed
KW
710straightforward as one might desire. See L<perlrecharclass/POSIX Character
711Classes> for details.
dcccc8ff 712See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
713variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8_safe>,
714C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8_safe>.
c99e91e9 715
0c82b6df 716=for apidoc Am|bool|isSPACE|char ch
2787a470 717Returns a boolean indicating whether the specified character is a
6aff1f14 718whitespace character. This is analogous
398d098a 719to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 720this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
721locale forms of this macro (the ones with C<LC> in their names) matched
722precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
723in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
724(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
725See the L<top of this section|/Character classification> for an explanation of
726variants
da8c1a98
KW
727C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8_safe>,
728C<isSPACE_LC>, C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8_safe>.
0c82b6df 729
398d098a
KW
730=for apidoc Am|bool|isPSXSPC|char ch
731(short for Posix Space)
779cf272
KW
732Starting in 5.18, this is identical in all its forms to the
733corresponding C<isSPACE()> macros.
398d098a
KW
734The locale forms of this macro are identical to their corresponding
735C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
736non-locale forms differ from their C<isSPACE()> forms only in that the
737C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
738Otherwise they are identical. Thus this macro is analogous to what
739C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 740See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
741variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8_safe>,
742C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 743
954c1994 744=for apidoc Am|bool|isUPPER|char ch
2787a470 745Returns a boolean indicating whether the specified character is an
6aff1f14 746uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 747See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
748variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8_safe>,
749C<isUPPER_LC>, C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8_safe>.
954c1994 750
243effed 751=for apidoc Am|bool|isPRINT|char ch
8eea39dd 752Returns a boolean indicating whether the specified character is a
6aff1f14 753printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
754See the L<top of this section|/Character classification> for an explanation of
755variants
da8c1a98
KW
756C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8_safe>,
757C<isPRINT_LC>, C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8_safe>.
243effed
KW
758
759=for apidoc Am|bool|isWORDCHAR|char ch
760Returns a boolean indicating whether the specified character is a character
761that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
762in a regular expression. A word character is an alphabetic character, a
763decimal digit, a connecting punctuation character (such as an underscore), or
764a "mark" character that attaches to one of those (like some sort of accent).
765C<isALNUM()> is a synonym provided for backward compatibility, even though a
766word character includes more than the standard C language meaning of
767alphanumeric.
dcccc8ff 768See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
769variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and
770C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and
771C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
772include the platform's native underscore.
8a58bdcf
KW
773
774=for apidoc Am|bool|isXDIGIT|char ch
775Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
776digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
777and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
778See the L<top of this section|/Character classification> for an explanation of
779variants
da8c1a98
KW
780C<isXDIGIT_uvchr>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>,
781and C<isXDIGIT_LC_utf8_safe>.
243effed 782
3c3ecf18
KW
783=for apidoc Am|bool|isIDFIRST|char ch
784Returns a boolean indicating whether the specified character can be the first
785character of an identifier. This is very close to, but not quite the same as
786the official Unicode property C<XID_Start>. The difference is that this
787returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
788See the L<top of this section|/Character classification> for an explanation of
789variants
da8c1a98
KW
790C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8_safe>,
791C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18
KW
792
793=for apidoc Am|bool|isIDCONT|char ch
794Returns a boolean indicating whether the specified character can be the
795second or succeeding character of an identifier. This is very close to, but
796not quite the same as the official Unicode property C<XID_Continue>. The
797difference is that this returns true only if the input character also matches
dcccc8ff
KW
798L</isWORDCHAR>. See the L<top of this section|/Character classification> for
799an
d0da05db 800explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
da8c1a98
KW
801C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
802C<isIDCONT_LC_utf8_safe>.
3c3ecf18 803
243effed 804=head1 Miscellaneous Functions
8eea39dd 805
95a59cab 806=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 807Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
808Behaviour is only well defined when isXDIGIT(*str) is true.
809
e7c1e6c1 810=head1 Character case changing
21da7284
KW
811Perl uses "full" Unicode case mappings. This means that converting a single
812character to another case may result in a sequence of more than one character.
813For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
814character sequence C<SS>. This presents some complications The lowercase of
815all characters in the range 0..255 is a single character, and thus
816C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
817return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
818an API that does allow every possible legal result to be returned.) Likewise
819no other function that is crippled by not being able to give the correct
820results for the full range of possible inputs has been implemented here.
e7c1e6c1 821
1f607577
KW
822=for apidoc Am|U8|toUPPER|U8 ch
823Converts the specified character to uppercase. If the input is anything but an
824ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 825C<toUPPER_A> is equivalent.
954c1994 826
d0da05db
KW
827=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
828Converts the code point C<cp> to its uppercase version, and
829stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
830point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
831that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
832bytes since the uppercase version may be longer than the original character.
833
834The first code point of the uppercased version is returned
21da7284
KW
835(but note, as explained at L<the top of this section|/Character case
836changing>, that there may be more.)
1f607577
KW
837
838=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
839Converts the UTF-8 encoded character at C<p> to its uppercase version, and
840stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
841that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
842bytes since the uppercase version may be longer than the original character.
843
844The first code point of the uppercased version is returned
21da7284
KW
845(but note, as explained at L<the top of this section|/Character case
846changing>, that there may be more).
1f607577
KW
847
848The input character at C<p> is assumed to be well-formed.
849
25200305
KW
850=for apidoc Am|U8|toFOLD|U8 ch
851Converts the specified character to foldcase. If the input is anything but an
852ASCII uppercase character, that input character itself is returned. Variant
853C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 854Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 855
d0da05db
KW
856=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
857Converts the code point C<cp> to its foldcase version, and
858stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
859point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
860that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
861bytes since the foldcase version may be longer than the original character.
862
863The first code point of the foldcased version is returned
21da7284
KW
864(but note, as explained at L<the top of this section|/Character case
865changing>, that there may be more).
1f607577
KW
866
867=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
868Converts the UTF-8 encoded character at C<p> to its foldcase version, and
869stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
870that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
871bytes since the foldcase version may be longer than the original character.
872
873The first code point of the foldcased version is returned
21da7284
KW
874(but note, as explained at L<the top of this section|/Character case
875changing>, that there may be more).
1f607577
KW
876
877The input character at C<p> is assumed to be well-formed.
878
879=for apidoc Am|U8|toLOWER|U8 ch
880Converts the specified character to lowercase. If the input is anything but an
881ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 882C<toLOWER_A> is equivalent.
954c1994 883
1f607577 884=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
885Converts the specified Latin1 character to lowercase. The results are
886undefined if the input doesn't fit in a byte.
1f607577
KW
887
888=for apidoc Am|U8|toLOWER_LC|U8 ch
889Converts the specified character to lowercase using the current locale's rules,
890if possible; otherwise returns the input character itself.
891
d0da05db
KW
892=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
893Converts the code point C<cp> to its lowercase version, and
894stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
895point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
896that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
897bytes since the lowercase version may be longer than the original character.
898
899The first code point of the lowercased 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
KW
902
903=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
904Converts the UTF-8 encoded character at C<p> to its lowercase version, and
905stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
906that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
907bytes since the lowercase version may be longer than the original character.
