This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
handy.h: add _memEQs() and _memNEs() for use with constant strings
[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
62946e08 484
8d063cd8
LW
485#define strNE(s1,s2) (strcmp(s1,s2))
486#define strEQ(s1,s2) (!strcmp(s1,s2))
487#define strLT(s1,s2) (strcmp(s1,s2) < 0)
488#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
489#define strGT(s1,s2) (strcmp(s1,s2) > 0)
490#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 491
8d063cd8
LW
492#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
493#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
378cc40b 494
62946e08
YO
495#define strNEs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1))
496#define strEQs(s1,s2) (!strncmp(s1,"" s2 "", sizeof(s2)-1))
497
36477c24
PP
498#ifdef HAS_MEMCMP
499# define memNE(s1,s2,l) (memcmp(s1,s2,l))
500# define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
501#else
502# define memNE(s1,s2,l) (bcmp(s1,s2,l))
503# define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
504#endif
505
085b7534 506/* memEQ and memNE where second comparand is a string constant */
568a785a 507#define memEQs(s1, l, s2) \
777fa2cb 508 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
568a785a
NC
509#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
510
777fa2cb
YO
511/* memEQ and memNE where second comparand is a string constant
512 * and we can assume the length of s1 is at least that of the string */
513#define _memEQs(s1, s2) \
514 (memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
515#define _memNEs(s1, s2) (memNE((s1),("" s2 ""),(sizeof(s2)-1)))
516
062b6850
KW
517#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
518#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
519#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
520#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
521
bbce6d69
PP
522/*
523 * Character classes.
524 *
525 * Unfortunately, the introduction of locales means that we
526 * can't trust isupper(), etc. to tell the truth. And when
527 * it comes to /\w+/ with tainting enabled, we *must* be able
528 * to trust our character classes.
529 *
530 * Therefore, the default tests in the text of Perl will be
531 * independent of locale. Any code that wants to depend on
532 * the current locale will use the tests that begin with "lc".
533 */
534
2304df62
AD
535#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
536# ifndef CTYPE256
537# define CTYPE256
538# endif
539#endif
540
954c1994 541/*
ccfc67b7 542
dcccc8ff 543=head1 Character classification
243effed
KW
544This section is about functions (really macros) that classify characters
545into types, such as punctuation versus alphabetic, etc. Most of these are
546analogous to regular expression character classes. (See
547L<perlrecharclass/POSIX Character Classes>.) There are several variants for
548each class. (Not all macros have all variants; each item below lists the
549ones valid for it.) None are affected by C<use bytes>, and only the ones
550with C<LC> in the name are affected by the current locale.
551
552The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
553C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
554represented by that octet is (or on non-ASCII platforms, corresponds to) an
555ASCII character in the named class based on platform, Unicode, and Perl rules.
556If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 557
6aff1f14
KW
558Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
559with no suffix C<"_A">.
4b9734bf
KW
560
561Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
562onto the platform. That is, the code points that are ASCII are unaffected,
563since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
564as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
565true when called with the code point 0xDF, which is a word character in both
4650c663 566ASCII and EBCDIC (though it represents different characters in each).
243effed 567
d0da05db 568Variant C<isFOO_uvchr> is like the C<isFOO_L1> variant, but accepts any UV code
243effed
KW
569point as input. If the code point is larger than 255, Unicode rules are used
570to determine if it is in the character class. For example,
d0da05db 571C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 572WITH MACRON in Unicode, and is a word character.
243effed 573
d0da05db 574Variant C<isFOO_utf8> is like C<isFOO_uvchr>, but the input is a pointer to a
dc6b0978
KW
575(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
576possibly containing embedded C<NUL> characters). The classification of just
577the first (possibly multi-byte) character in the string is tested.
243effed 578
31f05a37
KW
579Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
580result is based on the current locale, which is what C<LC> in the name stands
581for. If Perl can determine that the current locale is a UTF-8 locale, it uses
582the published Unicode rules; otherwise, it uses the C library function that
583gives the named classification. For example, C<isDIGIT_LC()> when not in a
584UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
585returned if the input won't fit into an octet. On some platforms where the C
586library function is known to be defective, Perl changes its result to follow
587the POSIX standard's rules.
