This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
gv.c: document where @- and @+ (and %- and %+ are initialized in gv.c)
[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)
a038e571
NC
420#define hv_fetchs(hv,key,lval) \
421 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
422 (lval) ? (HV_FETCH_JUST_SV | HV_FETCH_LVALUE) \
423 : HV_FETCH_JUST_SV, NULL, 0))
424
425#define hv_stores(hv,key,val) \
426 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
427 (HV_FETCH_ISSTORE|HV_FETCH_JUST_SV), (val), 0))
ba3a79e7 428
510966aa
Z
429#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
430
b96d8cd9
NC
431#define get_cvs(str, flags) \
432 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 433
954c1994 434/*
ccfc67b7
JH
435=head1 Miscellaneous Functions
436
954c1994 437=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
438Test two C<NUL>-terminated strings to see if they are different. Returns true
439or false.
954c1994
GS
440
441=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
442Test two C<NUL>-terminated strings to see if they are equal. Returns true or
443false.
954c1994
GS
444
445=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
446Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
447second, C<s2>. Returns true or false.
954c1994
GS
448
449=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
450Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
451equal to the second, C<s2>. Returns true or false.
954c1994
GS
452
453=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
454Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
455the second, C<s2>. Returns true or false.
954c1994
GS
456
457=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
458Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
459or equal to the second, C<s2>. Returns true or false.
954c1994
GS
460
461=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
462Test two C<NUL>-terminated strings to see if they are different. The C<len>
463parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
464wrapper for C<strncmp>).
465
466=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
467Test two C<NUL>-terminated strings to see if they are equal. The C<len>
468parameter indicates the number of bytes to compare. Returns true or false. (A
469wrapper for C<strncmp>).
954c1994 470
bd18bd40
KW
471=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
472Test two buffers (which may contain embedded C<NUL> characters, to see if they
473are equal. The C<len> parameter indicates the number of bytes to compare.
474Returns zero if equal, or non-zero if non-equal.
475
476=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
477Test two buffers (which may contain embedded C<NUL> characters, to see if they
478are not equal. The C<len> parameter indicates the number of bytes to compare.
479Returns zero if non-equal, or non-zero if equal.
480
954c1994
GS
481=cut
482*/
483
8d063cd8
LW
484#define strNE(s1,s2) (strcmp(s1,s2))
485#define strEQ(s1,s2) (!strcmp(s1,s2))
486#define strLT(s1,s2) (strcmp(s1,s2) < 0)
487#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
488#define strGT(s1,s2) (strcmp(s1,s2) > 0)
489#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
490#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
491#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
378cc40b 492
36477c24
PP
493#ifdef HAS_MEMCMP
494# define memNE(s1,s2,l) (memcmp(s1,s2,l))
495# define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
496#else
497# define memNE(s1,s2,l) (bcmp(s1,s2,l))
498# define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
499#endif
500
085b7534 501/* memEQ and memNE where second comparand is a string constant */
568a785a 502#define memEQs(s1, l, s2) \
8e6497b6 503 (sizeof(s2)-1 == l && memEQ(s1, ("" s2 ""), (sizeof(s2)-1)))
568a785a
NC
504#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
505
062b6850
KW
506#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
507#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
508#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
509#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
510
bbce6d69
PP
511/*
512 * Character classes.
513 *
514 * Unfortunately, the introduction of locales means that we
515 * can't trust isupper(), etc. to tell the truth. And when
516 * it comes to /\w+/ with tainting enabled, we *must* be able
517 * to trust our character classes.
518 *
519 * Therefore, the default tests in the text of Perl will be
520 * independent of locale. Any code that wants to depend on
521 * the current locale will use the tests that begin with "lc".
522 */
523
2304df62
AD
524#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
525# ifndef CTYPE256
526# define CTYPE256
527# endif
528#endif
529
954c1994 530/*
ccfc67b7 531
dcccc8ff 532=head1 Character classification
243effed
KW
533This section is about functions (really macros) that classify characters
534into types, such as punctuation versus alphabetic, etc. Most of these are
535analogous to regular expression character classes. (See
536L<perlrecharclass/POSIX Character Classes>.) There are several variants for
537each class. (Not all macros have all variants; each item below lists the
538ones valid for it.) None are affected by C<use bytes>, and only the ones
539with C<LC> in the name are affected by the current locale.
540
541The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
542C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
543represented by that octet is (or on non-ASCII platforms, corresponds to) an
544ASCII character in the named class based on platform, Unicode, and Perl rules.
545If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 546
6aff1f14
KW
547Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
548with no suffix C<"_A">.
4b9734bf
KW
549
550Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
551onto the platform. That is, the code points that are ASCII are unaffected,
552since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
553as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
554true when called with the code point 0xDF, which is a word character in both
4650c663 555ASCII and EBCDIC (though it represents different characters in each).
243effed 556
d0da05db 557Variant C<isFOO_uvchr> is like the C<isFOO_L1> variant, but accepts any UV code
243effed
KW
558point as input. If the code point is larger than 255, Unicode rules are used
559to determine if it is in the character class. For example,
d0da05db 560C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 561WITH MACRON in Unicode, and is a word character.
243effed 562
d0da05db 563Variant C<isFOO_utf8> is like C<isFOO_uvchr>, but the input is a pointer to a
dc6b0978
KW
564(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
565possibly containing embedded C<NUL> characters). The classification of just
566the first (possibly multi-byte) character in the string is tested.
243effed 567
31f05a37
KW
568Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
569result is based on the current locale, which is what C<LC> in the name stands
570for. If Perl can determine that the current locale is a UTF-8 locale, it uses
571the published Unicode rules; otherwise, it uses the C library function that
572gives the named classification. For example, C<isDIGIT_LC()> when not in a
573UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
574returned if the input won't fit into an octet. On some platforms where the C
575library function is known to be defective, Perl changes its result to follow
576the POSIX standard's rules.
243effed
KW
577
578Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
579returns the same as C<isFOO_LC> for input code points less than 256, and
580returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
581
b7d90381 582Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer
dc6b0978
KW
583to a (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
584possibly containing embedded C<NUL> characters). The classification of just
585the first (possibly multi-byte) character in the string is tested.
ccfc67b7 586
954c1994 587=for apidoc Am|bool|isALPHA|char ch
2787a470 588Returns a boolean indicating whether the specified character is an
6aff1f14 589alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
590See the L<top of this section|/Character classification> for an explanation of
591variants
d0da05db 592C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>, C<isALPHA_LC>,
243effed 593C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
8a58bdcf 594
15861f94
KW
595=for apidoc Am|bool|isALPHANUMERIC|char ch
596Returns a boolean indicating whether the specified character is a either an
6aff1f14 597alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
598See the L<top of this section|/Character classification> for an explanation of
599variants
d0da05db 600C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
e5ad6aba 601C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
15861f94
KW
602C<isALPHANUMERIC_LC_utf8>.
603
8a58bdcf
KW
604=for apidoc Am|bool|isASCII|char ch
605Returns a boolean indicating whether the specified character is one of the 128
243effed 606characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 607On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
608character corresponds to an ASCII character. Variants C<isASCII_A()> and
609C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
610See the L<top of this section|/Character classification> for an explanation of
611variants
d0da05db 612C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
243effed
KW
613C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
614library routine C<isascii()>. In these cases, the variants whose names contain
615C<LC> are the same as the corresponding ones without.
616
d98532ea
KW
617Also note, that because all ASCII characters are UTF-8 invariant (meaning they
618have the exact same representation (always a single byte) whether encoded in
619UTF-8 or not), C<isASCII> will give the correct results when called with any
620byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
621work properly on any string encoded or not in UTF-8.