908
909The first code point of the lowercased version is returned
21da7284
KW
910(but note, as explained at L<the top of this section|/Character case
911changing>, that there may be more).
1f607577
KW
912
913The input character at C<p> is assumed to be well-formed.
914
25200305
KW
915=for apidoc Am|U8|toTITLE|U8 ch
916Converts the specified character to titlecase. If the input is anything but an
917ASCII lowercase character, that input character itself is returned. Variant
b7d90381 918C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 919range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 920not a concept used in locale handling, so there is no functionality for that.)
25200305 921
d0da05db
KW
922=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
923Converts the code point C<cp> to its titlecase version, and
924stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
925point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
926that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
927bytes since the titlecase version may be longer than the original character.
928
929The first code point of the titlecased version is returned
21da7284
KW
930(but note, as explained at L<the top of this section|/Character case
931changing>, that there may be more).
1f607577
KW
932
933=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
934Converts the UTF-8 encoded character at C<p> to its titlecase version, and
935stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
936that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
937bytes since the titlecase version may be longer than the original character.
938
939The first code point of the titlecased version is returned
21da7284
KW
940(but note, as explained at L<the top of this section|/Character case
941changing>, that there may be more).
1f607577
KW
942
943The input character at C<p> is assumed to be well-formed.
944
954c1994 945=cut
353c9b6f 946
d0da05db 947XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
25200305 948really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 949
8a58bdcf 950Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
951patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
952
954c1994
GS
953*/
954
7c062697
KW
955/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
956 * is known only in the perl core, and this macro can be called from outside
957 * that */
958#ifdef HAS_QUAD
959# define WIDEST_UTYPE U64TYPE
960#else
961# define WIDEST_UTYPE U32
962#endif
963
3912bc88
KW
964/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
965 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
966 * bits of information are lost even on a 64-bit machine, but to get the
967 * compiler to optimize it out if possible. This is because Configure makes
968 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
969 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
970 *
971 * For Coverity, be always true, because otherwise Coverity thinks
972 * it finds several expressions that are always true, independent
973 * of operands. Well, they are, but that is kind of the point.
220c71bf 974 */
7e75d1a1 975#ifndef __COVERITY__
6c5b02ac
KW
976 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
977 * pointer) */
978#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 979 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
980#else
981#define FITS_IN_8_BITS(c) (1)
982#endif
cf301eb7 983
41f43cc2 984#ifdef EBCDIC
b6340bd0 985# ifndef _ALL_SOURCE
0852beac
KW
986 /* The native libc isascii() et.al. functions return the wrong results
987 * on at least z/OS unless this is defined. */
b6340bd0
KW
988# error _ALL_SOURCE should probably be defined
989# endif
41f43cc2 990#else
0852beac
KW
991 /* There is a simple definition of ASCII for ASCII platforms. But the
992 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 993 * other macros below.
3f3c579d
KW
994 *
995 * The cast here is used instead of '(c) >= 0', because some compilers emit
996 * a warning that that test is always true when the parameter is an
997 * unsigned type. khw supposes that it could be written as
998 * && ((c) == '\0' || (c) > 0)
999 * to avoid the message, but the cast will likely avoid extra branches even
1000 * with stupid compilers.
1001 *
1002 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1003 * a pointer) */
9c903d59 1004# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1005#endif
1006
38694112
KW
1007/* Take the eight possible bit patterns of the lower 3 bits and you get the
1008 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1009 * can be ignored. If the rest match '0', we have an octal */
1010#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1011
9fb1bf9d 1012#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1013
a500dc72
KW
1014/* Character class numbers. For internal core Perl use only. The ones less
1015 * than 32 are used in PL_charclass[] and the ones up through the one that
1016 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1017 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1018 * their actual definitions are here. If that file has a name not used here,
1019 * it won't compile.
1709d539
KW
1020 *
1021 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
1022 * frequency of their use. This gives a slight edge to exiting a loop earlier
1023 * (in reginclass() in regexec.c) */
1709d539
KW
1024# define _CC_WORDCHAR 0 /* \w and [:word:] */
1025# define _CC_DIGIT 1 /* \d and [:digit:] */
1026# define _CC_ALPHA 2 /* [:alpha:] */
1027# define _CC_LOWER 3 /* [:lower:] */
1028# define _CC_UPPER 4 /* [:upper:] */
1029# define _CC_PUNCT 5 /* [:punct:] */
1030# define _CC_PRINT 6 /* [:print:] */
15861f94 1031# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1032# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1033# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1709d539 1034
b0d691b2 1035#define _FIRST_NON_SWASH_CC 10
1709d539
KW
1036/* The character classes above are implemented with swashes. The second group
1037 * (just below) contains the ones implemented without. These are also sorted
1038 * in rough order of the frequency of their use, except that \v should be last,
1039 * as it isn't a real Posix character class, and some (small) inefficiencies in
1040 * regular expression handling would be introduced by putting it in the middle
1041 * of those that are. Also, cntrl and ascii come after the others as it may be
1042 * useful to group these which have no members that match above Latin1, (or
1043 * above ASCII in the latter case) */
1044
779cf272 1045# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1046# define _CC_BLANK 11 /* [:blank:] */
1047# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1048# define _CC_CNTRL 13 /* [:cntrl:] */
1049# define _CC_ASCII 14 /* [:ascii:] */
1050# define _CC_VERTSPACE 15 /* \v */
1709d539 1051
a0947d7b
KW
1052# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1053
1709d539 1054/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1055 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1056# define _CC_IDFIRST 16
1057# define _CC_CHARNAME_CONT 17
1058# define _CC_NONLATIN1_FOLD 18
1059# define _CC_NONLATIN1_SIMPLE_FOLD 19
1060# define _CC_QUOTEMETA 20
1061# define _CC_NON_FINAL_FOLD 21
1062# define _CC_IS_IN_SOME_FOLD 22
1063# define _CC_MNEMONIC_CNTRL 23
073c22b3
KW
1064
1065/* This next group is only used on EBCDIC platforms, so theoretically could be
1066 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
1067# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1068# define _CC_UTF8_IS_START 29
1069# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1070# define _CC_UTF8_IS_CONTINUATION 31
1071/* Unused: 24-27
f4cdb42c
KW
1072 * If more bits are needed, one could add a second word for non-64bit
1073 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1074 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1075 * is used only for optimization (as of this writing), and differs in the
1076 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1077 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1078 * would just cause /i regexes which match them to run less efficiently.