243effed
KW
588
589Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
590returns the same as C<isFOO_LC> for input code points less than 256, and
591returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
592
b7d90381 593Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer
dc6b0978
KW
594to a (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
595possibly containing embedded C<NUL> characters). The classification of just
596the first (possibly multi-byte) character in the string is tested.
ccfc67b7 597
954c1994 598=for apidoc Am|bool|isALPHA|char ch
2787a470 599Returns a boolean indicating whether the specified character is an
6aff1f14 600alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
601See the L<top of this section|/Character classification> for an explanation of
602variants
d0da05db 603C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>, C<isALPHA_LC>,
243effed 604C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
8a58bdcf 605
15861f94
KW
606=for apidoc Am|bool|isALPHANUMERIC|char ch
607Returns a boolean indicating whether the specified character is a either an
6aff1f14 608alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
609See the L<top of this section|/Character classification> for an explanation of
610variants
d0da05db 611C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
e5ad6aba 612C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
15861f94
KW
613C<isALPHANUMERIC_LC_utf8>.
614
8a58bdcf
KW
615=for apidoc Am|bool|isASCII|char ch
616Returns a boolean indicating whether the specified character is one of the 128
243effed 617characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 618On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
619character corresponds to an ASCII character. Variants C<isASCII_A()> and
620C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
621See the L<top of this section|/Character classification> for an explanation of
622variants
d0da05db 623C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
243effed
KW
624C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
625library routine C<isascii()>. In these cases, the variants whose names contain
626C<LC> are the same as the corresponding ones without.
627
d98532ea
KW
628Also note, that because all ASCII characters are UTF-8 invariant (meaning they
629have the exact same representation (always a single byte) whether encoded in
630UTF-8 or not), C<isASCII> will give the correct results when called with any
631byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
632work properly on any string encoded or not in UTF-8.
633
243effed
KW
634=for apidoc Am|bool|isBLANK|char ch
635Returns a boolean indicating whether the specified character is a
6aff1f14 636character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
637See the L<top of this section|/Character classification> for an explanation of
638variants
d0da05db 639C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>, C<isBLANK_LC>,
243effed
KW
640C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
641platforms do not have the C library routine C<isblank()>. In these cases, the
642variants whose names contain C<LC> are the same as the corresponding ones
643without.
644
645=for apidoc Am|bool|isCNTRL|char ch
646Returns a boolean indicating whether the specified character is a
6aff1f14 647control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
648See the L<top of this section|/Character classification> for an explanation of
649variants
d0da05db 650C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
6aff1f14
KW
651C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
652On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 653
954c1994 654=for apidoc Am|bool|isDIGIT|char ch
2787a470 655Returns a boolean indicating whether the specified character is a
6aff1f14 656digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 657Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
658See the L<top of this section|/Character classification> for an explanation of
659variants
d0da05db 660C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
243effed
KW
661C<isDIGIT_LC_utf8>.
662
663=for apidoc Am|bool|isGRAPH|char ch
664Returns a boolean indicating whether the specified character is a
6aff1f14 665graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff
KW
666See the L<top of this section|/Character classification> for an explanation of
667variants
d0da05db 668C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
243effed 669C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
954c1994 670
0c82b6df 671=for apidoc Am|bool|isLOWER|char ch
2787a470 672Returns a boolean indicating whether the specified character is a
6aff1f14 673lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
674See the L<top of this section|/Character classification> for an explanation of
675variants
d0da05db 676C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>, C<isLOWER_LC>,
243effed 677C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
0c82b6df 678
c99e91e9 679=for apidoc Am|bool|isOCTAL|char ch
2787a470 680Returns a boolean indicating whether the specified character is an
6aff1f14 681octal digit, [0-7].
243effed
KW
682The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
683C<isOCTAL>.
684
685=for apidoc Am|bool|isPUNCT|char ch
686Returns a boolean indicating whether the specified character is a
6aff1f14
KW
687punctuation character, analogous to C<m/[[:punct:]]/>.
688Note that the definition of what is punctuation isn't as
243effed
KW
689straightforward as one might desire. See L<perlrecharclass/POSIX Character
690Classes> for details.
dcccc8ff
KW
691See the L<top of this section|/Character classification> for an explanation of
692variants
d0da05db 693C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
243effed 694C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
c99e91e9 695
0c82b6df 696=for apidoc Am|bool|isSPACE|char ch
2787a470 697Returns a boolean indicating whether the specified character is a
6aff1f14 698whitespace character. This is analogous
398d098a 699to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 700this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
701locale forms of this macro (the ones with C<LC> in their names) matched
702precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
703in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
704(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
705See the L<top of this section|/Character classification> for an explanation of
706variants
d0da05db 707C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>, C<isSPACE_LC>,
243effed 708C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
0c82b6df 709
398d098a
KW
710=for apidoc Am|bool|isPSXSPC|char ch
711(short for Posix Space)
779cf272
KW
712Starting in 5.18, this is identical in all its forms to the
713corresponding C<isSPACE()> macros.