622
243effed
KW
623=for apidoc Am|bool|isBLANK|char ch
624Returns a boolean indicating whether the specified character is a
6aff1f14 625character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
626See the L<top of this section|/Character classification> for an explanation of
627variants
d0da05db 628C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>, C<isBLANK_LC>,
243effed
KW
629C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
630platforms do not have the C library routine C<isblank()>. In these cases, the
631variants whose names contain C<LC> are the same as the corresponding ones
632without.
633
634=for apidoc Am|bool|isCNTRL|char ch
635Returns a boolean indicating whether the specified character is a
6aff1f14 636control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
637See the L<top of this section|/Character classification> for an explanation of
638variants
d0da05db 639C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
6aff1f14
KW
640C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
641On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 642
954c1994 643=for apidoc Am|bool|isDIGIT|char ch
2787a470 644Returns a boolean indicating whether the specified character is a
6aff1f14 645digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 646Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
647See the L<top of this section|/Character classification> for an explanation of
648variants
d0da05db 649C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
243effed
KW
650C<isDIGIT_LC_utf8>.
651
652=for apidoc Am|bool|isGRAPH|char ch
653Returns a boolean indicating whether the specified character is a
6aff1f14 654graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff
KW
655See the L<top of this section|/Character classification> for an explanation of
656variants
d0da05db 657C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
243effed 658C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
954c1994 659
0c82b6df 660=for apidoc Am|bool|isLOWER|char ch
2787a470 661Returns a boolean indicating whether the specified character is a
6aff1f14 662lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
663See the L<top of this section|/Character classification> for an explanation of
664variants
d0da05db 665C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>, C<isLOWER_LC>,
243effed 666C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
0c82b6df 667
c99e91e9 668=for apidoc Am|bool|isOCTAL|char ch
2787a470 669Returns a boolean indicating whether the specified character is an
6aff1f14 670octal digit, [0-7].
243effed
KW
671The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
672C<isOCTAL>.
673
674=for apidoc Am|bool|isPUNCT|char ch
675Returns a boolean indicating whether the specified character is a
6aff1f14
KW
676punctuation character, analogous to C<m/[[:punct:]]/>.
677Note that the definition of what is punctuation isn't as
243effed
KW
678straightforward as one might desire. See L<perlrecharclass/POSIX Character
679Classes> for details.
dcccc8ff
KW
680See the L<top of this section|/Character classification> for an explanation of
681variants
d0da05db 682C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
243effed 683C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
c99e91e9 684
0c82b6df 685=for apidoc Am|bool|isSPACE|char ch
2787a470 686Returns a boolean indicating whether the specified character is a
6aff1f14 687whitespace character. This is analogous
398d098a 688to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 689this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
690locale forms of this macro (the ones with C<LC> in their names) matched
691precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
692in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
693(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
694See the L<top of this section|/Character classification> for an explanation of
695variants
d0da05db 696C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>, C<isSPACE_LC>,
243effed 697C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
0c82b6df 698
398d098a
KW
699=for apidoc Am|bool|isPSXSPC|char ch
700(short for Posix Space)
779cf272
KW
701Starting in 5.18, this is identical in all its forms to the
702corresponding C<isSPACE()> macros.
398d098a
KW
703The locale forms of this macro are identical to their corresponding
704C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
705non-locale forms differ from their C<isSPACE()> forms only in that the
706C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
707Otherwise they are identical. Thus this macro is analogous to what
708C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 709See the L<top of this section|/Character classification> for an explanation of
d0da05db 710variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
b7d90381 711C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
398d098a 712
954c1994 713=for apidoc Am|bool|isUPPER|char ch
2787a470 714Returns a boolean indicating whether the specified character is an
6aff1f14 715uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff
KW
716See the L<top of this section|/Character classification> for an explanation of
717variants
d0da05db 718C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>, C<isUPPER_LC>,
243effed 719C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
954c1994 720
243effed 721=for apidoc Am|bool|isPRINT|char ch
8eea39dd 722Returns a boolean indicating whether the specified character is a
6aff1f14 723printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
724See the L<top of this section|/Character classification> for an explanation of
725variants
d0da05db 726C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>, C<isPRINT_LC>,
243effed
KW
727C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
728
729=for apidoc Am|bool|isWORDCHAR|char ch
730Returns a boolean indicating whether the specified character is a character
731that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
732in a regular expression. A word character is an alphabetic character, a
733decimal digit, a connecting punctuation character (such as an underscore), or
734a "mark" character that attaches to one of those (like some sort of accent).
735C<isALNUM()> is a synonym provided for backward compatibility, even though a
736word character includes more than the standard C language meaning of
737alphanumeric.
dcccc8ff
KW
738See the L<top of this section|/Character classification> for an explanation of
739variants
d0da05db 740C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and C<isWORDCHAR_utf8>.
1a83413c
KW
741C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
742described there, but additionally include the platform's native underscore.
8a58bdcf
KW
743
744=for apidoc Am|bool|isXDIGIT|char ch
745Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
746digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
747and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
748See the L<top of this section|/Character classification> for an explanation of
749variants
d0da05db 750C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
243effed
KW
751C<isXDIGIT_LC_utf8>.
752
3c3ecf18
KW
753=for apidoc Am|bool|isIDFIRST|char ch
754Returns a boolean indicating whether the specified character can be the first
755character of an identifier. This is very close to, but not quite the same as
756the official Unicode property C<XID_Start>. The difference is that this
757returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
758See the L<top of this section|/Character classification> for an explanation of
759variants
d0da05db 760C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
3c3ecf18
KW
761C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
762
763=for apidoc Am|bool|isIDCONT|char ch
764Returns a boolean indicating whether the specified character can be the
765second or succeeding character of an identifier. This is very close to, but
766not quite the same as the official Unicode property C<XID_Continue>. The
767difference is that this returns true only if the input character also matches
dcccc8ff
KW
768L</isWORDCHAR>. See the L<top of this section|/Character classification> for
769an
d0da05db 770explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
3c3ecf18
KW
771C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
772C<isIDCONT_LC_utf8>.
773
243effed 774=head1 Miscellaneous Functions
8eea39dd 775
95a59cab 776=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 777Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
778Behaviour is only well defined when isXDIGIT(*str) is true.
779
e7c1e6c1
KW
780=head1 Character case changing
781
1f607577
KW
782=for apidoc Am|U8|toUPPER|U8 ch
783Converts the specified character to uppercase. If the input is anything but an
784ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 785C<toUPPER_A> is equivalent.
954c1994 786
d0da05db
KW
787=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
788Converts the code point C<cp> to its uppercase version, and
789stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
790point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
791that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
792bytes since the uppercase version may be longer than the original character.
793
794The first code point of the uppercased version is returned
795(but note, as explained just above, that there may be more.)
796
797=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
798Converts the UTF-8 encoded character at C<p> to its uppercase version, and
799stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
800that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
801bytes since the uppercase version may be longer than the original character.
802
803The first code point of the uppercased version is returned
804(but note, as explained just above, that there may be more.)
805
806The input character at C<p> is assumed to be well-formed.
807
25200305
KW
808=for apidoc Am|U8|toFOLD|U8 ch
809Converts the specified character to foldcase. If the input is anything but an
810ASCII uppercase character, that input character itself is returned. Variant
811C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 812Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 813
d0da05db
KW
814=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
815Converts the code point C<cp> to its foldcase version, and
816stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
817point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
818that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
819bytes since the foldcase version may be longer than the original character.