1079 * Similarly the EBCDIC-only bits are used just for speed, and could be
1080 * replaced by other means */
96ac0975 1081
3a371f2f
KW
1082#if defined(PERL_CORE) || defined(PERL_EXT)
1083/* An enum version of the character class numbers, to help compilers
1084 * optimize */
1085typedef enum {
3a371f2f 1086 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1087 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1088 _CC_ENUM_ASCII = _CC_ASCII,
1089 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1090 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1091 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1092 _CC_ENUM_DIGIT = _CC_DIGIT,
1093 _CC_ENUM_GRAPH = _CC_GRAPH,
1094 _CC_ENUM_LOWER = _CC_LOWER,
1095 _CC_ENUM_PRINT = _CC_PRINT,
1096 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1097 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1098 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1099 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1100 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1101 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1102} _char_class_number;
1103#endif
1104
63c61c3f 1105#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1106#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1107
b7d90381
KW
1108#if defined(PERL_IN_UTF8_C) \
1109 || defined(PERL_IN_REGCOMP_C) \
1110 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1111# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1112 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1113 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1114 #error Need to adjust order of swash_property_names[]
1115# endif
1116
b44daeeb
KW
1117/* This is declared static in each of the few files that this is #defined for
1118 * to keep them from being publicly accessible. Hence there is a small amount
1119 * of wasted space */
1120
922e8cb4
KW
1121static const char* const swash_property_names[] = {
1122 "XPosixWord",
1123 "XPosixDigit",
1124 "XPosixAlpha",
1125 "XPosixLower",
1126 "XPosixUpper",
1127 "XPosixPunct",
1128 "XPosixPrint",
1129 "XPosixAlnum",
b0d691b2
KW
1130 "XPosixGraph",
1131 "Cased"
922e8cb4
KW
1132};
1133#endif
1134
6635f04f 1135START_EXTERN_C
96ac0975
NC
1136# ifdef DOINIT
1137EXTCONST U32 PL_charclass[] = {
1138# include "l1_char_class_tab.h"
1139};
1140
1141# else /* ! DOINIT */
1142EXTCONST U32 PL_charclass[];
1143# endif
6635f04f 1144END_EXTERN_C
96ac0975 1145
265c1f46 1146 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1147# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1148
1149 /* For internal core Perl use only: the base macro for defining macros like
1150 * isALPHA */
ff7ecfc3 1151# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1152 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1153
f4cdb42c
KW
1154 /* The mask for the _A versions of the macros; it just adds in the bit for
1155 * ASCII. */
1156# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1157
4650c663
KW
1158 /* For internal core Perl use only: the base macro for defining macros like
1159 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1160 * the ASCII bit are present */
b7d90381
KW
1161# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1162 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1163 == _CC_mask_A(classnum)))
f4cdb42c 1164
f4cdb42c 1165# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1166# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1167# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1168# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1169# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1170# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1171# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1172# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1173# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1174# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1175# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1176# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1177# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1178 */
d95f8b6a 1179# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1180# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1181# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1182# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1183
1184 /* continuation character for legal NAME in \N{NAME} */
1185# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1186
1187# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1188# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1189# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1190# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1191# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1192# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1193# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1194# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1195# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1196# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1197
0852beac
KW
1198# ifdef EBCDIC
1199# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1200# endif
1201
f12c0118
KW
1202 /* Participates in a single-character fold with a character above 255 */
1203# 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)))
1204
1205 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1206# 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 1207
4eeeb416 1208# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1209# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1210 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1211# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1212 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1213# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1214 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1215#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1216
1217 /* If we don't have perl.h, we are compiling a utility program. Below we
1218 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1219 * to it. Most are coded based on first principles. These are written to
74665a89 1220 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
0852beac
KW
1221# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1222# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1223# define isSPACE_A(c) (isBLANK_A(c) \
1224 || (c) == '\n' \
1225 || (c) == '\r' \
1226 || (c) == '\v' \
0852beac 1227 || (c) == '\f')
74665a89
KW
1228 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1229 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1230 * performance (if optimization isn't on), and allow the same code to be
1231 * used for both platform types */
1232# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1233 && ( (c) <= 'i' \
1234 || ((c) >= 'j' && (c) <= 'r') \
1235 || (c) >= 's'))
1236# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1237 && ( (c) <= 'I' \
1238 || ((c) >= 'J' && (c) <= 'R') \
1239 || (c) >= 'S'))
a4d7a999
KW
1240# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1241# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1242# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1243# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
74665a89
KW
1244# define isXDIGIT_A(c) (isDIGIT_A(c) \
1245 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1246 || ((c) <= 'F' && (c) >= 'A'))
74665a89
KW
1247# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1248 || (c) == '#' || (c) == '$' || (c) == '%' \
1249 || (c) == '&' || (c) == '\'' || (c) == '(' \
1250 || (c) == ')' || (c) == '*' || (c) == '+' \
1251 || (c) == ',' || (c) == '.' || (c) == '/' \
1252 || (c) == ':' || (c) == ';' || (c) == '<' \
1253 || (c) == '=' || (c) == '>' || (c) == '?' \
1254 || (c) == '@' || (c) == '[' || (c) == '\\' \
1255 || (c) == ']' || (c) == '^' || (c) == '_' \
1256 || (c) == '`' || (c) == '{' || (c) == '|' \
1257 || (c) == '}' || (c) == '~')
1258# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1259# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1260
0852beac 1261# ifdef EBCDIC
74665a89
KW
1262 /* The below is accurate for the 3 EBCDIC code pages traditionally
1263 * supported by perl. The only difference between them in the controls
1264 * is the position of \n, and that is represented symbolically below */
1265# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1266 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1267 || (c) == '\t' || (c) == '\v' \
1268 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1269 || (c) == 7 /* U+7F DEL */ \
1270 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1271 /* DLE, DC[1-3] */ \
1272 || (c) == 0x18 /* U+18 CAN */ \
1273 || (c) == 0x19 /* U+19 EOM */ \
1274 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1275 || (c) == 0x26 /* U+17 ETB */ \
1276 || (c) == 0x27 /* U+1B ESC */ \
1277 || (c) == 0x2D /* U+05 ENQ */ \
1278 || (c) == 0x2E /* U+06 ACK */ \
1279 || (c) == 0x32 /* U+16 SYN */ \
1280 || (c) == 0x37 /* U+04 EOT */ \
1281 || (c) == 0x3C /* U+14 DC4 */ \
1282 || (c) == 0x3D /* U+15 NAK */ \
1283 || (c) == 0x3F)/* U+1A SUB */
0852beac 1284# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1285# else /* isASCII is already defined for ASCII platforms, so can use that to
1286 define isCNTRL */
1287# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1288# endif
1289
3ffc8c70 1290 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1291 * during debugging, and it seems best to keep them. We may be called
1292 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1293 * do anything anyway, so make it not a problem */
1294# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1295# define NATIVE_TO_LATIN1(ch) (ch)
1296# endif
3ded5eb0
KW
1297# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1298# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1299# define isBLANK_L1(c) (isBLANK_A(c) \