398d098a
KW
714The locale forms of this macro are identical to their corresponding
715C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
716non-locale forms differ from their C<isSPACE()> forms only in that the
717C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
718Otherwise they are identical. Thus this macro is analogous to what
719C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 720See the L<top of this section|/Character classification> for an explanation of
d0da05db 721variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
b7d90381 722C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
398d098a 723
954c1994 724=for apidoc Am|bool|isUPPER|char ch
2787a470 725Returns a boolean indicating whether the specified character is an
6aff1f14 726uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff
KW
727See the L<top of this section|/Character classification> for an explanation of
728variants
d0da05db 729C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>, C<isUPPER_LC>,
243effed 730C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
954c1994 731
243effed 732=for apidoc Am|bool|isPRINT|char ch
8eea39dd 733Returns a boolean indicating whether the specified character is a
6aff1f14 734printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
735See the L<top of this section|/Character classification> for an explanation of
736variants
d0da05db 737C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>, C<isPRINT_LC>,
243effed
KW
738C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
739
740=for apidoc Am|bool|isWORDCHAR|char ch
741Returns a boolean indicating whether the specified character is a character
742that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
743in a regular expression. A word character is an alphabetic character, a
744decimal digit, a connecting punctuation character (such as an underscore), or
745a "mark" character that attaches to one of those (like some sort of accent).
746C<isALNUM()> is a synonym provided for backward compatibility, even though a
747word character includes more than the standard C language meaning of
748alphanumeric.
dcccc8ff
KW
749See the L<top of this section|/Character classification> for an explanation of
750variants
d0da05db 751C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and C<isWORDCHAR_utf8>.
1a83413c
KW
752C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
753described there, but additionally include the platform's native underscore.
8a58bdcf
KW
754
755=for apidoc Am|bool|isXDIGIT|char ch
756Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
757digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
758and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
759See the L<top of this section|/Character classification> for an explanation of
760variants
d0da05db 761C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
243effed
KW
762C<isXDIGIT_LC_utf8>.
763
3c3ecf18
KW
764=for apidoc Am|bool|isIDFIRST|char ch
765Returns a boolean indicating whether the specified character can be the first
766character of an identifier. This is very close to, but not quite the same as
767the official Unicode property C<XID_Start>. The difference is that this
768returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
769See the L<top of this section|/Character classification> for an explanation of
770variants
d0da05db 771C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
3c3ecf18
KW
772C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
773
774=for apidoc Am|bool|isIDCONT|char ch
775Returns a boolean indicating whether the specified character can be the
776second or succeeding character of an identifier. This is very close to, but
777not quite the same as the official Unicode property C<XID_Continue>. The
778difference is that this returns true only if the input character also matches
dcccc8ff
KW
779L</isWORDCHAR>. See the L<top of this section|/Character classification> for
780an
d0da05db 781explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
3c3ecf18
KW
782C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
783C<isIDCONT_LC_utf8>.
784
243effed 785=head1 Miscellaneous Functions
8eea39dd 786
95a59cab 787=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 788Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
789Behaviour is only well defined when isXDIGIT(*str) is true.
790
e7c1e6c1
KW
791=head1 Character case changing
792
1f607577
KW
793=for apidoc Am|U8|toUPPER|U8 ch
794Converts the specified character to uppercase. If the input is anything but an
795ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 796C<toUPPER_A> is equivalent.
954c1994 797
d0da05db
KW
798=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
799Converts the code point C<cp> to its uppercase version, and
800stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
801point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
802that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
803bytes since the uppercase version may be longer than the original character.
804
805The first code point of the uppercased version is returned
806(but note, as explained just above, that there may be more.)
807
808=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
809Converts the UTF-8 encoded character at C<p> to its uppercase version, and
810stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
811that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
812bytes since the uppercase version may be longer than the original character.
813
814The first code point of the uppercased version is returned
815(but note, as explained just above, that there may be more.)
816
817The input character at C<p> is assumed to be well-formed.
818
25200305
KW
819=for apidoc Am|U8|toFOLD|U8 ch
820Converts the specified character to foldcase. If the input is anything but an
821ASCII uppercase character, that input character itself is returned. Variant
822C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 823Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 824
d0da05db
KW
825=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
826Converts the code point C<cp> to its foldcase version, and
827stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
828point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
829that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
830bytes since the foldcase version may be longer than the original character.
