This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Update Test-Simple to CPAN version 1.302073
[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
d0da05db 568Variant C<isFOO_utf8> is like C<isFOO_uvchr>, but the input is a pointer to a
dc6b0978
KW
569(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
570possibly containing embedded C<NUL> characters). The classification of just
571the first (possibly multi-byte) character in the string is tested.
243effed 572
31f05a37
KW
573Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
574result is based on the current locale, which is what C<LC> in the name stands
575for. If Perl can determine that the current locale is a UTF-8 locale, it uses
576the published Unicode rules; otherwise, it uses the C library function that
577gives the named classification. For example, C<isDIGIT_LC()> when not in a
578UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
579returned if the input won't fit into an octet. On some platforms where the C
580library function is known to be defective, Perl changes its result to follow
581the POSIX standard's rules.
243effed
KW
582
583Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
584returns the same as C<isFOO_LC> for input code points less than 256, and
585returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
586
b7d90381 587Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer
dc6b0978
KW
588to a (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
589possibly containing embedded C<NUL> characters). The classification of just
590the first (possibly multi-byte) character in the string is tested.
ccfc67b7 591
954c1994 592=for apidoc Am|bool|isALPHA|char ch
2787a470 593Returns a boolean indicating whether the specified character is an
6aff1f14 594alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
595See the L<top of this section|/Character classification> for an explanation of
596variants
d0da05db 597C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>, C<isALPHA_LC>,
243effed 598C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
8a58bdcf 599
15861f94
KW
600=for apidoc Am|bool|isALPHANUMERIC|char ch
601Returns a boolean indicating whether the specified character is a either an
6aff1f14 602alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
603See the L<top of this section|/Character classification> for an explanation of
604variants
d0da05db 605C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
e5ad6aba 606C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
15861f94
KW
607C<isALPHANUMERIC_LC_utf8>.
608
8a58bdcf
KW
609=for apidoc Am|bool|isASCII|char ch
610Returns a boolean indicating whether the specified character is one of the 128
243effed 611characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 612On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
613character corresponds to an ASCII character. Variants C<isASCII_A()> and
614C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
615See the L<top of this section|/Character classification> for an explanation of
616variants
d0da05db 617C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
243effed
KW
618C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
619library routine C<isascii()>. In these cases, the variants whose names contain
620C<LC> are the same as the corresponding ones without.
621
d98532ea
KW
622Also note, that because all ASCII characters are UTF-8 invariant (meaning they
623have the exact same representation (always a single byte) whether encoded in
624UTF-8 or not), C<isASCII> will give the correct results when called with any
625byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
626work properly on any string encoded or not in UTF-8.
627
243effed
KW
628=for apidoc Am|bool|isBLANK|char ch
629Returns a boolean indicating whether the specified character is a
6aff1f14 630character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
631See the L<top of this section|/Character classification> for an explanation of
632variants
d0da05db 633C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>, C<isBLANK_LC>,
243effed
KW
634C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
635platforms do not have the C library routine C<isblank()>. In these cases, the
636variants whose names contain C<LC> are the same as the corresponding ones
637without.
638
639=for apidoc Am|bool|isCNTRL|char ch
640Returns a boolean indicating whether the specified character is a
6aff1f14 641control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
642See the L<top of this section|/Character classification> for an explanation of
643variants
d0da05db 644C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
6aff1f14
KW
645C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
646On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 647
954c1994 648=for apidoc Am|bool|isDIGIT|char ch
2787a470 649Returns a boolean indicating whether the specified character is a
6aff1f14 650digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 651Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
652See the L<top of this section|/Character classification> for an explanation of
653variants
d0da05db 654C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
243effed
KW
655C<isDIGIT_LC_utf8>.
656
657=for apidoc Am|bool|isGRAPH|char ch
658Returns a boolean indicating whether the specified character is a
6aff1f14 659graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff
KW
660See the L<top of this section|/Character classification> for an explanation of
661variants
d0da05db 662C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
243effed 663C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
954c1994 664
0c82b6df 665=for apidoc Am|bool|isLOWER|char ch
2787a470 666Returns a boolean indicating whether the specified character is a
6aff1f14 667lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
668See the L<top of this section|/Character classification> for an explanation of
669variants
d0da05db 670C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>, C<isLOWER_LC>,
243effed 671C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
0c82b6df 672
c99e91e9 673=for apidoc Am|bool|isOCTAL|char ch
2787a470 674Returns a boolean indicating whether the specified character is an
6aff1f14 675octal digit, [0-7].
243effed
KW
676The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
677C<isOCTAL>.
678
679=for apidoc Am|bool|isPUNCT|char ch
680Returns a boolean indicating whether the specified character is a
6aff1f14
KW
681punctuation character, analogous to C<m/[[:punct:]]/>.
682Note that the definition of what is punctuation isn't as
243effed
KW
683straightforward as one might desire. See L<perlrecharclass/POSIX Character
684Classes> for details.
dcccc8ff
KW
685See the L<top of this section|/Character classification> for an explanation of
686variants
d0da05db 687C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
243effed 688C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
c99e91e9 689
0c82b6df 690=for apidoc Am|bool|isSPACE|char ch
2787a470 691Returns a boolean indicating whether the specified character is a
6aff1f14 692whitespace character. This is analogous
398d098a 693to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 694this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
695locale forms of this macro (the ones with C<LC> in their names) matched
696precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
697in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
698(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
699See the L<top of this section|/Character classification> for an explanation of
700variants
d0da05db 701C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>, C<isSPACE_LC>,
243effed 702C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
0c82b6df 703
398d098a
KW
704=for apidoc Am|bool|isPSXSPC|char ch
705(short for Posix Space)
779cf272
KW
706Starting in 5.18, this is identical in all its forms to the
707corresponding C<isSPACE()> macros.
398d098a
KW
708The locale forms of this macro are identical to their corresponding
709C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
710non-locale forms differ from their C<isSPACE()> forms only in that the
711C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
712Otherwise they are identical. Thus this macro is analogous to what
713C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 714See the L<top of this section|/Character classification> for an explanation of
d0da05db 715variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
b7d90381 716C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
398d098a 717
954c1994 718=for apidoc Am|bool|isUPPER|char ch
2787a470 719Returns a boolean indicating whether the specified character is an
6aff1f14 720uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff
KW
721See the L<top of this section|/Character classification> for an explanation of
722variants
d0da05db 723C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>, C<isUPPER_LC>,
243effed 724C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
954c1994 725
243effed 726=for apidoc Am|bool|isPRINT|char ch
8eea39dd 727Returns a boolean indicating whether the specified character is a
6aff1f14 728printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
729See the L<top of this section|/Character classification> for an explanation of
730variants
d0da05db 731C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>, C<isPRINT_LC>,
243effed
KW
732C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
733
734=for apidoc Am|bool|isWORDCHAR|char ch
735Returns a boolean indicating whether the specified character is a character
736that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
737in a regular expression. A word character is an alphabetic character, a
738decimal digit, a connecting punctuation character (such as an underscore), or
739a "mark" character that attaches to one of those (like some sort of accent).