820
821The first code point of the foldcased version is returned
822(but note, as explained just above, that there may be more.)
823
824=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
825Converts the UTF-8 encoded character at C<p> to its foldcase version, and
826stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
827that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
828bytes since the foldcase version may be longer than the original character.
829
830The first code point of the foldcased version is returned
831(but note, as explained just above, that there may be more.)
832
833The input character at C<p> is assumed to be well-formed.
834
835=for apidoc Am|U8|toLOWER|U8 ch
836Converts the specified character to lowercase. If the input is anything but an
837ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 838C<toLOWER_A> is equivalent.
954c1994 839
1f607577 840=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
841Converts the specified Latin1 character to lowercase. The results are
842undefined if the input doesn't fit in a byte.
1f607577
KW
843
844=for apidoc Am|U8|toLOWER_LC|U8 ch
845Converts the specified character to lowercase using the current locale's rules,
846if possible; otherwise returns the input character itself.
847
d0da05db
KW
848=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
849Converts the code point C<cp> to its lowercase version, and
850stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
851point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
852that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
853bytes since the lowercase version may be longer than the original character.
854
855The first code point of the lowercased version is returned
856(but note, as explained just above, that there may be more.)
857
858=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
859Converts the UTF-8 encoded character at C<p> to its lowercase version, and
860stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
861that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
862bytes since the lowercase version may be longer than the original character.
863
864The first code point of the lowercased version is returned
865(but note, as explained just above, that there may be more.)
866
867The input character at C<p> is assumed to be well-formed.
868
25200305
KW
869=for apidoc Am|U8|toTITLE|U8 ch
870Converts the specified character to titlecase. If the input is anything but an
871ASCII lowercase character, that input character itself is returned. Variant
b7d90381 872C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 873range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 874not a concept used in locale handling, so there is no functionality for that.)
25200305 875
d0da05db
KW
876=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
877Converts the code point C<cp> to its titlecase version, and
878stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
879point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
880that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
881bytes since the titlecase version may be longer than the original character.
882
883The first code point of the titlecased version is returned
884(but note, as explained just above, that there may be more.)
885
886=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
887Converts the UTF-8 encoded character at C<p> to its titlecase version, and
888stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
889that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
890bytes since the titlecase version may be longer than the original character.
891
892The first code point of the titlecased version is returned
893(but note, as explained just above, that there may be more.)
894
895The input character at C<p> is assumed to be well-formed.
896
954c1994 897=cut
353c9b6f 898
d0da05db 899XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
25200305 900really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 901
8a58bdcf 902Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
903patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
904
954c1994
GS
905*/
906
7c062697
KW
907/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
908 * is known only in the perl core, and this macro can be called from outside
909 * that */
910#ifdef HAS_QUAD
911# define WIDEST_UTYPE U64TYPE
912#else
913# define WIDEST_UTYPE U32
914#endif
915
3912bc88
KW
916/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
917 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
918 * bits of information are lost even on a 64-bit machine, but to get the
919 * compiler to optimize it out if possible. This is because Configure makes
920 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
921 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
922 *
923 * For Coverity, be always true, because otherwise Coverity thinks
924 * it finds several expressions that are always true, independent
925 * of operands. Well, they are, but that is kind of the point.
220c71bf 926 */
7e75d1a1 927#ifndef __COVERITY__
6c5b02ac
KW
928 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
929 * pointer) */
930#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 931 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
932#else
933#define FITS_IN_8_BITS(c) (1)
934#endif
cf301eb7 935
41f43cc2 936#ifdef EBCDIC
b6340bd0 937# ifndef _ALL_SOURCE
0852beac
KW
938 /* The native libc isascii() et.al. functions return the wrong results
939 * on at least z/OS unless this is defined. */
b6340bd0
KW
940# error _ALL_SOURCE should probably be defined
941# endif
41f43cc2 942#else
0852beac
KW
943 /* There is a simple definition of ASCII for ASCII platforms. But the
944 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 945 * other macros below.
3f3c579d
KW
946 *
947 * The cast here is used instead of '(c) >= 0', because some compilers emit
948 * a warning that that test is always true when the parameter is an
949 * unsigned type. khw supposes that it could be written as
950 * && ((c) == '\0' || (c) > 0)
951 * to avoid the message, but the cast will likely avoid extra branches even
952 * with stupid compilers.
953 *
954 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
955 * a pointer) */
9c903d59 956# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
957#endif
958
38694112
KW
959/* Take the eight possible bit patterns of the lower 3 bits and you get the
960 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
961 * can be ignored. If the rest match '0', we have an octal */
962#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 963
9fb1bf9d 964#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 965
a500dc72
KW
966/* Character class numbers. For internal core Perl use only. The ones less
967 * than 32 are used in PL_charclass[] and the ones up through the one that
968 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
969 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
970 * their actual definitions are here. If that file has a name not used here,
971 * it won't compile.
1709d539
KW
972 *
973 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
974 * frequency of their use. This gives a slight edge to exiting a loop earlier
975 * (in reginclass() in regexec.c) */
1709d539
KW
976# define _CC_WORDCHAR 0 /* \w and [:word:] */
977# define _CC_DIGIT 1 /* \d and [:digit:] */
978# define _CC_ALPHA 2 /* [:alpha:] */
979# define _CC_LOWER 3 /* [:lower:] */
980# define _CC_UPPER 4 /* [:upper:] */
981# define _CC_PUNCT 5 /* [:punct:] */
982# define _CC_PRINT 6 /* [:print:] */
15861f94 983# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 984# define _CC_GRAPH 8 /* [:graph:] */
359b005e 985# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1709d539 986
b0d691b2 987#define _FIRST_NON_SWASH_CC 10
1709d539
KW
988/* The character classes above are implemented with swashes. The second group
989 * (just below) contains the ones implemented without. These are also sorted
990 * in rough order of the frequency of their use, except that \v should be last,
991 * as it isn't a real Posix character class, and some (small) inefficiencies in
992 * regular expression handling would be introduced by putting it in the middle
993 * of those that are. Also, cntrl and ascii come after the others as it may be
994 * useful to group these which have no members that match above Latin1, (or
995 * above ASCII in the latter case) */
996
779cf272 997# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
998# define _CC_BLANK 11 /* [:blank:] */
999# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1000# define _CC_CNTRL 13 /* [:cntrl:] */
1001# define _CC_ASCII 14 /* [:ascii:] */
1002# define _CC_VERTSPACE 15 /* \v */
1709d539 1003
a0947d7b
KW
1004# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1005
1709d539 1006/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1007 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1008# define _CC_IDFIRST 16
1009# define _CC_CHARNAME_CONT 17
1010# define _CC_NONLATIN1_FOLD 18
1011# define _CC_NONLATIN1_SIMPLE_FOLD 19
1012# define _CC_QUOTEMETA 20
1013# define _CC_NON_FINAL_FOLD 21
1014# define _CC_IS_IN_SOME_FOLD 22
1015# define _CC_MNEMONIC_CNTRL 23
073c22b3
KW
1016
1017/* This next group is only used on EBCDIC platforms, so theoretically could be
1018 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
1019# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1020# define _CC_UTF8_IS_START 29
1021# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1022# define _CC_UTF8_IS_CONTINUATION 31
1023/* Unused: 24-27
f4cdb42c
KW
1024 * If more bits are needed, one could add a second word for non-64bit
1025 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1026 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1027 * is used only for optimization (as of this writing), and differs in the
1028 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1029 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1030 * would just cause /i regexes which match them to run less efficiently.