831
832The first code point of the foldcased version is returned
833(but note, as explained just above, that there may be more.)
834
835=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
836Converts the UTF-8 encoded character at C<p> to its foldcase version, and
837stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
838that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
839bytes since the foldcase version may be longer than the original character.
840
841The first code point of the foldcased version is returned
842(but note, as explained just above, that there may be more.)
843
844The input character at C<p> is assumed to be well-formed.
845
846=for apidoc Am|U8|toLOWER|U8 ch
847Converts the specified character to lowercase. If the input is anything but an
848ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 849C<toLOWER_A> is equivalent.
954c1994 850
1f607577 851=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
852Converts the specified Latin1 character to lowercase. The results are
853undefined if the input doesn't fit in a byte.
1f607577
KW
854
855=for apidoc Am|U8|toLOWER_LC|U8 ch
856Converts the specified character to lowercase using the current locale's rules,
857if possible; otherwise returns the input character itself.
858
d0da05db
KW
859=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
860Converts the code point C<cp> to its lowercase version, and
861stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
862point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
863that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
864bytes since the lowercase version may be longer than the original character.
865
866The first code point of the lowercased version is returned
867(but note, as explained just above, that there may be more.)
868
869=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
870Converts the UTF-8 encoded character at C<p> to its lowercase version, and
871stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
872that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
873bytes since the lowercase version may be longer than the original character.
874
875The first code point of the lowercased version is returned
876(but note, as explained just above, that there may be more.)
877
878The input character at C<p> is assumed to be well-formed.
879
25200305
KW
880=for apidoc Am|U8|toTITLE|U8 ch
881Converts the specified character to titlecase. If the input is anything but an
882ASCII lowercase character, that input character itself is returned. Variant
b7d90381 883C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 884range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 885not a concept used in locale handling, so there is no functionality for that.)
25200305 886
d0da05db
KW
887=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
888Converts the code point C<cp> to its titlecase version, and
889stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
890point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
891that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
892bytes since the titlecase version may be longer than the original character.
893
894The first code point of the titlecased version is returned
895(but note, as explained just above, that there may be more.)
896
897=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
898Converts the UTF-8 encoded character at C<p> to its titlecase version, and
899stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
900that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
901bytes since the titlecase version may be longer than the original character.
902
903The first code point of the titlecased version is returned
904(but note, as explained just above, that there may be more.)
905
906The input character at C<p> is assumed to be well-formed.
907
954c1994 908=cut
353c9b6f 909
d0da05db 910XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
25200305 911really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 912
8a58bdcf 913Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
914patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
915
954c1994
GS
916*/
917
7c062697
KW
918/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
919 * is known only in the perl core, and this macro can be called from outside
920 * that */
921#ifdef HAS_QUAD
922# define WIDEST_UTYPE U64TYPE
923#else
924# define WIDEST_UTYPE U32
925#endif
926
3912bc88
KW
927/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
928 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
929 * bits of information are lost even on a 64-bit machine, but to get the
930 * compiler to optimize it out if possible. This is because Configure makes
931 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
932 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
933 *
934 * For Coverity, be always true, because otherwise Coverity thinks
935 * it finds several expressions that are always true, independent
936 * of operands. Well, they are, but that is kind of the point.
220c71bf 937 */
7e75d1a1 938#ifndef __COVERITY__
6c5b02ac
KW
939 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
940 * pointer) */
941#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 942 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
943#else
944#define FITS_IN_8_BITS(c) (1)
945#endif
cf301eb7 946
41f43cc2 947#ifdef EBCDIC
b6340bd0 948# ifndef _ALL_SOURCE
0852beac
KW
949 /* The native libc isascii() et.al. functions return the wrong results
950 * on at least z/OS unless this is defined. */
b6340bd0
KW
951# error _ALL_SOURCE should probably be defined
952# endif
41f43cc2 953#else
0852beac
KW
954 /* There is a simple definition of ASCII for ASCII platforms. But the
955 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 956 * other macros below.
3f3c579d
KW
957 *
958 * The cast here is used instead of '(c) >= 0', because some compilers emit
959 * a warning that that test is always true when the parameter is an
960 * unsigned type. khw supposes that it could be written as
961 * && ((c) == '\0' || (c) > 0)
962 * to avoid the message, but the cast will likely avoid extra branches even
963 * with stupid compilers.