740C<isALNUM()> is a synonym provided for backward compatibility, even though a
741word character includes more than the standard C language meaning of
742alphanumeric.
dcccc8ff
KW
743See the L<top of this section|/Character classification> for an explanation of
744variants
d0da05db 745C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and C<isWORDCHAR_utf8>.
1a83413c
KW
746C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
747described there, but additionally include the platform's native underscore.
8a58bdcf
KW
748
749=for apidoc Am|bool|isXDIGIT|char ch
750Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
751digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
752and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
753See the L<top of this section|/Character classification> for an explanation of
754variants
d0da05db 755C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
243effed
KW
756C<isXDIGIT_LC_utf8>.
757
3c3ecf18
KW
758=for apidoc Am|bool|isIDFIRST|char ch
759Returns a boolean indicating whether the specified character can be the first
760character of an identifier. This is very close to, but not quite the same as
761the official Unicode property C<XID_Start>. The difference is that this
762returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
763See the L<top of this section|/Character classification> for an explanation of
764variants
d0da05db 765C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
3c3ecf18
KW
766C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
767
768=for apidoc Am|bool|isIDCONT|char ch
769Returns a boolean indicating whether the specified character can be the
770second or succeeding character of an identifier. This is very close to, but
771not quite the same as the official Unicode property C<XID_Continue>. The
772difference is that this returns true only if the input character also matches
dcccc8ff
KW
773L</isWORDCHAR>. See the L<top of this section|/Character classification> for
774an
d0da05db 775explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
3c3ecf18
KW
776C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
777C<isIDCONT_LC_utf8>.
778
243effed 779=head1 Miscellaneous Functions
8eea39dd 780
95a59cab 781=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 782Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
783Behaviour is only well defined when isXDIGIT(*str) is true.
784
e7c1e6c1
KW
785=head1 Character case changing
786
1f607577
KW
787=for apidoc Am|U8|toUPPER|U8 ch
788Converts the specified character to uppercase. If the input is anything but an
789ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 790C<toUPPER_A> is equivalent.
954c1994 791
d0da05db
KW
792=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
793Converts the code point C<cp> to its uppercase version, and
794stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
795point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
796that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
797bytes since the uppercase version may be longer than the original character.
798
799The first code point of the uppercased version is returned
800(but note, as explained just above, that there may be more.)
801
802=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
803Converts the UTF-8 encoded character at C<p> to its uppercase version, and
804stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
805that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
806bytes since the uppercase version may be longer than the original character.
807
808The first code point of the uppercased version is returned
809(but note, as explained just above, that there may be more.)
810
811The input character at C<p> is assumed to be well-formed.
812
25200305
KW
813=for apidoc Am|U8|toFOLD|U8 ch
814Converts the specified character to foldcase. If the input is anything but an
815ASCII uppercase character, that input character itself is returned. Variant
816C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 817Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 818
d0da05db
KW
819=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
820Converts the code point C<cp> to its foldcase version, and
821stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
822point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
823that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
824bytes since the foldcase version may be longer than the original character.
825
826The first code point of the foldcased version is returned
827(but note, as explained just above, that there may be more.)
828
829=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
830Converts the UTF-8 encoded character at C<p> to its foldcase version, and
831stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
832that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
833bytes since the foldcase version may be longer than the original character.
834
835The first code point of the foldcased version is returned
836(but note, as explained just above, that there may be more.)
837
838The input character at C<p> is assumed to be well-formed.
839
840=for apidoc Am|U8|toLOWER|U8 ch
841Converts the specified character to lowercase. If the input is anything but an
842ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 843C<toLOWER_A> is equivalent.
954c1994 844
1f607577 845=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
846Converts the specified Latin1 character to lowercase. The results are
847undefined if the input doesn't fit in a byte.
1f607577
KW
848
849=for apidoc Am|U8|toLOWER_LC|U8 ch
850Converts the specified character to lowercase using the current locale's rules,
851if possible; otherwise returns the input character itself.
852
d0da05db
KW
853=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
854Converts the code point C<cp> to its lowercase version, and
855stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
856point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
857that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
858bytes since the lowercase version may be longer than the original character.
859
860The first code point of the lowercased version is returned
861(but note, as explained just above, that there may be more.)
862
863=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
864Converts the UTF-8 encoded character at C<p> to its lowercase version, and
865stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
866that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
867bytes since the lowercase version may be longer than the original character.
868
869The first code point of the lowercased version is returned
870(but note, as explained just above, that there may be more.)
871
872The input character at C<p> is assumed to be well-formed.
873
25200305
KW
874=for apidoc Am|U8|toTITLE|U8 ch
875Converts the specified character to titlecase. If the input is anything but an
876ASCII lowercase character, that input character itself is returned. Variant
b7d90381 877C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 878range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 879not a concept used in locale handling, so there is no functionality for that.)
25200305 880
d0da05db
KW
881=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
882Converts the code point C<cp> to its titlecase version, and
883stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
884point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
885that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
886bytes since the titlecase version may be longer than the original character.
887
888The first code point of the titlecased version is returned
889(but note, as explained just above, that there may be more.)
890
891=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
892Converts the UTF-8 encoded character at C<p> to its titlecase version, and
893stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
894that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
895bytes since the titlecase version may be longer than the original character.
896
897The first code point of the titlecased version is returned
898(but note, as explained just above, that there may be more.)
899
900The input character at C<p> is assumed to be well-formed.
901
954c1994 902=cut
353c9b6f 903
d0da05db 904XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
25200305 905really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 906
8a58bdcf 907Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
908patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
909
954c1994
GS
910*/
911
7c062697
KW
912/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
913 * is known only in the perl core, and this macro can be called from outside
914 * that */
915#ifdef HAS_QUAD
916# define WIDEST_UTYPE U64TYPE
917#else
918# define WIDEST_UTYPE U32
919#endif
920
3912bc88
KW
921/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
922 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
923 * bits of information are lost even on a 64-bit machine, but to get the
924 * compiler to optimize it out if possible. This is because Configure makes
925 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
926 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
927 *
928 * For Coverity, be always true, because otherwise Coverity thinks
929 * it finds several expressions that are always true, independent
930 * of operands. Well, they are, but that is kind of the point.
220c71bf 931 */
7e75d1a1 932#ifndef __COVERITY__
6c5b02ac
KW
933 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
934 * pointer) */
935#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 936 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
937#else
938#define FITS_IN_8_BITS(c) (1)
939#endif
cf301eb7 940
41f43cc2 941#ifdef EBCDIC
b6340bd0 942# ifndef _ALL_SOURCE
0852beac
KW
943 /* The native libc isascii() et.al. functions return the wrong results
944 * on at least z/OS unless this is defined. */
b6340bd0
KW
945# error _ALL_SOURCE should probably be defined
946# endif
41f43cc2 947#else
0852beac
KW
948 /* There is a simple definition of ASCII for ASCII platforms. But the
949 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 950 * other macros below.