1300 || (FITS_IN_8_BITS(c) \
1301 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1302# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1303# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1304# define isLOWER_L1(c) (isLOWER_A(c) \
1305 || (FITS_IN_8_BITS(c) \
1306 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1307 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1308 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1309 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1310 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1311# define isPRINT_L1(c) (isPRINT_A(c) \
1312 || (FITS_IN_8_BITS(c) \
1313 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1314# define isPUNCT_L1(c) (isPUNCT_A(c) \
1315 || (FITS_IN_8_BITS(c) \
1316 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1317 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1318 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1319 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1320 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1321 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1322 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1323# define isSPACE_L1(c) (isSPACE_A(c) \
1324 || (FITS_IN_8_BITS(c) \
1325 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1326 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1327# define isUPPER_L1(c) (isUPPER_A(c) \
1328 || (FITS_IN_8_BITS(c) \
1329 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1330 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1331 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1332# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1333# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1334# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1335 || isBLANK_L1(c) \
1336 || (c) == '-' \
1337 || (c) == '(' \
1338 || (c) == ')')
1339 /* The following are not fully accurate in the above-ASCII range. I (khw)
1340 * don't think it's necessary to be so for the purposes where this gets
1341 * compiled */
1342# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1343# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1344
1345 /* And these aren't accurate at all. They are useful only for above
1346 * Latin1, which utilities and bootstrapping don't deal with */
1347# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1348# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1349# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1350
1351 /* Many of the macros later in this file are defined in terms of these. By
1352 * implementing them with a function, which converts the class number into
1353 * a call to the desired macro, all of the later ones work. However, that
1354 * function won't be actually defined when building a utility program (no
1355 * perl.h), and so a compiler error will be generated if one is attempted
1356 * to be used. And the above-Latin1 code points require Unicode tables to
1357 * be present, something unlikely to be the case when bootstrapping */
1358# define _generic_isCC(c, classnum) \
1359 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1360# define _generic_isCC_A(c, classnum) \
1361 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1362#endif /* End of no perl.h H_PERL */
8a58bdcf 1363
e66b99e9
KW
1364#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1365#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1366#define isASCII_A(c) isASCII(c)
1367#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1368#define isBLANK(c) isBLANK_A(c)
1369#define isCNTRL(c) isCNTRL_A(c)
1370#define isDIGIT(c) isDIGIT_A(c)
1371#define isGRAPH(c) isGRAPH_A(c)
1372#define isIDFIRST(c) isIDFIRST_A(c)
1373#define isLOWER(c) isLOWER_A(c)
1374#define isPRINT(c) isPRINT_A(c)
779cf272 1375#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1376#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1377#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1378#define isPUNCT(c) isPUNCT_A(c)
1379#define isSPACE(c) isSPACE_A(c)
1380#define isUPPER(c) isUPPER_A(c)
1381#define isWORDCHAR(c) isWORDCHAR_A(c)
1382#define isXDIGIT(c) isXDIGIT_A(c)
1383
1384/* ASCII casing. These could also be written as
1385 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1386 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1387 which uses table lookup and mask instead of subtraction. (This would
1388 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1389#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1390#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1391
25200305
KW
1392/* In the ASCII range, these are equivalent to what they're here defined to be.
1393 * But by creating these definitions, other code doesn't have to be aware of
1394 * this detail */
1395#define toFOLD(c) toLOWER(c)
25200305
KW
1396#define toTITLE(c) toUPPER(c)
1397
c753c8d3
KW
1398#define toLOWER_A(c) toLOWER(c)
1399#define toUPPER_A(c) toUPPER(c)
25200305
KW
1400#define toFOLD_A(c) toFOLD(c)
1401#define toTITLE_A(c) toTITLE(c)
1a0901db 1402
4650c663 1403/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1404#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1405 ? (c) \
f4cd282c 1406 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1407#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1408
1a0901db 1409/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1410 * all mapped to one of them, and these need special handling; returns the
1411 * input itself if is out-of-range */
b2bf251f 1412#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1413 ? (c) \
f4cd282c 1414 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1415#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1416
beab9ebe
KW
1417/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1418
1419/* For internal core Perl use only: the base macro for defining macros like
1420 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1421 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1422 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1423 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1424 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1425 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1426 * code points under locale rules, use the macros based on _generic_LC_uvchr
1427 * instead of this. */
beab9ebe
KW
1428#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1429 (! FITS_IN_8_BITS(c) \
1430 ? 0 \
31f05a37
KW
1431 : IN_UTF8_CTYPE_LOCALE \
1432 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1433 : cBOOL(non_utf8))
1434
1435/* For internal core Perl use only: a helper macro for defining macros like
1436 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1437 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1438 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1439 * _generic_LC_base for more info */
1440#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1441 _generic_LC_base(c,utf8_locale_classnum, \
1442 non_utf8_func( (_LC_CAST) (c)))
1443
1444/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1445 * 'c' is the platform's native underscore character */
1446#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1447 _generic_LC_base(c, utf8_locale_classnum, \
1448 (non_utf8_func( (_LC_CAST) (c)) \
1449 || (char)(c) == '_'))
1450
1451/* These next three are also for internal core Perl use only: case-change
1452 * helper macros */
1453#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1454 ? (c) \
31f05a37
KW
1455 : (IN_UTF8_CTYPE_LOCALE) \
1456 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1457 : (cast)function((cast)(c)))
beab9ebe 1458
31f05a37
KW
1459/* Note that the result can be larger than a byte in a UTF-8 locale. It
1460 * returns a single value, so can't adequately return the upper case of LATIN
1461 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1462 * values "SS"); instead it asserts against that under DEBUGGING, and
1463 * otherwise returns its input */
beab9ebe
KW
1464#define _generic_toUPPER_LC(c, function, cast) \
1465 (! FITS_IN_8_BITS(c) \
1466 ? (c) \
31f05a37 1467 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1468 ? (cast)function((cast)(c)) \
31f05a37
KW
1469 : ((((U8)(c)) == MICRO_SIGN) \
1470 ? GREEK_CAPITAL_LETTER_MU \
1471 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1472 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1473 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1474 ? (__ASSERT_(0) (c)) \
1475 : PL_mod_latin1_uc[ (U8) (c) ])))))
1476
1477/* Note that the result can be larger than a byte in a UTF-8 locale. It
1478 * returns a single value, so can't adequately return the fold case of LATIN
1479 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1480 * values "ss"); instead it asserts against that under DEBUGGING, and
1481 * otherwise returns its input */
beab9ebe 1482#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1483 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1484 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1485 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1486 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1487 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1488
84061b6a 1489/* Use the libc versions for these if available. */
f05550c0 1490#if defined(HAS_ISASCII)
84061b6a
KW
1491# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1492#else
1493# define isASCII_LC(c) isASCII(c)
1494#endif
1495
f05550c0 1496#if defined(HAS_ISBLANK)
84061b6a 1497# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1498#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1499# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1500#endif
1501
f05550c0 1502#define _LC_CAST U8
bbce6d69 1503
f05550c0 1504#ifdef WIN32
375f5f06
KW
1505 /* The Windows functions don't bother to follow the POSIX standard, which
1506 * for example says that something can't both be a printable and a control.