1031 * Similarly the EBCDIC-only bits are used just for speed, and could be
1032 * replaced by other means */
96ac0975 1033
3a371f2f
KW
1034#if defined(PERL_CORE) || defined(PERL_EXT)
1035/* An enum version of the character class numbers, to help compilers
1036 * optimize */
1037typedef enum {
3a371f2f 1038 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1039 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1040 _CC_ENUM_ASCII = _CC_ASCII,
1041 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1042 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1043 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1044 _CC_ENUM_DIGIT = _CC_DIGIT,
1045 _CC_ENUM_GRAPH = _CC_GRAPH,
1046 _CC_ENUM_LOWER = _CC_LOWER,
1047 _CC_ENUM_PRINT = _CC_PRINT,
1048 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1049 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1050 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1051 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1052 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1053 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1054} _char_class_number;
1055#endif
1056
63c61c3f 1057#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1058#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1059
b7d90381
KW
1060#if defined(PERL_IN_UTF8_C) \
1061 || defined(PERL_IN_REGCOMP_C) \
1062 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1063# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1064 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1065 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1066 #error Need to adjust order of swash_property_names[]
1067# endif
1068
b44daeeb
KW
1069/* This is declared static in each of the few files that this is #defined for
1070 * to keep them from being publicly accessible. Hence there is a small amount
1071 * of wasted space */
1072
922e8cb4
KW
1073static const char* const swash_property_names[] = {
1074 "XPosixWord",
1075 "XPosixDigit",
1076 "XPosixAlpha",
1077 "XPosixLower",
1078 "XPosixUpper",
1079 "XPosixPunct",
1080 "XPosixPrint",
1081 "XPosixAlnum",
b0d691b2
KW
1082 "XPosixGraph",
1083 "Cased"
922e8cb4
KW
1084};
1085#endif
1086
6635f04f 1087START_EXTERN_C
96ac0975
NC
1088# ifdef DOINIT
1089EXTCONST U32 PL_charclass[] = {
1090# include "l1_char_class_tab.h"
1091};
1092
1093# else /* ! DOINIT */
1094EXTCONST U32 PL_charclass[];
1095# endif
6635f04f 1096END_EXTERN_C
96ac0975 1097
265c1f46 1098 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1099# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1100
1101 /* For internal core Perl use only: the base macro for defining macros like
1102 * isALPHA */
ff7ecfc3 1103# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1104 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1105
f4cdb42c
KW
1106 /* The mask for the _A versions of the macros; it just adds in the bit for
1107 * ASCII. */
1108# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1109
4650c663
KW
1110 /* For internal core Perl use only: the base macro for defining macros like
1111 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1112 * the ASCII bit are present */
b7d90381
KW
1113# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1114 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1115 == _CC_mask_A(classnum)))
f4cdb42c 1116
f4cdb42c 1117# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1118# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1119# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1120# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1121# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1122# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1123# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1124# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1125# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1126# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1127# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1128# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1129# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1130 */
d95f8b6a 1131# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1132# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1133# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1134# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1135
1136 /* continuation character for legal NAME in \N{NAME} */
1137# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1138
1139# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1140# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1141# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1142# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1143# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1144# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1145# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1146# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1147# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1148# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1149
0852beac
KW
1150# ifdef EBCDIC
1151# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1152# endif
1153
f12c0118
KW
1154 /* Participates in a single-character fold with a character above 255 */
1155# 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)))
1156
1157 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1158# 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 1159
4eeeb416 1160# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1161# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1162 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1163# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1164 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1165# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1166 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1167#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1168
1169 /* If we don't have perl.h, we are compiling a utility program. Below we
1170 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1171 * to it. Most are coded based on first principles. These are written to
74665a89 1172 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
0852beac
KW
1173# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1174# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1175# define isSPACE_A(c) (isBLANK_A(c) \
1176 || (c) == '\n' \
1177 || (c) == '\r' \
1178 || (c) == '\v' \
0852beac 1179 || (c) == '\f')
74665a89
KW
1180 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1181 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1182 * performance (if optimization isn't on), and allow the same code to be
1183 * used for both platform types */
1184# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1185 && ( (c) <= 'i' \
1186 || ((c) >= 'j' && (c) <= 'r') \
1187 || (c) >= 's'))
1188# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1189 && ( (c) <= 'I' \
1190 || ((c) >= 'J' && (c) <= 'R') \
1191 || (c) >= 'S'))
a4d7a999
KW
1192# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1193# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1194# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1195# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
74665a89
KW
1196# define isXDIGIT_A(c) (isDIGIT_A(c) \
1197 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1198 || ((c) <= 'F' && (c) >= 'A'))
74665a89
KW
1199# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1200 || (c) == '#' || (c) == '$' || (c) == '%' \
1201 || (c) == '&' || (c) == '\'' || (c) == '(' \
1202 || (c) == ')' || (c) == '*' || (c) == '+' \
1203 || (c) == ',' || (c) == '.' || (c) == '/' \
1204 || (c) == ':' || (c) == ';' || (c) == '<' \
1205 || (c) == '=' || (c) == '>' || (c) == '?' \
1206 || (c) == '@' || (c) == '[' || (c) == '\\' \
1207 || (c) == ']' || (c) == '^' || (c) == '_' \
1208 || (c) == '`' || (c) == '{' || (c) == '|' \
1209 || (c) == '}' || (c) == '~')
1210# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1211# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1212
0852beac 1213# ifdef EBCDIC
74665a89
KW
1214 /* The below is accurate for the 3 EBCDIC code pages traditionally
1215 * supported by perl. The only difference between them in the controls
1216 * is the position of \n, and that is represented symbolically below */
1217# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1218 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1219 || (c) == '\t' || (c) == '\v' \
1220 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1221 || (c) == 7 /* U+7F DEL */ \
1222 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1223 /* DLE, DC[1-3] */ \
1224 || (c) == 0x18 /* U+18 CAN */ \
1225 || (c) == 0x19 /* U+19 EOM */ \
1226 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1227 || (c) == 0x26 /* U+17 ETB */ \
1228 || (c) == 0x27 /* U+1B ESC */ \
1229 || (c) == 0x2D /* U+05 ENQ */ \
1230 || (c) == 0x2E /* U+06 ACK */ \
1231 || (c) == 0x32 /* U+16 SYN */ \
1232 || (c) == 0x37 /* U+04 EOT */ \
1233 || (c) == 0x3C /* U+14 DC4 */ \
1234 || (c) == 0x3D /* U+15 NAK */ \
1235 || (c) == 0x3F)/* U+1A SUB */
0852beac 1236# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1237# else /* isASCII is already defined for ASCII platforms, so can use that to
1238 define isCNTRL */
1239# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1240# endif
1241
3ffc8c70 1242 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1243 * during debugging, and it seems best to keep them. We may be called
1244 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1245 * do anything anyway, so make it not a problem */
1246# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1247# define NATIVE_TO_LATIN1(ch) (ch)
1248# endif
3ded5eb0
KW
1249# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1250# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1251# define isBLANK_L1(c) (isBLANK_A(c) \