3f3c579d
KW
951 *
952 * The cast here is used instead of '(c) >= 0', because some compilers emit
953 * a warning that that test is always true when the parameter is an
954 * unsigned type. khw supposes that it could be written as
955 * && ((c) == '\0' || (c) > 0)
956 * to avoid the message, but the cast will likely avoid extra branches even
957 * with stupid compilers.
958 *
959 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
960 * a pointer) */
9c903d59 961# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
962#endif
963
38694112
KW
964/* Take the eight possible bit patterns of the lower 3 bits and you get the
965 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
966 * can be ignored. If the rest match '0', we have an octal */
967#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 968
9fb1bf9d 969#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 970
a500dc72
KW
971/* Character class numbers. For internal core Perl use only. The ones less
972 * than 32 are used in PL_charclass[] and the ones up through the one that
973 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
974 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
975 * their actual definitions are here. If that file has a name not used here,
976 * it won't compile.
1709d539
KW
977 *
978 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
979 * frequency of their use. This gives a slight edge to exiting a loop earlier
980 * (in reginclass() in regexec.c) */
1709d539
KW
981# define _CC_WORDCHAR 0 /* \w and [:word:] */
982# define _CC_DIGIT 1 /* \d and [:digit:] */
983# define _CC_ALPHA 2 /* [:alpha:] */
984# define _CC_LOWER 3 /* [:lower:] */
985# define _CC_UPPER 4 /* [:upper:] */
986# define _CC_PUNCT 5 /* [:punct:] */
987# define _CC_PRINT 6 /* [:print:] */
15861f94 988# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 989# define _CC_GRAPH 8 /* [:graph:] */
359b005e 990# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1709d539 991
b0d691b2 992#define _FIRST_NON_SWASH_CC 10
1709d539
KW
993/* The character classes above are implemented with swashes. The second group
994 * (just below) contains the ones implemented without. These are also sorted
995 * in rough order of the frequency of their use, except that \v should be last,
996 * as it isn't a real Posix character class, and some (small) inefficiencies in
997 * regular expression handling would be introduced by putting it in the middle
998 * of those that are. Also, cntrl and ascii come after the others as it may be
999 * useful to group these which have no members that match above Latin1, (or
1000 * above ASCII in the latter case) */
1001
779cf272 1002# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1003# define _CC_BLANK 11 /* [:blank:] */
1004# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1005# define _CC_CNTRL 13 /* [:cntrl:] */
1006# define _CC_ASCII 14 /* [:ascii:] */
1007# define _CC_VERTSPACE 15 /* \v */
1709d539 1008
a0947d7b
KW
1009# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1010
1709d539 1011/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1012 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1013# define _CC_IDFIRST 16
1014# define _CC_CHARNAME_CONT 17
1015# define _CC_NONLATIN1_FOLD 18
1016# define _CC_NONLATIN1_SIMPLE_FOLD 19
1017# define _CC_QUOTEMETA 20
1018# define _CC_NON_FINAL_FOLD 21
1019# define _CC_IS_IN_SOME_FOLD 22
1020# define _CC_MNEMONIC_CNTRL 23
073c22b3
KW
1021
1022/* This next group is only used on EBCDIC platforms, so theoretically could be
1023 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
1024# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1025# define _CC_UTF8_IS_START 29
1026# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1027# define _CC_UTF8_IS_CONTINUATION 31
1028/* Unused: 24-27
f4cdb42c
KW
1029 * If more bits are needed, one could add a second word for non-64bit
1030 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1031 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1032 * is used only for optimization (as of this writing), and differs in the
1033 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1034 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1035 * would just cause /i regexes which match them to run less efficiently.
1036 * Similarly the EBCDIC-only bits are used just for speed, and could be
1037 * replaced by other means */
96ac0975 1038
3a371f2f
KW
1039#if defined(PERL_CORE) || defined(PERL_EXT)
1040/* An enum version of the character class numbers, to help compilers
1041 * optimize */
1042typedef enum {
3a371f2f 1043 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1044 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1045 _CC_ENUM_ASCII = _CC_ASCII,
1046 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1047 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1048 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1049 _CC_ENUM_DIGIT = _CC_DIGIT,
1050 _CC_ENUM_GRAPH = _CC_GRAPH,
1051 _CC_ENUM_LOWER = _CC_LOWER,
1052 _CC_ENUM_PRINT = _CC_PRINT,
1053 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1054 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1055 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1056 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1057 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1058 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1059} _char_class_number;
1060#endif
1061
63c61c3f 1062#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1063#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1064
b7d90381
KW
1065#if defined(PERL_IN_UTF8_C) \
1066 || defined(PERL_IN_REGCOMP_C) \
1067 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1068# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1069 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1070 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1071 #error Need to adjust order of swash_property_names[]
1072# endif
1073
b44daeeb
KW
1074/* This is declared static in each of the few files that this is #defined for
1075 * to keep them from being publicly accessible. Hence there is a small amount
1076 * of wasted space */
1077
922e8cb4
KW
1078static const char* const swash_property_names[] = {
1079 "XPosixWord",
1080 "XPosixDigit",
1081 "XPosixAlpha",
1082 "XPosixLower",
1083 "XPosixUpper",
1084 "XPosixPunct",
1085 "XPosixPrint",
1086 "XPosixAlnum",
b0d691b2
KW
1087 "XPosixGraph",
1088 "Cased"
922e8cb4
KW
1089};
1090#endif
1091
6635f04f 1092START_EXTERN_C
96ac0975
NC
1093# ifdef DOINIT
1094EXTCONST U32 PL_charclass[] = {
1095# include "l1_char_class_tab.h"
1096};
1097
1098# else /* ! DOINIT */
1099EXTCONST U32 PL_charclass[];
1100# endif
6635f04f 1101END_EXTERN_C
96ac0975 1102
265c1f46 1103 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1104# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1105
1106 /* For internal core Perl use only: the base macro for defining macros like
1107 * isALPHA */
ff7ecfc3 1108# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1109 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1110
f4cdb42c
KW
1111 /* The mask for the _A versions of the macros; it just adds in the bit for
1112 * ASCII. */
1113# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1114
4650c663
KW
1115 /* For internal core Perl use only: the base macro for defining macros like
1116 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1117 * the ASCII bit are present */
b7d90381
KW
1118# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1119 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1120 == _CC_mask_A(classnum)))
f4cdb42c 1121
f4cdb42c 1122# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1123# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1124# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1125# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1126# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1127# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1128# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1129# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1130# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1131# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1132# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1133# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1134# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1135 */
d95f8b6a 1136# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1137# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1138# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1139# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1140
1141 /* continuation character for legal NAME in \N{NAME} */
1142# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1143
1144# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1145# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1146# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1147# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1148# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1149# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1150# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1151# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1152# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1153# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1154
0852beac
KW
1155# ifdef EBCDIC
1156# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1157# endif