1507 * But Windows treats the \t control as a printable, and does such things
1508 * as making superscripts into both digits and punctuation. This tames
1509 * these flaws by assuming that the definitions of both controls and space
1510 * are correct, and then making sure that other definitions don't have
1511 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1512 * Not all possible weirdnesses are checked for, just the ones that were
1513 * detected on actual Microsoft code pages */
1514
b7d90381
KW
1515# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1516# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1517
1518# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1519 && isALPHANUMERIC_LC(c))
1520# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1521 ! isPUNCT_LC(c))
1522# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1523 isALPHANUMERIC_LC(c))
1524# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1525# define isIDFIRST_LC(c) (((c) == '_') \
1526 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1527# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1528# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1529# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1530# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1531# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1532# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1533 && isALPHANUMERIC_LC(c))
f05550c0
BF
1534
1535# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1536# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1537# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1538
1539#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1540 /* For most other platforms */
beab9ebe 1541
f05550c0
BF
1542# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1543# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1544# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1545# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1546# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1547# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1548# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1549# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1550# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1551# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1552# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1553# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1554# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1555
1556
1557# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1558# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1559# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1560
1561#else /* The final fallback position */
1562
b7d90381
KW
1563# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1564# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1565# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1566# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1567# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1568# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1569# define isLOWER_LC(c) (isascii(c) && islower(c))
1570# define isPRINT_LC(c) (isascii(c) && isprint(c))
1571# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1572# define isSPACE_LC(c) (isascii(c) && isspace(c))
1573# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1574# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1575# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1576
1577# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1578# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1579# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1580
f05550c0 1581#endif
55204971 1582
eba68aa0
KW
1583#define isIDCONT(c) isWORDCHAR(c)
1584#define isIDCONT_A(c) isWORDCHAR_A(c)
1585#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1586#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1587#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1588
4650c663 1589/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1590 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1591 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1592 * classes where there is a macro or function 'above_latin1' that takes the
1593 * single argument 'c' and returns the desired value. These exist for those
1594 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1595 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1596 * for classes where that overhead is faster than a direct lookup.
d0da05db 1597 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1598 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1599 * have duplicate checks here, so could create versions of the macros that
1600 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1601
1602/* Note that all ignore 'use bytes' */
d0da05db 1603#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
cd500f2f
KW
1604 ? _generic_isCC(c, classnum) \
1605 : above_latin1(c))
d0da05db 1606#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
cd500f2f 1607 ? _generic_isCC(c, classnum) \
922e8cb4 1608 : _is_uni_FOO(classnum, c))
d0da05db
KW
1609#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1610#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1611#define isASCII_uvchr(c) isASCII(c)
1612#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1613#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1614#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1615#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1616#define isIDCONT_uvchr(c) _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1617#define isIDFIRST_uvchr(c) _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
1618#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1619#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1620
1621#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1622#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1623#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1624
1625#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1626#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1627#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1628#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1629
1630#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1631#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1632#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1633#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1634
1635/* For backwards compatibility, even though '_uni' should mean official Unicode
1636 * code points, in Perl it means native for those below 256 */
1637#define isALPHA_uni(c) isALPHA_uvchr(c)
1638#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1639#define isASCII_uni(c) isASCII_uvchr(c)
1640#define isBLANK_uni(c) isBLANK_uvchr(c)
1641#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1642#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1643#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1644#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1645#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1646#define isLOWER_uni(c) isLOWER_uvchr(c)
1647#define isPRINT_uni(c) isPRINT_uvchr(c)
1648#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1649#define isSPACE_uni(c) isSPACE_uvchr(c)
1650#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1651#define isUPPER_uni(c) isUPPER_uvchr(c)
1652#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1653#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1654#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1655#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1656#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1657#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1658#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1659
4650c663
KW
1660/* For internal core Perl use only: the base macros for defining macros like
1661 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1662 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1663 * simple class definitions; the other for more complex. These are like
d0da05db 1664 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1665#define _generic_LC_uvchr(latin1, above_latin1, c) \
1666 (c < 256 ? latin1(c) : above_latin1(c))
1667#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1668 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1669
1670#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1671#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1672 _CC_ALPHANUMERIC, c)
b7d90381
KW
1673#define isASCII_LC_uvchr(c) isASCII_LC(c)
1674#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1675 is_HORIZWS_cp_high, c)
feeab5a9 1676#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1677#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1678#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1679#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1680 _is_uni_perl_idcont, c)
b7d90381 1681#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1682 _is_uni_perl_idstart, c)
1683#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1684#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1685#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1686#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1687#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1688 is_XPERLSPACE_cp_high, c)
cd500f2f 1689#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1690#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1691 _CC_WORDCHAR, c)
b7d90381
KW
1692#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1693 is_XDIGIT_cp_high, c)
e712593e 1694
b7d90381 1695#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1696
4650c663
KW
1697/* For internal core Perl use only: the base macros for defining macros like
1698 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1699 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1700 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1701 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1702 * same representation whether utf8 or not. Note that it assumes that the utf8
1703 * has been validated, and ignores 'use bytes' */
cd500f2f 1704#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1705 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1706 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1707 ? _generic_isCC( \
a62b247b 1708 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
aea0c08c 1709 *((p)+1 )), \
94bb8c36 1710 classnum) \
252810d0 1711 : utf8)
da8c1a98
KW
1712
1713/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
1714 * they don't otherwise go out of their way to look for malformed UTF-8. If
1715 * they can return accurate results without knowing if the input is otherwise
1716 * malformed, they do so. For example isASCII is accurate in spite of any
1717 * non-length malformations because it looks only at a single byte. Likewise
1718 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
1719 * variant ones return FALSE. But, if the input has to be well-formed in order
1720 * for the results to be accurate, the macros will test and if malformed will
1721 * call a routine to die
1722 *
1723 * Except for toke.c, the macros do assume that e > p, asserting that on
1724 * DEBUGGING builds. Much code that calls these depends on this being true,
1725 * for other reasons. toke.c is treated specially as using the regular
1726 * assertion breaks it in many ways. All strings that these operate on there
1727 * are supposed to have an extra NUL character at the end, so that *e = \0. A
1728 * bunch of code in toke.c assumes that this is true, so the assertion allows
1729 * for that */
1730#ifdef PERL_IN_TOKE_C
1731# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
1732#else
1733# define _utf8_safe_assert(p,e) ((e) > (p))
1734#endif
1735
1736#define _generic_utf8_safe(classnum, p, e, above_latin1) \
1737 (__ASSERT_(_utf8_safe_assert(p, e)) \
1738 (UTF8_IS_INVARIANT(*(p))) \
1739 ? _generic_isCC(*(p), classnum) \
1740 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1741 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1742 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
1743 classnum) \
1744 : (_force_out_malformed_utf8_message( \
1745 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1746 : above_latin1))
b7d90381
KW
1747/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1748 * 'above_latin1' can be a macro */
cd500f2f
KW
1749#define _generic_func_utf8(classnum, above_latin1, p) \
1750 _generic_utf8(classnum, p, above_latin1(p))
da8c1a98
KW
1751#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
1752 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
1753#define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1754 _generic_utf8_safe(classnum, p, e, \
1755 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1756 ? (_force_out_malformed_utf8_message( \
1757 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1758 : above_latin1(p)))
4650c663 1759/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1760 * 'above_latin1' parameter */
922e8cb4 1761#define _generic_swash_utf8(classnum, p) \
cd500f2f 1762 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
da8c1a98
KW
1763#define _generic_swash_utf8_safe(classnum, p, e) \
1764_generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e))
922e8cb4 1765
cc8ab7c0 1766/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1767 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1768 * class is TRUE for. Hence it can skip the tests for this range.
1769 * 'above_latin1' should include its arguments */
3ded5eb0 1770#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1771 (UTF8_IS_INVARIANT(*(p)) \
1772 ? _generic_isCC(*(p), classnum) \
1773 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1774 ? above_latin1 \
cc8ab7c0
KW
1775 : 0)
1776
da8c1a98
KW
1777#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
1778 (__ASSERT_(_utf8_safe_assert(p, e)) \
1779 (UTF8_IS_INVARIANT(*(p))) \
1780 ? _generic_isCC(*(p), classnum) \
1781 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1782 ? 0 /* Note that doesn't check validity for latin1 */ \
1783 : above_latin1)
1784
84238efa
KW
1785/* NOTE that some of these macros have very similar ones in regcharclass.h.