1252 || (FITS_IN_8_BITS(c) \
1253 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1254# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1255# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1256# define isLOWER_L1(c) (isLOWER_A(c) \
1257 || (FITS_IN_8_BITS(c) \
1258 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1259 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1260 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1261 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1262 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1263# define isPRINT_L1(c) (isPRINT_A(c) \
1264 || (FITS_IN_8_BITS(c) \
1265 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1266# define isPUNCT_L1(c) (isPUNCT_A(c) \
1267 || (FITS_IN_8_BITS(c) \
1268 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1269 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1270 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1271 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1272 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1273 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1274 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1275# define isSPACE_L1(c) (isSPACE_A(c) \
1276 || (FITS_IN_8_BITS(c) \
1277 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1278 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1279# define isUPPER_L1(c) (isUPPER_A(c) \
1280 || (FITS_IN_8_BITS(c) \
1281 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1282 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1283 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1284# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1285# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1286# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1287 || isBLANK_L1(c) \
1288 || (c) == '-' \
1289 || (c) == '(' \
1290 || (c) == ')')
1291 /* The following are not fully accurate in the above-ASCII range. I (khw)
1292 * don't think it's necessary to be so for the purposes where this gets
1293 * compiled */
1294# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1295# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1296
1297 /* And these aren't accurate at all. They are useful only for above
1298 * Latin1, which utilities and bootstrapping don't deal with */
1299# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1300# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1301# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1302
1303 /* Many of the macros later in this file are defined in terms of these. By
1304 * implementing them with a function, which converts the class number into
1305 * a call to the desired macro, all of the later ones work. However, that
1306 * function won't be actually defined when building a utility program (no
1307 * perl.h), and so a compiler error will be generated if one is attempted
1308 * to be used. And the above-Latin1 code points require Unicode tables to
1309 * be present, something unlikely to be the case when bootstrapping */
1310# define _generic_isCC(c, classnum) \
1311 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1312# define _generic_isCC_A(c, classnum) \
1313 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1314#endif /* End of no perl.h H_PERL */
8a58bdcf 1315
e66b99e9
KW
1316#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1317#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1318#define isASCII_A(c) isASCII(c)
1319#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1320#define isBLANK(c) isBLANK_A(c)
1321#define isCNTRL(c) isCNTRL_A(c)
1322#define isDIGIT(c) isDIGIT_A(c)
1323#define isGRAPH(c) isGRAPH_A(c)
1324#define isIDFIRST(c) isIDFIRST_A(c)
1325#define isLOWER(c) isLOWER_A(c)
1326#define isPRINT(c) isPRINT_A(c)
779cf272 1327#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1328#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1329#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1330#define isPUNCT(c) isPUNCT_A(c)
1331#define isSPACE(c) isSPACE_A(c)
1332#define isUPPER(c) isUPPER_A(c)
1333#define isWORDCHAR(c) isWORDCHAR_A(c)
1334#define isXDIGIT(c) isXDIGIT_A(c)
1335
1336/* ASCII casing. These could also be written as
1337 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1338 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1339 which uses table lookup and mask instead of subtraction. (This would
1340 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1341#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1342#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1343
25200305
KW
1344/* In the ASCII range, these are equivalent to what they're here defined to be.
1345 * But by creating these definitions, other code doesn't have to be aware of
1346 * this detail */
1347#define toFOLD(c) toLOWER(c)
25200305
KW
1348#define toTITLE(c) toUPPER(c)
1349
c753c8d3
KW
1350#define toLOWER_A(c) toLOWER(c)
1351#define toUPPER_A(c) toUPPER(c)
25200305
KW
1352#define toFOLD_A(c) toFOLD(c)
1353#define toTITLE_A(c) toTITLE(c)
1a0901db 1354
4650c663 1355/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1356#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1357 ? (c) \
f4cd282c 1358 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1359#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1360
1a0901db 1361/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1362 * all mapped to one of them, and these need special handling; returns the
1363 * input itself if is out-of-range */
b2bf251f 1364#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1365 ? (c) \
f4cd282c 1366 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1367#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1368
beab9ebe
KW
1369/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1370
1371/* For internal core Perl use only: the base macro for defining macros like
1372 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1373 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1374 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1375 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1376 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1377 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1378 * code points under locale rules, use the macros based on _generic_LC_uvchr
1379 * instead of this. */
beab9ebe
KW
1380#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1381 (! FITS_IN_8_BITS(c) \
1382 ? 0 \
31f05a37
KW
1383 : IN_UTF8_CTYPE_LOCALE \
1384 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1385 : cBOOL(non_utf8))
1386
1387/* For internal core Perl use only: a helper macro for defining macros like
1388 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1389 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1390 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1391 * _generic_LC_base for more info */
1392#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1393 _generic_LC_base(c,utf8_locale_classnum, \
1394 non_utf8_func( (_LC_CAST) (c)))
1395
1396/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1397 * 'c' is the platform's native underscore character */
1398#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1399 _generic_LC_base(c, utf8_locale_classnum, \
1400 (non_utf8_func( (_LC_CAST) (c)) \
1401 || (char)(c) == '_'))
1402
1403/* These next three are also for internal core Perl use only: case-change
1404 * helper macros */
1405#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1406 ? (c) \
31f05a37
KW
1407 : (IN_UTF8_CTYPE_LOCALE) \
1408 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1409 : (cast)function((cast)(c)))
beab9ebe 1410
31f05a37
KW
1411/* Note that the result can be larger than a byte in a UTF-8 locale. It
1412 * returns a single value, so can't adequately return the upper case of LATIN
1413 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1414 * values "SS"); instead it asserts against that under DEBUGGING, and
1415 * otherwise returns its input */
beab9ebe
KW
1416#define _generic_toUPPER_LC(c, function, cast) \
1417 (! FITS_IN_8_BITS(c) \
1418 ? (c) \
31f05a37 1419 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1420 ? (cast)function((cast)(c)) \
31f05a37
KW
1421 : ((((U8)(c)) == MICRO_SIGN) \
1422 ? GREEK_CAPITAL_LETTER_MU \
1423 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1424 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1425 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1426 ? (__ASSERT_(0) (c)) \
1427 : PL_mod_latin1_uc[ (U8) (c) ])))))
1428
1429/* Note that the result can be larger than a byte in a UTF-8 locale. It
1430 * returns a single value, so can't adequately return the fold case of LATIN
1431 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1432 * values "ss"); instead it asserts against that under DEBUGGING, and
1433 * otherwise returns its input */
beab9ebe 1434#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1435 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1436 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1437 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1438 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1439 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1440
84061b6a 1441/* Use the libc versions for these if available. */
f05550c0 1442#if defined(HAS_ISASCII)
84061b6a
KW
1443# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1444#else
1445# define isASCII_LC(c) isASCII(c)
1446#endif
1447
f05550c0 1448#if defined(HAS_ISBLANK)
84061b6a 1449# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1450#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1451# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1452#endif
1453
f05550c0 1454#define _LC_CAST U8
bbce6d69 1455
f05550c0 1456#ifdef WIN32
375f5f06
KW
1457 /* The Windows functions don't bother to follow the POSIX standard, which
1458 * for example says that something can't both be a printable and a control.