1158
f12c0118
KW
1159 /* Participates in a single-character fold with a character above 255 */
1160# 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)))
1161
1162 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1163# 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 1164
4eeeb416 1165# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1166# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1167 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1168# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1169 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1170# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1171 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1172#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1173
1174 /* If we don't have perl.h, we are compiling a utility program. Below we
1175 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1176 * to it. Most are coded based on first principles. These are written to
74665a89 1177 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
0852beac
KW
1178# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1179# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1180# define isSPACE_A(c) (isBLANK_A(c) \
1181 || (c) == '\n' \
1182 || (c) == '\r' \
1183 || (c) == '\v' \
0852beac 1184 || (c) == '\f')
74665a89
KW
1185 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1186 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1187 * performance (if optimization isn't on), and allow the same code to be
1188 * used for both platform types */
1189# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1190 && ( (c) <= 'i' \
1191 || ((c) >= 'j' && (c) <= 'r') \
1192 || (c) >= 's'))
1193# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1194 && ( (c) <= 'I' \
1195 || ((c) >= 'J' && (c) <= 'R') \
1196 || (c) >= 'S'))
a4d7a999
KW
1197# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1198# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1199# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1200# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
74665a89
KW
1201# define isXDIGIT_A(c) (isDIGIT_A(c) \
1202 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1203 || ((c) <= 'F' && (c) >= 'A'))
74665a89
KW
1204# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1205 || (c) == '#' || (c) == '$' || (c) == '%' \
1206 || (c) == '&' || (c) == '\'' || (c) == '(' \
1207 || (c) == ')' || (c) == '*' || (c) == '+' \
1208 || (c) == ',' || (c) == '.' || (c) == '/' \
1209 || (c) == ':' || (c) == ';' || (c) == '<' \
1210 || (c) == '=' || (c) == '>' || (c) == '?' \
1211 || (c) == '@' || (c) == '[' || (c) == '\\' \
1212 || (c) == ']' || (c) == '^' || (c) == '_' \
1213 || (c) == '`' || (c) == '{' || (c) == '|' \
1214 || (c) == '}' || (c) == '~')
1215# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1216# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1217
0852beac 1218# ifdef EBCDIC
74665a89
KW
1219 /* The below is accurate for the 3 EBCDIC code pages traditionally
1220 * supported by perl. The only difference between them in the controls
1221 * is the position of \n, and that is represented symbolically below */
1222# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1223 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1224 || (c) == '\t' || (c) == '\v' \
1225 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1226 || (c) == 7 /* U+7F DEL */ \
1227 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1228 /* DLE, DC[1-3] */ \
1229 || (c) == 0x18 /* U+18 CAN */ \
1230 || (c) == 0x19 /* U+19 EOM */ \
1231 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1232 || (c) == 0x26 /* U+17 ETB */ \
1233 || (c) == 0x27 /* U+1B ESC */ \
1234 || (c) == 0x2D /* U+05 ENQ */ \
1235 || (c) == 0x2E /* U+06 ACK */ \
1236 || (c) == 0x32 /* U+16 SYN */ \
1237 || (c) == 0x37 /* U+04 EOT */ \
1238 || (c) == 0x3C /* U+14 DC4 */ \
1239 || (c) == 0x3D /* U+15 NAK */ \
1240 || (c) == 0x3F)/* U+1A SUB */
0852beac 1241# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1242# else /* isASCII is already defined for ASCII platforms, so can use that to
1243 define isCNTRL */
1244# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1245# endif
1246
3ffc8c70 1247 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1248 * during debugging, and it seems best to keep them. We may be called
1249 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1250 * do anything anyway, so make it not a problem */
1251# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1252# define NATIVE_TO_LATIN1(ch) (ch)
1253# endif
3ded5eb0
KW
1254# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1255# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1256# define isBLANK_L1(c) (isBLANK_A(c) \
1257 || (FITS_IN_8_BITS(c) \
1258 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1259# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1260# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1261# define isLOWER_L1(c) (isLOWER_A(c) \
1262 || (FITS_IN_8_BITS(c) \
1263 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1264 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1265 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1266 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1267 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1268# define isPRINT_L1(c) (isPRINT_A(c) \
1269 || (FITS_IN_8_BITS(c) \
1270 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1271# define isPUNCT_L1(c) (isPUNCT_A(c) \
1272 || (FITS_IN_8_BITS(c) \
1273 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1274 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1275 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1276 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1277 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1278 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1279 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1280# define isSPACE_L1(c) (isSPACE_A(c) \
1281 || (FITS_IN_8_BITS(c) \
1282 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1283 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1284# define isUPPER_L1(c) (isUPPER_A(c) \
1285 || (FITS_IN_8_BITS(c) \
1286 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1287 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1288 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1289# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1290# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1291# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1292 || isBLANK_L1(c) \
1293 || (c) == '-' \
1294 || (c) == '(' \
1295 || (c) == ')')
1296 /* The following are not fully accurate in the above-ASCII range. I (khw)
1297 * don't think it's necessary to be so for the purposes where this gets
1298 * compiled */
1299# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1300# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1301
1302 /* And these aren't accurate at all. They are useful only for above
1303 * Latin1, which utilities and bootstrapping don't deal with */
1304# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1305# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1306# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1307
1308 /* Many of the macros later in this file are defined in terms of these. By
1309 * implementing them with a function, which converts the class number into
1310 * a call to the desired macro, all of the later ones work. However, that
1311 * function won't be actually defined when building a utility program (no
1312 * perl.h), and so a compiler error will be generated if one is attempted
1313 * to be used. And the above-Latin1 code points require Unicode tables to
1314 * be present, something unlikely to be the case when bootstrapping */
1315# define _generic_isCC(c, classnum) \
1316 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1317# define _generic_isCC_A(c, classnum) \
1318 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1319#endif /* End of no perl.h H_PERL */
8a58bdcf 1320
e66b99e9
KW
1321#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1322#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1323#define isASCII_A(c) isASCII(c)
1324#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1325#define isBLANK(c) isBLANK_A(c)
1326#define isCNTRL(c) isCNTRL_A(c)
1327#define isDIGIT(c) isDIGIT_A(c)
1328#define isGRAPH(c) isGRAPH_A(c)
1329#define isIDFIRST(c) isIDFIRST_A(c)
1330#define isLOWER(c) isLOWER_A(c)
1331#define isPRINT(c) isPRINT_A(c)
779cf272 1332#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1333#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1334#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1335#define isPUNCT(c) isPUNCT_A(c)
1336#define isSPACE(c) isSPACE_A(c)
1337#define isUPPER(c) isUPPER_A(c)
1338#define isWORDCHAR(c) isWORDCHAR_A(c)
1339#define isXDIGIT(c) isXDIGIT_A(c)
1340
1341/* ASCII casing. These could also be written as
1342 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1343 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1344 which uses table lookup and mask instead of subtraction. (This would
1345 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1346#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1347#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1348
25200305
KW
1349/* In the ASCII range, these are equivalent to what they're here defined to be.