1786 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1787 * there, differing in name only by an underscore from the one here
1788 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1789 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1790 * points; the regcharclass.h ones are implemented as a series of
1791 * "if-else-if-else ..." */
1792
b7d90381
KW
1793#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1794#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1795#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
509fb054
KW
1796 utf8, the non-utf8 macro works
1797 */
b7d90381 1798#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2 1799
da8c1a98
KW
1800#define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e)
1801#define isALPHANUMERIC_utf8_safe(p, e) \
1802 _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e)
1803#define isASCII_utf8_safe(p, e) \
1804 /* Because ASCII is invariant under utf8, the non-utf8 macro \
1805 * works */ \
1806 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
1807#define isBLANK_utf8_safe(p, e) \
1808 _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
1809
e8fa43e2
KW
1810#ifdef EBCDIC
1811 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1812 * more efficient macro instead of the more general one */
88794300 1813# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
da8c1a98
KW
1814# define isCNTRL_utf8_safe(p, e) \
1815 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p))
e8fa43e2 1816#else
da8c1a98
KW
1817# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1818# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
1819#endif
1820
cd500f2f
KW
1821#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1822 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1823#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1824#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1825 _is_utf8_perl_idcont, p)
da8c1a98
KW
1826#define isDIGIT_utf8_safe(p, e) \
1827 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
1828 _is_utf8_FOO_with_len(_CC_DIGIT, p, e))
1829#define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e)
1830#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
1831 _is_utf8_perl_idcont_with_len, p, e)
e5dcd934 1832
c11ff943
KW
1833/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1834 * IDFIRST is an alnum. See
c8362b00 1835 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1836 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1837 * This used to be not the XID version, but we decided to go with the more
1838 * modern Unicode definition */
b7d90381 1839#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1840 _is_utf8_perl_idstart, p)
e5dcd934 1841
b7d90381
KW
1842#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1843#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1844#define isPSXSPC_utf8(p) isSPACE_utf8(p)
1845#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1846#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1847#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1848#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1849#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1850#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
cd500f2f 1851 is_XDIGIT_high(p))
da8c1a98
KW
1852#define isIDFIRST_utf8_safe(p, e) \
1853 _generic_func_utf8_safe(_CC_IDFIRST, \
1854 _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e))
1855
1856#define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e)
1857#define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e)
1858#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
1859#define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e)
1860#define isSPACE_utf8_safe(p, e) \
1861 _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
1862#define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e)
1863#define isVERTWS_utf8_safe(p, e) \
1864 _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
1865#define isWORDCHAR_utf8_safe(p, e) \
1866 _generic_swash_utf8_safe(_CC_WORDCHAR, p, e)
1867#define isXDIGIT_utf8_safe(p, e) \
1868 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
1869 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1870 ? (_force_out_malformed_utf8_message( \
1871 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1872 : is_XDIGIT_high(p)))
a0ed51b3 1873
7d2b61dd 1874#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1875#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1876#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1877#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1878
4650c663
KW
1879/* For internal core Perl use only: the base macros for defining macros like
1880 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1881 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1882 * 'macro' parameter */
94bb8c36
KW
1883#define _generic_LC_utf8(macro, p, utf8) \
1884 (UTF8_IS_INVARIANT(*(p)) \
1885 ? macro(*(p)) \
1886 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
a62b247b 1887 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
9201a572
KW
1888 : utf8)
1889
cd500f2f
KW
1890#define _generic_LC_swash_utf8(macro, classnum, p) \
1891 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1892#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1893 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1894
b7d90381 1895#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
cd500f2f 1896 _CC_ALPHANUMERIC, p)
b7d90381
KW
1897#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1898#define isASCII_LC_utf8(p) isASCII_LC(*p)
1899#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1900 is_HORIZWS_high, p)
1901#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1902#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1903#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1904#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1905 _is_utf8_perl_idcont, p)
1906#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1907 _is_utf8_perl_idstart, p)
1908#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1909#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1910#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1911#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1912#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1913 is_XPERLSPACE_high, p)
1914#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
cd500f2f
KW
1915#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1916 _CC_WORDCHAR, p)
b7d90381
KW
1917#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1918 is_XDIGIT_high, p)
da8c1a98
KW
1919/* For internal core Perl use only: the base macros for defining macros like
1920 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
1921 * point in 'p' is within the 0-255 range, it uses locale rules from the
1922 * passed-in 'macro' parameter */
1923#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
1924 (__ASSERT_(_utf8_safe_assert(p, e)) \
1925 (UTF8_IS_INVARIANT(*(p))) \
1926 ? macro(*(p)) \
1927 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1928 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1929 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
1930 : (_force_out_malformed_utf8_message( \
1931 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1932 : above_latin1))
1933
1934#define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \
1935 _generic_LC_utf8_safe(macro, p, e, \
1936 _is_utf8_FOO_with_len(classnum, p, e))
1937
1938#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
1939 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
1940
1941#define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1942 _generic_LC_utf8_safe(classnum, p, e, \
1943 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1944 ? (_force_out_malformed_utf8_message( \
1945 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1946 : above_latin1(p)))
1947
1948#define isALPHANUMERIC_LC_utf8_safe(p, e) \
1949 _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \
1950 _CC_ALPHANUMERIC, p, e)
1951#define isALPHA_LC_utf8_safe(p, e) \
1952 _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
1953#define isASCII_LC_utf8_safe(p, e) \
1954 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
1955#define isBLANK_LC_utf8_safe(p, e) \
1956 _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
1957#define isCNTRL_LC_utf8_safe(p, e) \
1958 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
1959#define isDIGIT_LC_utf8_safe(p, e) \
1960 _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
1961#define isGRAPH_LC_utf8_safe(p, e) \
1962 _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
1963#define isIDCONT_LC_utf8_safe(p, e) \
1964 _generic_LC_func_utf8_safe(isIDCONT_LC, \
1965 _is_utf8_perl_idcont_with_len, p, e)
1966#define isIDFIRST_LC_utf8_safe(p, e) \
1967 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
1968 _is_utf8_perl_idstart_with_len, p, e)
1969#define isLOWER_LC_utf8_safe(p, e) \
1970 _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
1971#define isPRINT_LC_utf8_safe(p, e) \
1972 _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
1973#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
1974#define isPUNCT_LC_utf8_safe(p, e) \
1975 _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
1976#define isSPACE_LC_utf8_safe(p, e) \
1977 _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
1978#define isUPPER_LC_utf8_safe(p, e) \
1979 _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
1980#define isWORDCHAR_LC_utf8_safe(p, e) \
1981 _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
1982#define isXDIGIT_LC_utf8_safe(p, e) \
1983 _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 1984
fbc19f27
KW
1985/* Macros for backwards compatibility and for completeness when the ASCII and
1986 * Latin1 values are identical */
b7d90381
KW
1987#define isALPHAU(c) isALPHA_L1(c)
1988#define isDIGIT_L1(c) isDIGIT_A(c)
1989#define isOCTAL(c) isOCTAL_A(c)
1990#define isOCTAL_L1(c) isOCTAL_A(c)