1459 * But Windows treats the \t control as a printable, and does such things
1460 * as making superscripts into both digits and punctuation. This tames
1461 * these flaws by assuming that the definitions of both controls and space
1462 * are correct, and then making sure that other definitions don't have
1463 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1464 * Not all possible weirdnesses are checked for, just the ones that were
1465 * detected on actual Microsoft code pages */
1466
b7d90381
KW
1467# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1468# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1469
1470# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1471 && isALPHANUMERIC_LC(c))
1472# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1473 ! isPUNCT_LC(c))
1474# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1475 isALPHANUMERIC_LC(c))
1476# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1477# define isIDFIRST_LC(c) (((c) == '_') \
1478 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1479# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1480# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1481# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1482# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1483# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1484# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1485 && isALPHANUMERIC_LC(c))
f05550c0
BF
1486
1487# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1488# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1489# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1490
1491#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1492 /* For most other platforms */
beab9ebe 1493
f05550c0
BF
1494# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1495# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1496# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1497# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1498# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1499# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1500# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1501# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1502# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1503# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1504# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1505# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1506# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1507
1508
1509# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1510# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1511# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1512
1513#else /* The final fallback position */
1514
b7d90381
KW
1515# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1516# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1517# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1518# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1519# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1520# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1521# define isLOWER_LC(c) (isascii(c) && islower(c))
1522# define isPRINT_LC(c) (isascii(c) && isprint(c))
1523# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1524# define isSPACE_LC(c) (isascii(c) && isspace(c))
1525# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1526# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1527# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1528
1529# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1530# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1531# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1532
f05550c0 1533#endif
55204971 1534
eba68aa0
KW
1535#define isIDCONT(c) isWORDCHAR(c)
1536#define isIDCONT_A(c) isWORDCHAR_A(c)
1537#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1538#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1539#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1540
4650c663 1541/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1542 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1543 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1544 * classes where there is a macro or function 'above_latin1' that takes the
1545 * single argument 'c' and returns the desired value. These exist for those
1546 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1547 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1548 * for classes where that overhead is faster than a direct lookup.
d0da05db 1549 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1550 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1551 * have duplicate checks here, so could create versions of the macros that
1552 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1553
1554/* Note that all ignore 'use bytes' */
d0da05db 1555#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
cd500f2f
KW
1556 ? _generic_isCC(c, classnum) \
1557 : above_latin1(c))
d0da05db 1558#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
cd500f2f 1559 ? _generic_isCC(c, classnum) \
922e8cb4 1560 : _is_uni_FOO(classnum, c))
d0da05db
KW
1561#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1562#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1563#define isASCII_uvchr(c) isASCII(c)
1564#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1565#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1566#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1567#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1568#define isIDCONT_uvchr(c) _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1569#define isIDFIRST_uvchr(c) _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
1570#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1571#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1572
1573#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1574#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1575#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1576
1577#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1578#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1579#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1580#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1581
1582#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1583#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1584#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1585#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1586
1587/* For backwards compatibility, even though '_uni' should mean official Unicode
1588 * code points, in Perl it means native for those below 256 */
1589#define isALPHA_uni(c) isALPHA_uvchr(c)
1590#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1591#define isASCII_uni(c) isASCII_uvchr(c)
1592#define isBLANK_uni(c) isBLANK_uvchr(c)
1593#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1594#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1595#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1596#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1597#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1598#define isLOWER_uni(c) isLOWER_uvchr(c)
1599#define isPRINT_uni(c) isPRINT_uvchr(c)
1600#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1601#define isSPACE_uni(c) isSPACE_uvchr(c)
1602#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1603#define isUPPER_uni(c) isUPPER_uvchr(c)
1604#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1605#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1606#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1607#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1608#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1609#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1610#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1611
4650c663
KW
1612/* For internal core Perl use only: the base macros for defining macros like
1613 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1614 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1615 * simple class definitions; the other for more complex. These are like
d0da05db 1616 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1617#define _generic_LC_uvchr(latin1, above_latin1, c) \
1618 (c < 256 ? latin1(c) : above_latin1(c))
1619#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1620 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1621
1622#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1623#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1624 _CC_ALPHANUMERIC, c)
b7d90381
KW
1625#define isASCII_LC_uvchr(c) isASCII_LC(c)
1626#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1627 is_HORIZWS_cp_high, c)
feeab5a9 1628#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1629#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1630#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1631#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1632 _is_uni_perl_idcont, c)
b7d90381 1633#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1634 _is_uni_perl_idstart, c)
1635#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1636#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1637#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1638#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1639#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1640 is_XPERLSPACE_cp_high, c)
cd500f2f 1641#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1642#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1643 _CC_WORDCHAR, c)
b7d90381
KW
1644#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1645 is_XDIGIT_cp_high, c)
e712593e 1646
b7d90381 1647#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1648
4650c663
KW
1649/* For internal core Perl use only: the base macros for defining macros like
1650 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1651 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1652 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1653 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1654 * same representation whether utf8 or not. Note that it assumes that the utf8
1655 * has been validated, and ignores 'use bytes' */
cd500f2f 1656#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1657 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1658 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1659 ? _generic_isCC( \
a62b247b 1660 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
aea0c08c 1661 *((p)+1 )), \
94bb8c36 1662 classnum) \
252810d0 1663 : utf8)
b7d90381
KW
1664/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1665 * 'above_latin1' can be a macro */
cd500f2f
KW
1666#define _generic_func_utf8(classnum, above_latin1, p) \
1667 _generic_utf8(classnum, p, above_latin1(p))
4650c663 1668/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1669 * 'above_latin1' parameter */
922e8cb4 1670#define _generic_swash_utf8(classnum, p) \
cd500f2f 1671 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
922e8cb4 1672
cc8ab7c0 1673/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1674 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1675 * class is TRUE for. Hence it can skip the tests for this range.
1676 * 'above_latin1' should include its arguments */
3ded5eb0 1677#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1678 (UTF8_IS_INVARIANT(*(p)) \
1679 ? _generic_isCC(*(p), classnum) \
1680 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1681 ? above_latin1 \
cc8ab7c0
KW
1682 : 0)
1683
84238efa
KW
1684/* NOTE that some of these macros have very similar ones in regcharclass.h.