1350 * But by creating these definitions, other code doesn't have to be aware of
1351 * this detail */
1352#define toFOLD(c) toLOWER(c)
25200305
KW
1353#define toTITLE(c) toUPPER(c)
1354
c753c8d3
KW
1355#define toLOWER_A(c) toLOWER(c)
1356#define toUPPER_A(c) toUPPER(c)
25200305
KW
1357#define toFOLD_A(c) toFOLD(c)
1358#define toTITLE_A(c) toTITLE(c)
1a0901db 1359
4650c663 1360/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1361#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1362 ? (c) \
f4cd282c 1363 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1364#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1365
1a0901db 1366/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1367 * all mapped to one of them, and these need special handling; returns the
1368 * input itself if is out-of-range */
b2bf251f 1369#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1370 ? (c) \
f4cd282c 1371 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1372#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1373
beab9ebe
KW
1374/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1375
1376/* For internal core Perl use only: the base macro for defining macros like
1377 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1378 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1379 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1380 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1381 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1382 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1383 * code points under locale rules, use the macros based on _generic_LC_uvchr
1384 * instead of this. */
beab9ebe
KW
1385#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1386 (! FITS_IN_8_BITS(c) \
1387 ? 0 \
31f05a37
KW
1388 : IN_UTF8_CTYPE_LOCALE \
1389 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1390 : cBOOL(non_utf8))
1391
1392/* For internal core Perl use only: a helper macro for defining macros like
1393 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1394 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1395 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1396 * _generic_LC_base for more info */
1397#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1398 _generic_LC_base(c,utf8_locale_classnum, \
1399 non_utf8_func( (_LC_CAST) (c)))
1400
1401/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1402 * 'c' is the platform's native underscore character */
1403#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1404 _generic_LC_base(c, utf8_locale_classnum, \
1405 (non_utf8_func( (_LC_CAST) (c)) \
1406 || (char)(c) == '_'))
1407
1408/* These next three are also for internal core Perl use only: case-change
1409 * helper macros */
1410#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1411 ? (c) \
31f05a37
KW
1412 : (IN_UTF8_CTYPE_LOCALE) \
1413 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1414 : (cast)function((cast)(c)))
beab9ebe 1415
31f05a37
KW
1416/* Note that the result can be larger than a byte in a UTF-8 locale. It
1417 * returns a single value, so can't adequately return the upper case of LATIN
1418 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1419 * values "SS"); instead it asserts against that under DEBUGGING, and
1420 * otherwise returns its input */
beab9ebe
KW
1421#define _generic_toUPPER_LC(c, function, cast) \
1422 (! FITS_IN_8_BITS(c) \
1423 ? (c) \
31f05a37 1424 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1425 ? (cast)function((cast)(c)) \
31f05a37
KW
1426 : ((((U8)(c)) == MICRO_SIGN) \
1427 ? GREEK_CAPITAL_LETTER_MU \
1428 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1429 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1430 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1431 ? (__ASSERT_(0) (c)) \
1432 : PL_mod_latin1_uc[ (U8) (c) ])))))
1433
1434/* Note that the result can be larger than a byte in a UTF-8 locale. It
1435 * returns a single value, so can't adequately return the fold case of LATIN
1436 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1437 * values "ss"); instead it asserts against that under DEBUGGING, and
1438 * otherwise returns its input */
beab9ebe 1439#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1440 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1441 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1442 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1443 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1444 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1445
84061b6a 1446/* Use the libc versions for these if available. */
f05550c0 1447#if defined(HAS_ISASCII)
84061b6a
KW
1448# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1449#else
1450# define isASCII_LC(c) isASCII(c)
1451#endif
1452
f05550c0 1453#if defined(HAS_ISBLANK)
84061b6a 1454# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1455#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1456# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1457#endif
1458
f05550c0 1459#define _LC_CAST U8
bbce6d69 1460
f05550c0 1461#ifdef WIN32
375f5f06
KW
1462 /* The Windows functions don't bother to follow the POSIX standard, which
1463 * for example says that something can't both be a printable and a control.
1464 * But Windows treats the \t control as a printable, and does such things
1465 * as making superscripts into both digits and punctuation. This tames
1466 * these flaws by assuming that the definitions of both controls and space
1467 * are correct, and then making sure that other definitions don't have
1468 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1469 * Not all possible weirdnesses are checked for, just the ones that were
1470 * detected on actual Microsoft code pages */
1471
b7d90381
KW
1472# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1473# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1474
1475# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1476 && isALPHANUMERIC_LC(c))
1477# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1478 ! isPUNCT_LC(c))
1479# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1480 isALPHANUMERIC_LC(c))
1481# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1482# define isIDFIRST_LC(c) (((c) == '_') \
1483 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1484# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1485# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1486# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1487# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1488# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1489# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1490 && isALPHANUMERIC_LC(c))
f05550c0
BF
1491
1492# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1493# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1494# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1495
1496#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1497 /* For most other platforms */
beab9ebe 1498
f05550c0
BF
1499# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1500# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1501# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1502# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1503# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1504# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1505# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1506# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1507# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1508# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1509# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1510# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1511# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1512
1513
1514# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1515# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1516# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1517
1518#else /* The final fallback position */
1519
b7d90381
KW
1520# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1521# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1522# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1523# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1524# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1525# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1526# define isLOWER_LC(c) (isascii(c) && islower(c))
1527# define isPRINT_LC(c) (isascii(c) && isprint(c))
1528# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1529# define isSPACE_LC(c) (isascii(c) && isspace(c))
1530# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1531# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1532# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1533
1534# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1535# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1536# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1537
f05550c0 1538#endif
55204971 1539
eba68aa0
KW
1540#define isIDCONT(c) isWORDCHAR(c)
1541#define isIDCONT_A(c) isWORDCHAR_A(c)
1542#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1543#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1544#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1545
4650c663 1546/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1547 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1548 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1549 * classes where there is a macro or function 'above_latin1' that takes the
1550 * single argument 'c' and returns the desired value. These exist for those
1551 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1552 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1553 * for classes where that overhead is faster than a direct lookup.