1991#define isXDIGIT_L1(c) isXDIGIT_A(c)
1992#define isALNUM(c) isWORDCHAR(c)
1993#define isALNUMU(c) isWORDCHAR_L1(c)
1994#define isALNUM_LC(c) isWORDCHAR_LC(c)
1995#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 1996#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
1997#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1998#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1999#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2000#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2001#define isALNUMC(c) isALPHANUMERIC(c)
2002#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2003#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2004#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 2005#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 2006#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 2007
2bd1cbf6
KW
2008/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2009 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2010 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2011 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2012 * the outlier from the block that contains the other controls, just like
2013 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2014 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2015 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2016 * etc. */
2bd1cbf6 2017#ifndef EBCDIC
75763b3a 2018# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2019#else
75763b3a
KW
2020# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2021 ((isPRINT_A(c)) \
2022 ? (UNLIKELY((c) == '?') \
2023 ? QUESTION_MARK_CTRL \
2024 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2025 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2026 ? '?' \
2027 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2028#endif
bbce6d69 2029
dea28490
JJ
2030/* Line numbers are unsigned, 32 bits. */
2031typedef U32 line_t;
e5dcd934 2032#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2033
91152fc1
DG
2034/* Helpful alias for version prescan */
2035#define is_LAX_VERSION(a,b) \
2036 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2037
2038#define is_STRICT_VERSION(a,b) \
2039 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2040
2041#define BADVERSION(a,b,c) \
2042 if (b) { \
2043 *b = c; \
2044 } \
2045 return a;
8c52afec 2046
cb27eebd
KW
2047/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2048 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2049 * which is advanced. The input is validated only by an assert() in DEBUGGING
2050 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2051 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2052#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2053 ? (c) \
2054 : ((c) + 9))))
902008b8
KW
2055#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2056 ? (*(s)++) \
2057 : (*(s)++ + 9))))
95a59cab 2058
cb27eebd
KW
2059/* Converts a character known to represent an octal digit (0-7) to its numeric
2060 * value. The input is validated only by an assert() in DEBUGGING builds. In
2061 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2062#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2063
305b8651
KW
2064/* Efficiently returns a boolean as to if two native characters are equivalent
2065 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2066 * the ALPHA in the name is to remind you of that. This is asserted() in
2067 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2068 * works (on valid input) for both non- and UTF-8-encoded bytes.
2069 *
2070 * When one of the inputs is a compile-time constant and gets folded by the
2071 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2072 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2073 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2074 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2075 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2076 * just a single 0, in the bit position where the upper- and lowercase differ.
2077 * */
305b8651
KW
2078#define isALPHA_FOLD_EQ(c1, c2) \
2079 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2080 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2081#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2082
8e84507e 2083/*
ccfc67b7
JH
2084=head1 Memory Management
2085
a02a5408 2086=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2087The XSUB-writer's interface to the C C<malloc> function.
2088
596f7718 2089Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2090
c5008215
JC
2091In 5.9.3, Newx() and friends replace the older New() API, and drops
2092the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2093themselves. This aid has been superseded by a new build option,
d10b4965 2094PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2095there for use in XS modules supporting older perls.
2096
a02a5408 2097=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2098The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2099cast. See also C<L</Newx>>.
954c1994 2100
596f7718 2101Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2102
a02a5408 2103=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2104The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2105memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2106
596f7718 2107Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2108
954c1994
GS
2109=for apidoc Am|void|Renew|void* ptr|int nitems|type
2110The XSUB-writer's interface to the C C<realloc> function.
2111
596f7718 2112Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2113
954c1994
GS
2114=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2115The XSUB-writer's interface to the C C<realloc> function, with
2116cast.
2117
596f7718 2118Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2119
49b8b560 2120=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2121The XSUB-writer's interface to the C C<free> function.
2122
596f7718 2123This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2124
954c1994
GS
2125=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2126The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2127source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2128C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2129
e90e2364 2130=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2131Like C<Move> but returns C<dest>. Useful
72d33970 2132for encouraging compilers to tail-call
e90e2364
NC
2133optimise.
2134
954c1994
GS
2135=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2136The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2137source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2138C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2139
e90e2364
NC
2140=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2141
796b6530 2142Like C<Copy> but returns C<dest>. Useful
72d33970 2143for encouraging compilers to tail-call
e90e2364
NC
2144optimise.
2145
954c1994
GS
2146=for apidoc Am|void|Zero|void* dest|int nitems|type
2147
2148The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2149destination, C<nitems> is the number of items, and C<type> is the type.
2150
e90e2364
NC
2151=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2152
72d33970
FC
2153Like C<Zero> but returns dest. Useful
2154for encouraging compilers to tail-call
e90e2364
NC
2155optimise.
2156
da5d8dbb 2157=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2158This is an architecture-independent macro to copy one structure to another.
954c1994 2159
7e337ee0
JH
2160=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2161
2162Fill up memory with a byte pattern (a byte repeated over and over
2163again) that hopefully catches attempts to access uninitialized memory.
2164
2165=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2166
2167PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2168
1c12ffb4 2169=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2170
2171PoisonWith(0xEF) for catching access to freed memory.
2172
9965345d
JH
2173=for apidoc Am|void|Poison|void* dest|int nitems|type
2174
7e337ee0 2175PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2176
2177=cut */
954c1994 2178
561b68a9
SH
2179/* Maintained for backwards-compatibility only. Use newSV() instead. */
2180#ifndef PERL_CORE
ff06c60c 2181#define NEWSV(x,len) newSV(len)
561b68a9 2182#endif
ff06c60c 2183
19a94d75
JH
2184#define MEM_SIZE_MAX ((MEM_SIZE)~0)
2185
e6bdf523 2186
27d5b266 2187#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2188
2189/* This expression will be constant-folded at compile time. It checks
2190 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2191 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2192 * As well as avoiding the need for a run-time check in some cases, it's
2193 * designed to avoid compiler warnings like:
2194 * comparison is always false due to limited range of data type
73e8ff00
DM
2195 * It's mathematically equivalent to
2196 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2197 */
2198
2199# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
004073ba 2200 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
e6bdf523 2201
88f9f128 2202/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2203 * compiler warnings. We *want* to write the expression as
2204 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2205 * (for some compile-time constant C), but even when the LHS
2206 * constant-folds to false at compile-time, g++ insists on emitting
2207 * warnings about the RHS (e.g. "comparison is always false"), so instead
2208 * we write it as
e6bdf523 2209 *
d98e5cde 2210 * (cond ? n : X) > C
88f9f128 2211 *
d98e5cde
DM
2212 * where X is a constant with X > C always false. Choosing a value for X
2213 * is tricky. If 0, some compilers will complain about 0 > C always being
2214 * false; if 1, Coverity complains when n happens to be the constant value
2215 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2216 * for X and hope that nothing else whines.