1685 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1686 * there, differing in name only by an underscore from the one here
1687 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1688 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1689 * points; the regcharclass.h ones are implemented as a series of
1690 * "if-else-if-else ..." */
1691
b7d90381
KW
1692#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1693#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1694#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
509fb054
KW
1695 utf8, the non-utf8 macro works
1696 */
b7d90381 1697#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2
KW
1698
1699#ifdef EBCDIC
1700 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1701 * more efficient macro instead of the more general one */
88794300 1702# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
e8fa43e2
KW
1703#else
1704# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1705#endif
1706
cd500f2f
KW
1707#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1708 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1709#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1710#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1711 _is_utf8_perl_idcont, p)
e5dcd934 1712
c11ff943
KW
1713/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1714 * IDFIRST is an alnum. See
c8362b00 1715 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1716 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1717 * This used to be not the XID version, but we decided to go with the more
1718 * modern Unicode definition */
b7d90381 1719#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1720 _is_utf8_perl_idstart, p)
e5dcd934 1721
b7d90381
KW
1722#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1723#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1724#define isPSXSPC_utf8(p) isSPACE_utf8(p)
1725#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1726#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1727#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1728#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1729#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1730#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
cd500f2f 1731 is_XDIGIT_high(p))
a0ed51b3 1732
7d2b61dd 1733#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1734#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1735#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1736#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1737
4650c663
KW
1738/* For internal core Perl use only: the base macros for defining macros like
1739 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1740 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1741 * 'macro' parameter */
94bb8c36
KW
1742#define _generic_LC_utf8(macro, p, utf8) \
1743 (UTF8_IS_INVARIANT(*(p)) \
1744 ? macro(*(p)) \
1745 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
a62b247b 1746 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
9201a572
KW
1747 : utf8)
1748
cd500f2f
KW
1749#define _generic_LC_swash_utf8(macro, classnum, p) \
1750 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1751#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1752 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1753
b7d90381 1754#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
cd500f2f 1755 _CC_ALPHANUMERIC, p)
b7d90381
KW
1756#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1757#define isASCII_LC_utf8(p) isASCII_LC(*p)
1758#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1759 is_HORIZWS_high, p)
1760#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1761#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1762#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1763#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1764 _is_utf8_perl_idcont, p)
1765#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1766 _is_utf8_perl_idstart, p)
1767#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1768#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1769#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1770#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1771#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1772 is_XPERLSPACE_high, p)
1773#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
cd500f2f
KW
1774#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1775 _CC_WORDCHAR, p)
b7d90381
KW
1776#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1777 is_XDIGIT_high, p)
aaa51d5e 1778
fbc19f27
KW
1779/* Macros for backwards compatibility and for completeness when the ASCII and
1780 * Latin1 values are identical */
b7d90381
KW
1781#define isALPHAU(c) isALPHA_L1(c)
1782#define isDIGIT_L1(c) isDIGIT_A(c)
1783#define isOCTAL(c) isOCTAL_A(c)
1784#define isOCTAL_L1(c) isOCTAL_A(c)
1785#define isXDIGIT_L1(c) isXDIGIT_A(c)
1786#define isALNUM(c) isWORDCHAR(c)
1787#define isALNUMU(c) isWORDCHAR_L1(c)
1788#define isALNUM_LC(c) isWORDCHAR_LC(c)
1789#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 1790#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
1791#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1792#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1793#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1794#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1795#define isALNUMC(c) isALPHANUMERIC(c)
1796#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1797#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 1798#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 1799#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 1800#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1801
2bd1cbf6
KW
1802/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1803 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1804 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1805 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1806 * the outlier from the block that contains the other controls, just like
1807 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1808 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
1809 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1810 * etc. */
2bd1cbf6 1811#ifndef EBCDIC
75763b3a 1812# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 1813#else
75763b3a
KW
1814# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1815 ((isPRINT_A(c)) \
1816 ? (UNLIKELY((c) == '?') \
1817 ? QUESTION_MARK_CTRL \
1818 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1819 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1820 ? '?' \
1821 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 1822#endif
bbce6d69 1823
dea28490
JJ
1824/* Line numbers are unsigned, 32 bits. */
1825typedef U32 line_t;
e5dcd934 1826#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 1827
91152fc1
DG
1828/* Helpful alias for version prescan */
1829#define is_LAX_VERSION(a,b) \
1830 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1831
1832#define is_STRICT_VERSION(a,b) \
1833 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1834
1835#define BADVERSION(a,b,c) \
1836 if (b) { \
1837 *b = c; \
1838 } \
1839 return a;
8c52afec 1840
cb27eebd
KW
1841/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1842 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1843 * which is advanced. The input is validated only by an assert() in DEBUGGING
1844 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1845 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1846#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1847 ? (c) \
1848 : ((c) + 9))))
902008b8
KW
1849#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1850 ? (*(s)++) \
1851 : (*(s)++ + 9))))
95a59cab 1852
cb27eebd
KW
1853/* Converts a character known to represent an octal digit (0-7) to its numeric
1854 * value. The input is validated only by an assert() in DEBUGGING builds. In
1855 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1856#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1857
305b8651
KW
1858/* Efficiently returns a boolean as to if two native characters are equivalent
1859 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1860 * the ALPHA in the name is to remind you of that. This is asserted() in
1861 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1862 * works (on valid input) for both non- and UTF-8-encoded bytes.
1863 *
1864 * When one of the inputs is a compile-time constant and gets folded by the
1865 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1866 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1867 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
1868 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1869 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1870 * just a single 0, in the bit position where the upper- and lowercase differ.
1871 * */
305b8651
KW
1872#define isALPHA_FOLD_EQ(c1, c2) \
1873 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1874 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1875#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1876
8e84507e 1877/*
ccfc67b7
JH
1878=head1 Memory Management
1879
a02a5408 1880=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
1881The XSUB-writer's interface to the C C<malloc> function.
1882
596f7718 1883Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1884
c5008215
JC
1885In 5.9.3, Newx() and friends replace the older New() API, and drops
1886the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 1887themselves. This aid has been superseded by a new build option,
d10b4965 1888PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
1889there for use in XS modules supporting older perls.
1890
a02a5408 1891=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 1892The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 1893cast. See also C<L</Newx>>.
954c1994 1894
596f7718 1895Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1896
a02a5408 1897=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 1898The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 1899memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 1900
596f7718 1901Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1902
954c1994
GS
1903=for apidoc Am|void|Renew|void* ptr|int nitems|type
1904The XSUB-writer's interface to the C C<realloc> function.
1905
596f7718 1906Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1907
954c1994
GS
1908=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1909The XSUB-writer's interface to the C C<realloc> function, with
1910cast.
1911
596f7718 1912Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1913
49b8b560 1914=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
1915The XSUB-writer's interface to the C C<free> function.
1916
596f7718 1917This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 1918
954c1994
GS
1919=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1920The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 1921source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1922C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 1923
e90e2364 1924=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 1925Like C<Move> but returns C<dest>. Useful
72d33970 1926for encouraging compilers to tail-call
e90e2364
NC
1927optimise.
1928
954c1994
GS
1929=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1930The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 1931source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1932C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 1933
e90e2364
NC
1934=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1935
796b6530 1936Like C<Copy> but returns C<dest>. Useful
72d33970 1937for encouraging compilers to tail-call
e90e2364
NC
1938optimise.
1939
954c1994
GS
1940=for apidoc Am|void|Zero|void* dest|int nitems|type
1941
1942The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1943destination, C<nitems> is the number of items, and C<type> is the type.
1944
e90e2364
NC
1945=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1946
72d33970
FC
1947Like C<Zero> but returns dest. Useful
1948for encouraging compilers to tail-call
e90e2364
NC
1949optimise.
1950
da5d8dbb 1951=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 1952This is an architecture-independent macro to copy one structure to another.
954c1994 1953
7e337ee0
JH
1954=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1955
1956Fill up memory with a byte pattern (a byte repeated over and over
1957again) that hopefully catches attempts to access uninitialized memory.
1958
1959=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1960
1961PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1962
1c12ffb4 1963=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
1964
1965PoisonWith(0xEF) for catching access to freed memory.
1966
9965345d
JH
1967=for apidoc Am|void|Poison|void* dest|int nitems|type
1968
7e337ee0 1969PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
1970
1971=cut */
954c1994 1972
561b68a9
SH
1973/* Maintained for backwards-compatibility only. Use newSV() instead. */
1974#ifndef PERL_CORE
ff06c60c 1975#define NEWSV(x,len) newSV(len)
561b68a9 1976#endif
ff06c60c 1977
19a94d75
JH
1978#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1979
e6bdf523 1980
27d5b266 1981#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
1982
1983/* This expression will be constant-folded at compile time. It checks
1984 * whether or not the type of the count n is so small (e.g. U8 or U16, or
1985 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
1986 * As well as avoiding the need for a run-time check in some cases, it's
1987 * designed to avoid compiler warnings like:
1988 * comparison is always false due to limited range of data type
73e8ff00
DM
1989 * It's mathematically equivalent to
1990 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
1991 */
1992
1993# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
004073ba 1994 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
e6bdf523 1995
88f9f128 1996/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
1997 * compiler warnings. We *want* to write the expression as
1998 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
1999 * (for some compile-time constant C), but even when the LHS
2000 * constant-folds to false at compile-time, g++ insists on emitting
2001 * warnings about the RHS (e.g. "comparison is always false"), so instead
2002 * we write it as
e6bdf523 2003 *
d98e5cde 2004 * (cond ? n : X) > C
88f9f128 2005 *
d98e5cde
DM
2006 * where X is a constant with X > C always false. Choosing a value for X
2007 * is tricky. If 0, some compilers will complain about 0 > C always being
2008 * false; if 1, Coverity complains when n happens to be the constant value
2009 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2010 * for X and hope that nothing else whines.