d0da05db 1554 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1555 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1556 * have duplicate checks here, so could create versions of the macros that
1557 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1558
1559/* Note that all ignore 'use bytes' */
d0da05db 1560#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
cd500f2f
KW
1561 ? _generic_isCC(c, classnum) \
1562 : above_latin1(c))
d0da05db 1563#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
cd500f2f 1564 ? _generic_isCC(c, classnum) \
922e8cb4 1565 : _is_uni_FOO(classnum, c))
d0da05db
KW
1566#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1567#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1568#define isASCII_uvchr(c) isASCII(c)
1569#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1570#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1571#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1572#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1573#define isIDCONT_uvchr(c) _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1574#define isIDFIRST_uvchr(c) _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
1575#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1576#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1577
1578#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1579#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1580#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1581
1582#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1583#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1584#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1585#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1586
1587#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1588#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1589#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1590#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1591
1592/* For backwards compatibility, even though '_uni' should mean official Unicode
1593 * code points, in Perl it means native for those below 256 */
1594#define isALPHA_uni(c) isALPHA_uvchr(c)
1595#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1596#define isASCII_uni(c) isASCII_uvchr(c)
1597#define isBLANK_uni(c) isBLANK_uvchr(c)
1598#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1599#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1600#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1601#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1602#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1603#define isLOWER_uni(c) isLOWER_uvchr(c)
1604#define isPRINT_uni(c) isPRINT_uvchr(c)
1605#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1606#define isSPACE_uni(c) isSPACE_uvchr(c)
1607#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1608#define isUPPER_uni(c) isUPPER_uvchr(c)
1609#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1610#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1611#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1612#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1613#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1614#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1615#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1616
4650c663
KW
1617/* For internal core Perl use only: the base macros for defining macros like
1618 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1619 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1620 * simple class definitions; the other for more complex. These are like
d0da05db 1621 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1622#define _generic_LC_uvchr(latin1, above_latin1, c) \
1623 (c < 256 ? latin1(c) : above_latin1(c))
1624#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1625 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1626
1627#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1628#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1629 _CC_ALPHANUMERIC, c)
b7d90381
KW
1630#define isASCII_LC_uvchr(c) isASCII_LC(c)
1631#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1632 is_HORIZWS_cp_high, c)
feeab5a9 1633#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1634#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1635#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1636#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1637 _is_uni_perl_idcont, c)
b7d90381 1638#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1639 _is_uni_perl_idstart, c)
1640#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1641#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1642#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1643#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1644#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1645 is_XPERLSPACE_cp_high, c)
cd500f2f 1646#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1647#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1648 _CC_WORDCHAR, c)
b7d90381
KW
1649#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1650 is_XDIGIT_cp_high, c)
e712593e 1651
b7d90381 1652#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1653
4650c663
KW
1654/* For internal core Perl use only: the base macros for defining macros like
1655 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1656 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1657 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1658 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1659 * same representation whether utf8 or not. Note that it assumes that the utf8
1660 * has been validated, and ignores 'use bytes' */
cd500f2f 1661#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1662 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1663 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1664 ? _generic_isCC( \
a62b247b 1665 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
aea0c08c 1666 *((p)+1 )), \
94bb8c36 1667 classnum) \
252810d0 1668 : utf8)
b7d90381
KW
1669/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1670 * 'above_latin1' can be a macro */
cd500f2f
KW
1671#define _generic_func_utf8(classnum, above_latin1, p) \
1672 _generic_utf8(classnum, p, above_latin1(p))
4650c663 1673/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1674 * 'above_latin1' parameter */
922e8cb4 1675#define _generic_swash_utf8(classnum, p) \
cd500f2f 1676 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
922e8cb4 1677
cc8ab7c0 1678/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1679 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1680 * class is TRUE for. Hence it can skip the tests for this range.
1681 * 'above_latin1' should include its arguments */
3ded5eb0 1682#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1683 (UTF8_IS_INVARIANT(*(p)) \
1684 ? _generic_isCC(*(p), classnum) \
1685 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1686 ? above_latin1 \
cc8ab7c0
KW
1687 : 0)
1688
84238efa
KW
1689/* NOTE that some of these macros have very similar ones in regcharclass.h.
1690 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1691 * there, differing in name only by an underscore from the one here
1692 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1693 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1694 * points; the regcharclass.h ones are implemented as a series of
1695 * "if-else-if-else ..." */
1696
b7d90381
KW
1697#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1698#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1699#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
509fb054
KW
1700 utf8, the non-utf8 macro works
1701 */
b7d90381 1702#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2
KW
1703
1704#ifdef EBCDIC
1705 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1706 * more efficient macro instead of the more general one */
88794300 1707# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
e8fa43e2
KW
1708#else
1709# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1710#endif
1711
cd500f2f
KW
1712#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1713 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1714#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1715#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1716 _is_utf8_perl_idcont, p)
e5dcd934 1717
c11ff943
KW
1718/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1719 * IDFIRST is an alnum. See
c8362b00 1720 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1721 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1722 * This used to be not the XID version, but we decided to go with the more
1723 * modern Unicode definition */
b7d90381 1724#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1725 _is_utf8_perl_idstart, p)
e5dcd934 1726
b7d90381
KW
1727#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1728#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1729#define isPSXSPC_utf8(p) isSPACE_utf8(p)
1730#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1731#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1732#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1733#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1734#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1735#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
cd500f2f 1736 is_XDIGIT_high(p))
a0ed51b3 1737
7d2b61dd 1738#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1739#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1740#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1741#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1742
4650c663
KW
1743/* For internal core Perl use only: the base macros for defining macros like
1744 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1745 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1746 * 'macro' parameter */
94bb8c36
KW
1747#define _generic_LC_utf8(macro, p, utf8) \
1748 (UTF8_IS_INVARIANT(*(p)) \
1749 ? macro(*(p)) \
1750 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
a62b247b 1751 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
9201a572
KW
1752 : utf8)
1753
cd500f2f
KW
1754#define _generic_LC_swash_utf8(macro, classnum, p) \
1755 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1756#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1757 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1758
b7d90381 1759#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
cd500f2f 1760 _CC_ALPHANUMERIC, p)
b7d90381
KW
1761#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1762#define isASCII_LC_utf8(p) isASCII_LC(*p)
1763#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1764 is_HORIZWS_high, p)
1765#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1766#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1767#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1768#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1769 _is_utf8_perl_idcont, p)
1770#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1771 _is_utf8_perl_idstart, p)
1772#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1773#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1774#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1775#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1776#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1777 is_XPERLSPACE_high, p)
1778#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
cd500f2f
KW
1779#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1780 _CC_WORDCHAR, p)
b7d90381
KW
1781#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1782 is_XDIGIT_high, p)
aaa51d5e 1783
fbc19f27
KW
1784/* Macros for backwards compatibility and for completeness when the ASCII and
1785 * Latin1 values are identical */
b7d90381
KW
1786#define isALPHAU(c) isALPHA_L1(c)
1787#define isDIGIT_L1(c) isDIGIT_A(c)
1788#define isOCTAL(c) isOCTAL_A(c)
1789#define isOCTAL_L1(c) isOCTAL_A(c)
1790#define isXDIGIT_L1(c) isXDIGIT_A(c)
1791#define isALNUM(c) isWORDCHAR(c)
1792#define isALNUMU(c) isWORDCHAR_L1(c)
1793#define isALNUM_LC(c) isWORDCHAR_LC(c)
1794#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 1795#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
1796#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1797#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1798#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1799#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1800#define isALNUMC(c) isALPHANUMERIC(c)
1801#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1802#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 1803#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 1804#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 1805#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1806
2bd1cbf6
KW
1807/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1808 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1809 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1810 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1811 * the outlier from the block that contains the other controls, just like
1812 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1813 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
1814 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1815 * etc. */
2bd1cbf6 1816#ifndef EBCDIC
75763b3a 1817# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 1818#else
75763b3a
KW
1819# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1820 ((isPRINT_A(c)) \
1821 ? (UNLIKELY((c) == '?') \
1822 ? QUESTION_MARK_CTRL \
1823 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1824 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1825 ? '?' \
1826 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 1827#endif
bbce6d69 1828
dea28490
JJ
1829/* Line numbers are unsigned, 32 bits. */
1830typedef U32 line_t;
e5dcd934 1831#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 1832
91152fc1
DG
1833/* Helpful alias for version prescan */
1834#define is_LAX_VERSION(a,b) \
1835 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1836
1837#define is_STRICT_VERSION(a,b) \
1838 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1839
1840#define BADVERSION(a,b,c) \
1841 if (b) { \
1842 *b = c; \
1843 } \
1844 return a;
8c52afec 1845
cb27eebd
KW
1846/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1847 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1848 * which is advanced. The input is validated only by an assert() in DEBUGGING
1849 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1850 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1851#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1852 ? (c) \
1853 : ((c) + 9))))
902008b8
KW
1854#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1855 ? (*(s)++) \
1856 : (*(s)++ + 9))))
95a59cab 1857
cb27eebd
KW
1858/* Converts a character known to represent an octal digit (0-7) to its numeric
1859 * value. The input is validated only by an assert() in DEBUGGING builds. In
1860 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1861#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1862
305b8651
KW
1863/* Efficiently returns a boolean as to if two native characters are equivalent
1864 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1865 * the ALPHA in the name is to remind you of that. This is asserted() in
1866 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1867 * works (on valid input) for both non- and UTF-8-encoded bytes.