e6bdf523
DM
2217 */
2218
2219# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2220 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2221 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2222
2223# define MEM_WRAP_CHECK(n,t) \
2224 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2225 && (croak_memory_wrap(),0))
2226
2227# define MEM_WRAP_CHECK_1(n,t,a) \
2228 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2229 && (Perl_croak_nocontext("%s",(a)),0))
2230
8b44ba4c 2231#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2232
d1decf2b 2233#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
2234#else
2235
410319be
NC
2236#define MEM_WRAP_CHECK(n,t)
2237#define MEM_WRAP_CHECK_1(n,t,a)
2238#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
2239#define MEM_WRAP_CHECK_(n,t)
2240
6fff79ce 2241#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 2242
1936d2a7 2243#endif
8b44ba4c 2244
fe4f188c 2245#ifdef PERL_MEM_LOG
46c6c7e2 2246/*
9f653bb5 2247 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2248 * go through functions, which are handy for debugging breakpoints, but
2249 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2250 * line number, and C function name if available) passed in. This info can
2251 * then be used for logging the calls, for which one gets a sample
73d1d973 2252 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2253 *
46c6c7e2 2254 * Known problems:
94e892a6 2255 * - not all memory allocs get logged, only those
46c6c7e2 2256 * that go through Newx() and derivatives (while all
94e892a6 2257 * Safefrees do get logged)
46c6c7e2
JH
2258 * - __FILE__ and __LINE__ do not work everywhere
2259 * - __func__ or __FUNCTION__ even less so
2260 * - I think more goes on after the perlio frees but
2261 * the thing is that STDERR gets closed (as do all
2262 * the file descriptors)
2263 * - no deeper calling stack than the caller of the Newx()
2264 * or the kind, but do I look like a C reflection/introspection
2265 * utility to you?
2266 * - the function prototypes for the logging functions
2267 * probably should maybe be somewhere else than handy.h
2268 * - one could consider inlining (macrofying) the logging
2269 * for speed, but I am too lazy
2270 * - one could imagine recording the allocations in a hash,
2271 * (keyed by the allocation address?), and maintain that
2272 * through reallocs and frees, but how to do that without
2273 * any News() happening...?
73d1d973 2274 * - lots of -Ddefines to get useful/controllable output
b953482e 2275 * - lots of ENV reads
46c6c7e2
JH
2276 */
2277
0b0ab801 2278# ifdef PERL_CORE
73d1d973 2279# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2280enum mem_log_type {
2281 MLT_ALLOC,
2282 MLT_REALLOC,
d7a2c63c
MHM
2283 MLT_FREE,
2284 MLT_NEW_SV,
2285 MLT_DEL_SV
0b0ab801
MHM
2286};
2287# endif
12754f92 2288# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2289void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2290void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2291# endif
0b0ab801
MHM
2292# endif
2293
fe4f188c
JH
2294#endif
2295
2296#ifdef PERL_MEM_LOG
d1401ee9
MHM
2297#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2298#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2299#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2300#endif
2301
2302#ifndef MEM_LOG_ALLOC
2303#define MEM_LOG_ALLOC(n,t,a) (a)
2304#endif
2305#ifndef MEM_LOG_REALLOC
2306#define MEM_LOG_REALLOC(n,t,v,a) (a)
2307#endif
2308#ifndef MEM_LOG_FREE
2309#define MEM_LOG_FREE(a) (a)
2310#endif
2311
d1401ee9
MHM
2312#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2313#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2314#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2315
2316#ifndef PERL_CORE
a02a5408
JC
2317/* pre 5.9.x compatibility */
2318#define New(x,v,n,t) Newx(v,n,t)
2319#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2320#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2321#endif
a02a5408 2322
ff68c719 2323#define Renew(v,n,t) \
d1401ee9 2324 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2325#define Renewc(v,n,t,c) \
d1401ee9 2326 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2327
2328#ifdef PERL_POISON
2329#define Safefree(d) \
06c0cc96 2330 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2331#else
fe4f188c 2332#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2333#endif
55497cff 2334
8b44ba4c
NC
2335#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2336#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2337#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2338
8b44ba4c
NC
2339#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2340#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2341#ifdef HAS_MEMSET
8b44ba4c 2342#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2343#else
8b44ba4c
NC
2344/* Using bzero(), which returns void. */
2345#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2346#endif
2347
7e337ee0
JH
2348#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2349#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2350#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2351#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2352
caa674f3
DD
2353#ifdef PERL_POISON
2354# define PERL_POISON_EXPR(x) x
2355#else
2356# define PERL_POISON_EXPR(x)
2357#endif
2358
2304df62 2359#ifdef USE_STRUCT_COPY
ff68c719 2360#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2361#else
2362#define StructCopy(s,d,t) Copy(s,d,1,t)
2363#endif
2cc61e15 2364
c3caa5c3
JH
2365/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2366 * want your zero-based indices to be less than but not equal to).
2367 *
2368 * C_ARRAY_END is one past the last: half-open/half-closed range,
2369 * not last-inclusive range. */
622913ab 2370#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2371#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2372
2cc61e15
DD
2373#ifdef NEED_VA_COPY
2374# ifdef va_copy
2375# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2376# else
a1866d1b
JH
2377# if defined(__va_copy)
2378# define Perl_va_copy(s, d) __va_copy(d, s)
2379# else
2380# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2381# endif
2cc61e15
DD
2382# endif
2383#endif
2384
472d47bc
SB
2385/* convenience debug macros */
2386#ifdef USE_ITHREADS
2387#define pTHX_FORMAT "Perl interpreter: 0x%p"
2388#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2389#define pTHX_VALUE_ (void *)my_perl,
2390#define pTHX_VALUE (void *)my_perl
2391#define pTHX__VALUE_ ,(void *)my_perl,
2392#define pTHX__VALUE ,(void *)my_perl
472d47bc 2393#else
3609ea0d 2394#define pTHX_FORMAT
472d47bc 2395#define pTHX__FORMAT
3609ea0d 2396#define pTHX_VALUE_
472d47bc 2397#define pTHX_VALUE
3609ea0d 2398#define pTHX__VALUE_
472d47bc
SB
2399#define pTHX__VALUE
2400#endif /* USE_ITHREADS */
3609ea0d 2401
2acdbac1
NC
2402/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2403 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2404 CPAN::Unpack show any users outside the core. */
2405#ifdef PERL_CORE
b7d90381
KW
2406# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2407 "Use of " s " is deprecated")
2acdbac1
NC
2408#endif
2409
dfff4baf
BF
2410/* Internal macros to deal with gids and uids */
2411#ifdef PERL_CORE
2412
2413# if Uid_t_size > IVSIZE
2414# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2415# define SvUID(sv) SvNV(sv)
2416# else
2417# if Uid_t_sign <= 0
2418# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2419# define SvUID(sv) SvIV(sv)
2420# else
2421# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2422# define SvUID(sv) SvUV(sv)
2423# endif
2424# endif /* Uid_t_size */
2425
2426# if Gid_t_size > IVSIZE
2427# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2428# define SvGID(sv) SvNV(sv)
2429# else
2430# if Gid_t_sign <= 0
2431# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2432# define SvGID(sv) SvIV(sv)
2433# else
2434# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2435# define SvGID(sv) SvUV(sv)
2436# endif
2437# endif /* Gid_t_size */
2438
2439#endif
2440
9d745869
KW
2441#endif /* HANDY_H */
2442
e9a8c099 2443/*
14d04a33 2444 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2445 */