e6bdf523
DM
2011 */
2012
2013# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2014 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2015 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2016
2017# define MEM_WRAP_CHECK(n,t) \
2018 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2019 && (croak_memory_wrap(),0))
2020
2021# define MEM_WRAP_CHECK_1(n,t,a) \
2022 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2023 && (Perl_croak_nocontext("%s",(a)),0))
2024
8b44ba4c 2025#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2026
d1decf2b 2027#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
2028#else
2029
410319be
NC
2030#define MEM_WRAP_CHECK(n,t)
2031#define MEM_WRAP_CHECK_1(n,t,a)
2032#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
2033#define MEM_WRAP_CHECK_(n,t)
2034
6fff79ce 2035#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 2036
1936d2a7 2037#endif
8b44ba4c 2038
fe4f188c 2039#ifdef PERL_MEM_LOG
46c6c7e2 2040/*
9f653bb5 2041 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2042 * go through functions, which are handy for debugging breakpoints, but
2043 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2044 * line number, and C function name if available) passed in. This info can
2045 * then be used for logging the calls, for which one gets a sample
73d1d973 2046 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2047 *
46c6c7e2 2048 * Known problems:
94e892a6 2049 * - not all memory allocs get logged, only those
46c6c7e2 2050 * that go through Newx() and derivatives (while all
94e892a6 2051 * Safefrees do get logged)
46c6c7e2
JH
2052 * - __FILE__ and __LINE__ do not work everywhere
2053 * - __func__ or __FUNCTION__ even less so
2054 * - I think more goes on after the perlio frees but
2055 * the thing is that STDERR gets closed (as do all
2056 * the file descriptors)
2057 * - no deeper calling stack than the caller of the Newx()
2058 * or the kind, but do I look like a C reflection/introspection
2059 * utility to you?
2060 * - the function prototypes for the logging functions
2061 * probably should maybe be somewhere else than handy.h
2062 * - one could consider inlining (macrofying) the logging
2063 * for speed, but I am too lazy
2064 * - one could imagine recording the allocations in a hash,
2065 * (keyed by the allocation address?), and maintain that
2066 * through reallocs and frees, but how to do that without
2067 * any News() happening...?
73d1d973 2068 * - lots of -Ddefines to get useful/controllable output
b953482e 2069 * - lots of ENV reads
46c6c7e2
JH
2070 */
2071
0b0ab801 2072# ifdef PERL_CORE
73d1d973 2073# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2074enum mem_log_type {
2075 MLT_ALLOC,
2076 MLT_REALLOC,
d7a2c63c
MHM
2077 MLT_FREE,
2078 MLT_NEW_SV,
2079 MLT_DEL_SV
0b0ab801
MHM
2080};
2081# endif
12754f92 2082# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2083void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2084void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2085# endif
0b0ab801
MHM
2086# endif
2087
fe4f188c
JH
2088#endif
2089
2090#ifdef PERL_MEM_LOG
d1401ee9
MHM
2091#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2092#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2093#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2094#endif
2095
2096#ifndef MEM_LOG_ALLOC
2097#define MEM_LOG_ALLOC(n,t,a) (a)
2098#endif
2099#ifndef MEM_LOG_REALLOC
2100#define MEM_LOG_REALLOC(n,t,v,a) (a)
2101#endif
2102#ifndef MEM_LOG_FREE
2103#define MEM_LOG_FREE(a) (a)
2104#endif
2105
d1401ee9
MHM
2106#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2107#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2108#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2109
2110#ifndef PERL_CORE
a02a5408
JC
2111/* pre 5.9.x compatibility */
2112#define New(x,v,n,t) Newx(v,n,t)
2113#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2114#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2115#endif
a02a5408 2116
ff68c719 2117#define Renew(v,n,t) \
d1401ee9 2118 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2119#define Renewc(v,n,t,c) \
d1401ee9 2120 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2121
2122#ifdef PERL_POISON
2123#define Safefree(d) \
06c0cc96 2124 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2125#else
fe4f188c 2126#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2127#endif
55497cff 2128
8b44ba4c
NC
2129#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2130#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2131#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2132
8b44ba4c
NC
2133#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2134#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2135#ifdef HAS_MEMSET
8b44ba4c 2136#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2137#else
8b44ba4c
NC
2138/* Using bzero(), which returns void. */
2139#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2140#endif
2141
7e337ee0
JH
2142#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2143#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2144#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2145#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2146
caa674f3
DD
2147#ifdef PERL_POISON
2148# define PERL_POISON_EXPR(x) x
2149#else
2150# define PERL_POISON_EXPR(x)
2151#endif
2152
2304df62 2153#ifdef USE_STRUCT_COPY
ff68c719 2154#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2155#else
2156#define StructCopy(s,d,t) Copy(s,d,1,t)
2157#endif
2cc61e15 2158
c3caa5c3
JH
2159/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2160 * want your zero-based indices to be less than but not equal to).
2161 *
2162 * C_ARRAY_END is one past the last: half-open/half-closed range,
2163 * not last-inclusive range. */
622913ab 2164#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2165#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2166
2cc61e15
DD
2167#ifdef NEED_VA_COPY
2168# ifdef va_copy
2169# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2170# else
a1866d1b
JH
2171# if defined(__va_copy)
2172# define Perl_va_copy(s, d) __va_copy(d, s)
2173# else
2174# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2175# endif
2cc61e15
DD
2176# endif
2177#endif
2178
472d47bc
SB
2179/* convenience debug macros */
2180#ifdef USE_ITHREADS
2181#define pTHX_FORMAT "Perl interpreter: 0x%p"
2182#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2183#define pTHX_VALUE_ (void *)my_perl,
2184#define pTHX_VALUE (void *)my_perl
2185#define pTHX__VALUE_ ,(void *)my_perl,
2186#define pTHX__VALUE ,(void *)my_perl
472d47bc 2187#else
3609ea0d 2188#define pTHX_FORMAT
472d47bc 2189#define pTHX__FORMAT
3609ea0d 2190#define pTHX_VALUE_
472d47bc 2191#define pTHX_VALUE
3609ea0d 2192#define pTHX__VALUE_
472d47bc
SB
2193#define pTHX__VALUE
2194#endif /* USE_ITHREADS */
3609ea0d 2195
2acdbac1
NC
2196/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2197 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2198 CPAN::Unpack show any users outside the core. */
2199#ifdef PERL_CORE
b7d90381
KW
2200# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2201 "Use of " s " is deprecated")
2acdbac1
NC
2202#endif
2203
dfff4baf
BF
2204/* Internal macros to deal with gids and uids */
2205#ifdef PERL_CORE
2206
2207# if Uid_t_size > IVSIZE
2208# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2209# define SvUID(sv) SvNV(sv)
2210# else
2211# if Uid_t_sign <= 0
2212# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2213# define SvUID(sv) SvIV(sv)
2214# else
2215# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2216# define SvUID(sv) SvUV(sv)
2217# endif
2218# endif /* Uid_t_size */
2219
2220# if Gid_t_size > IVSIZE
2221# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2222# define SvGID(sv) SvNV(sv)
2223# else
2224# if Gid_t_sign <= 0
2225# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2226# define SvGID(sv) SvIV(sv)
2227# else
2228# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2229# define SvGID(sv) SvUV(sv)
2230# endif
2231# endif /* Gid_t_size */
2232
2233#endif
2234
9d745869
KW
2235#endif /* HANDY_H */
2236
e9a8c099 2237/*
14d04a33 2238 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2239 */