1868 *
1869 * When one of the inputs is a compile-time constant and gets folded by the
1870 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1871 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1872 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
1873 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1874 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1875 * just a single 0, in the bit position where the upper- and lowercase differ.
1876 * */
305b8651
KW
1877#define isALPHA_FOLD_EQ(c1, c2) \
1878 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1879 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1880#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1881
8e84507e 1882/*
ccfc67b7
JH
1883=head1 Memory Management
1884
a02a5408 1885=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
1886The XSUB-writer's interface to the C C<malloc> function.
1887
596f7718 1888Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1889
c5008215
JC
1890In 5.9.3, Newx() and friends replace the older New() API, and drops
1891the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 1892themselves. This aid has been superseded by a new build option,
d10b4965 1893PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
1894there for use in XS modules supporting older perls.
1895
a02a5408 1896=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 1897The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 1898cast. See also C<L</Newx>>.
954c1994 1899
596f7718 1900Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1901
a02a5408 1902=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 1903The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 1904memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 1905
596f7718 1906Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1907
954c1994
GS
1908=for apidoc Am|void|Renew|void* ptr|int nitems|type
1909The XSUB-writer's interface to the C C<realloc> function.
1910
596f7718 1911Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1912
954c1994
GS
1913=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1914The XSUB-writer's interface to the C C<realloc> function, with
1915cast.
1916
596f7718 1917Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1918
49b8b560 1919=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
1920The XSUB-writer's interface to the C C<free> function.
1921
596f7718 1922This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 1923
954c1994
GS
1924=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1925The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 1926source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1927C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 1928
e90e2364 1929=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 1930Like C<Move> but returns C<dest>. Useful
72d33970 1931for encouraging compilers to tail-call
e90e2364
NC
1932optimise.
1933
954c1994
GS
1934=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1935The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 1936source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1937C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 1938
e90e2364
NC
1939=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1940
796b6530 1941Like C<Copy> but returns C<dest>. Useful
72d33970 1942for encouraging compilers to tail-call
e90e2364
NC
1943optimise.
1944
954c1994
GS
1945=for apidoc Am|void|Zero|void* dest|int nitems|type
1946
1947The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1948destination, C<nitems> is the number of items, and C<type> is the type.
1949
e90e2364
NC
1950=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1951
72d33970
FC
1952Like C<Zero> but returns dest. Useful
1953for encouraging compilers to tail-call
e90e2364
NC
1954optimise.
1955
da5d8dbb 1956=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 1957This is an architecture-independent macro to copy one structure to another.
954c1994 1958
7e337ee0
JH
1959=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1960
1961Fill up memory with a byte pattern (a byte repeated over and over
1962again) that hopefully catches attempts to access uninitialized memory.
1963
1964=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1965
1966PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1967
1c12ffb4 1968=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
1969
1970PoisonWith(0xEF) for catching access to freed memory.
1971
9965345d
JH
1972=for apidoc Am|void|Poison|void* dest|int nitems|type
1973
7e337ee0 1974PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
1975
1976=cut */
954c1994 1977
561b68a9
SH
1978/* Maintained for backwards-compatibility only. Use newSV() instead. */
1979#ifndef PERL_CORE
ff06c60c 1980#define NEWSV(x,len) newSV(len)
561b68a9 1981#endif
ff06c60c 1982
19a94d75
JH
1983#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1984
e6bdf523 1985
27d5b266 1986#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
1987
1988/* This expression will be constant-folded at compile time. It checks
1989 * whether or not the type of the count n is so small (e.g. U8 or U16, or
1990 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
1991 * As well as avoiding the need for a run-time check in some cases, it's
1992 * designed to avoid compiler warnings like:
1993 * comparison is always false due to limited range of data type
73e8ff00
DM
1994 * It's mathematically equivalent to
1995 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
1996 */
1997
1998# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
004073ba 1999 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
e6bdf523 2000
88f9f128 2001/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2002 * compiler warnings. We *want* to write the expression as
2003 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2004 * (for some compile-time constant C), but even when the LHS
2005 * constant-folds to false at compile-time, g++ insists on emitting
2006 * warnings about the RHS (e.g. "comparison is always false"), so instead
2007 * we write it as
e6bdf523 2008 *
d98e5cde 2009 * (cond ? n : X) > C
88f9f128 2010 *
d98e5cde
DM
2011 * where X is a constant with X > C always false. Choosing a value for X
2012 * is tricky. If 0, some compilers will complain about 0 > C always being
2013 * false; if 1, Coverity complains when n happens to be the constant value
2014 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2015 * for X and hope that nothing else whines.
e6bdf523
DM
2016 */
2017
2018# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2019 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2020 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2021
2022# define MEM_WRAP_CHECK(n,t) \
2023 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2024 && (croak_memory_wrap(),0))
2025
2026# define MEM_WRAP_CHECK_1(n,t,a) \
2027 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2028 && (Perl_croak_nocontext("%s",(a)),0))
2029
8b44ba4c 2030#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2031
d1decf2b 2032#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
2033#else
2034
410319be
NC
2035#define MEM_WRAP_CHECK(n,t)
2036#define MEM_WRAP_CHECK_1(n,t,a)
2037#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
2038#define MEM_WRAP_CHECK_(n,t)
2039
6fff79ce 2040#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 2041
1936d2a7 2042#endif
8b44ba4c 2043
fe4f188c 2044#ifdef PERL_MEM_LOG
46c6c7e2 2045/*
9f653bb5 2046 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2047 * go through functions, which are handy for debugging breakpoints, but
2048 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2049 * line number, and C function name if available) passed in. This info can
2050 * then be used for logging the calls, for which one gets a sample
73d1d973 2051 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2052 *
46c6c7e2 2053 * Known problems:
94e892a6 2054 * - not all memory allocs get logged, only those
46c6c7e2 2055 * that go through Newx() and derivatives (while all
94e892a6 2056 * Safefrees do get logged)
46c6c7e2
JH
2057 * - __FILE__ and __LINE__ do not work everywhere
2058 * - __func__ or __FUNCTION__ even less so
2059 * - I think more goes on after the perlio frees but
2060 * the thing is that STDERR gets closed (as do all
2061 * the file descriptors)
2062 * - no deeper calling stack than the caller of the Newx()
2063 * or the kind, but do I look like a C reflection/introspection
2064 * utility to you?
2065 * - the function prototypes for the logging functions
2066 * probably should maybe be somewhere else than handy.h
2067 * - one could consider inlining (macrofying) the logging
2068 * for speed, but I am too lazy
2069 * - one could imagine recording the allocations in a hash,
2070 * (keyed by the allocation address?), and maintain that
2071 * through reallocs and frees, but how to do that without
2072 * any News() happening...?
73d1d973 2073 * - lots of -Ddefines to get useful/controllable output
b953482e 2074 * - lots of ENV reads
46c6c7e2
JH
2075 */
2076
0b0ab801 2077# ifdef PERL_CORE
73d1d973 2078# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2079enum mem_log_type {
2080 MLT_ALLOC,
2081 MLT_REALLOC,
d7a2c63c
MHM
2082 MLT_FREE,
2083 MLT_NEW_SV,
2084 MLT_DEL_SV
0b0ab801
MHM
2085};
2086# endif
12754f92 2087# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2088void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2089void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2090# endif
0b0ab801
MHM
2091# endif
2092
fe4f188c
JH
2093#endif
2094
2095#ifdef PERL_MEM_LOG
d1401ee9
MHM
2096#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2097#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2098#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2099#endif
2100
2101#ifndef MEM_LOG_ALLOC
2102#define MEM_LOG_ALLOC(n,t,a) (a)
2103#endif
2104#ifndef MEM_LOG_REALLOC
2105#define MEM_LOG_REALLOC(n,t,v,a) (a)
2106#endif
2107#ifndef MEM_LOG_FREE
2108#define MEM_LOG_FREE(a) (a)
2109#endif
2110
d1401ee9
MHM
2111#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2112#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2113#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2114
2115#ifndef PERL_CORE
a02a5408
JC
2116/* pre 5.9.x compatibility */
2117#define New(x,v,n,t) Newx(v,n,t)
2118#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2119#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2120#endif
a02a5408 2121
ff68c719 2122#define Renew(v,n,t) \
d1401ee9 2123 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2124#define Renewc(v,n,t,c) \
d1401ee9 2125 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2126
2127#ifdef PERL_POISON
2128#define Safefree(d) \
06c0cc96 2129 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2130#else
fe4f188c 2131#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2132#endif
55497cff 2133
8b44ba4c
NC
2134#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2135#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2136#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2137
8b44ba4c
NC
2138#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2139#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2140#ifdef HAS_MEMSET
8b44ba4c 2141#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2142#else
8b44ba4c
NC
2143/* Using bzero(), which returns void. */
2144#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2145#endif
2146
7e337ee0
JH
2147#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2148#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2149#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2150#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2151
caa674f3
DD
2152#ifdef PERL_POISON
2153# define PERL_POISON_EXPR(x) x
2154#else
2155# define PERL_POISON_EXPR(x)
2156#endif
2157
2304df62 2158#ifdef USE_STRUCT_COPY
ff68c719 2159#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2160#else
2161#define StructCopy(s,d,t) Copy(s,d,1,t)
2162#endif
2cc61e15 2163
c3caa5c3
JH
2164/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2165 * want your zero-based indices to be less than but not equal to).
2166 *
2167 * C_ARRAY_END is one past the last: half-open/half-closed range,
2168 * not last-inclusive range. */
622913ab 2169#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2170#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2171
2cc61e15
DD
2172#ifdef NEED_VA_COPY
2173# ifdef va_copy
2174# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2175# else
a1866d1b
JH
2176# if defined(__va_copy)
2177# define Perl_va_copy(s, d) __va_copy(d, s)
2178# else
2179# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2180# endif
2cc61e15
DD
2181# endif
2182#endif
2183
472d47bc
SB
2184/* convenience debug macros */
2185#ifdef USE_ITHREADS
2186#define pTHX_FORMAT "Perl interpreter: 0x%p"
2187#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2188#define pTHX_VALUE_ (void *)my_perl,
2189#define pTHX_VALUE (void *)my_perl
2190#define pTHX__VALUE_ ,(void *)my_perl,
2191#define pTHX__VALUE ,(void *)my_perl
472d47bc 2192#else
3609ea0d 2193#define pTHX_FORMAT
472d47bc 2194#define pTHX__FORMAT
3609ea0d 2195#define pTHX_VALUE_
472d47bc 2196#define pTHX_VALUE
3609ea0d 2197#define pTHX__VALUE_
472d47bc
SB
2198#define pTHX__VALUE
2199#endif /* USE_ITHREADS */
3609ea0d 2200
2acdbac1
NC
2201/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2202 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2203 CPAN::Unpack show any users outside the core. */
2204#ifdef PERL_CORE
b7d90381
KW
2205# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2206 "Use of " s " is deprecated")
2acdbac1
NC
2207#endif
2208
dfff4baf
BF
2209/* Internal macros to deal with gids and uids */
2210#ifdef PERL_CORE
2211
2212# if Uid_t_size > IVSIZE
2213# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2214# define SvUID(sv) SvNV(sv)
2215# else
2216# if Uid_t_sign <= 0
2217# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2218# define SvUID(sv) SvIV(sv)
2219# else
2220# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2221# define SvUID(sv) SvUV(sv)
2222# endif
2223# endif /* Uid_t_size */
2224
2225# if Gid_t_size > IVSIZE
2226# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2227# define SvGID(sv) SvNV(sv)
2228# else
2229# if Gid_t_sign <= 0
2230# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2231# define SvGID(sv) SvIV(sv)
2232# else
2233# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2234# define SvGID(sv) SvUV(sv)
2235# endif
2236# endif /* Gid_t_size */
2237
2238#endif
2239
9d745869
KW
2240#endif /* HANDY_H */
2241
e9a8c099 2242/*
14d04a33 2243 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2244 */