964 *
965 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
966 * a pointer) */
9c903d59 967# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
968#endif
969
38694112
KW
970/* Take the eight possible bit patterns of the lower 3 bits and you get the
971 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
972 * can be ignored. If the rest match '0', we have an octal */
973#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 974
9fb1bf9d 975#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 976
a500dc72
KW
977/* Character class numbers. For internal core Perl use only. The ones less
978 * than 32 are used in PL_charclass[] and the ones up through the one that
979 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
980 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
981 * their actual definitions are here. If that file has a name not used here,
982 * it won't compile.
1709d539
KW
983 *
984 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
985 * frequency of their use. This gives a slight edge to exiting a loop earlier
986 * (in reginclass() in regexec.c) */
1709d539
KW
987# define _CC_WORDCHAR 0 /* \w and [:word:] */
988# define _CC_DIGIT 1 /* \d and [:digit:] */
989# define _CC_ALPHA 2 /* [:alpha:] */
990# define _CC_LOWER 3 /* [:lower:] */
991# define _CC_UPPER 4 /* [:upper:] */
992# define _CC_PUNCT 5 /* [:punct:] */
993# define _CC_PRINT 6 /* [:print:] */
15861f94 994# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 995# define _CC_GRAPH 8 /* [:graph:] */
359b005e 996# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1709d539 997
b0d691b2 998#define _FIRST_NON_SWASH_CC 10
1709d539
KW
999/* The character classes above are implemented with swashes. The second group
1000 * (just below) contains the ones implemented without. These are also sorted
1001 * in rough order of the frequency of their use, except that \v should be last,
1002 * as it isn't a real Posix character class, and some (small) inefficiencies in
1003 * regular expression handling would be introduced by putting it in the middle
1004 * of those that are. Also, cntrl and ascii come after the others as it may be
1005 * useful to group these which have no members that match above Latin1, (or
1006 * above ASCII in the latter case) */
1007
779cf272 1008# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1009# define _CC_BLANK 11 /* [:blank:] */
1010# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1011# define _CC_CNTRL 13 /* [:cntrl:] */
1012# define _CC_ASCII 14 /* [:ascii:] */
1013# define _CC_VERTSPACE 15 /* \v */
1709d539 1014
a0947d7b
KW
1015# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1016
1709d539 1017/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1018 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1019# define _CC_IDFIRST 16
1020# define _CC_CHARNAME_CONT 17
1021# define _CC_NONLATIN1_FOLD 18
1022# define _CC_NONLATIN1_SIMPLE_FOLD 19
1023# define _CC_QUOTEMETA 20
1024# define _CC_NON_FINAL_FOLD 21
1025# define _CC_IS_IN_SOME_FOLD 22
1026# define _CC_MNEMONIC_CNTRL 23
073c22b3
KW
1027
1028/* This next group is only used on EBCDIC platforms, so theoretically could be
1029 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
1030# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1031# define _CC_UTF8_IS_START 29
1032# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1033# define _CC_UTF8_IS_CONTINUATION 31
1034/* Unused: 24-27
f4cdb42c
KW
1035 * If more bits are needed, one could add a second word for non-64bit
1036 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1037 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1038 * is used only for optimization (as of this writing), and differs in the
1039 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1040 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1041 * would just cause /i regexes which match them to run less efficiently.
1042 * Similarly the EBCDIC-only bits are used just for speed, and could be
1043 * replaced by other means */
96ac0975 1044
3a371f2f
KW
1045#if defined(PERL_CORE) || defined(PERL_EXT)
1046/* An enum version of the character class numbers, to help compilers
1047 * optimize */
1048typedef enum {
3a371f2f 1049 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1050 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1051 _CC_ENUM_ASCII = _CC_ASCII,
1052 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1053 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1054 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1055 _CC_ENUM_DIGIT = _CC_DIGIT,
1056 _CC_ENUM_GRAPH = _CC_GRAPH,
1057 _CC_ENUM_LOWER = _CC_LOWER,
1058 _CC_ENUM_PRINT = _CC_PRINT,
1059 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1060 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1061 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1062 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1063 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1064 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1065} _char_class_number;
1066#endif
1067
63c61c3f 1068#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1069#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1070
b7d90381
KW
1071#if defined(PERL_IN_UTF8_C) \
1072 || defined(PERL_IN_REGCOMP_C) \
1073 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1074# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1075 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1076 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1077 #error Need to adjust order of swash_property_names[]
1078# endif
1079
b44daeeb
KW
1080/* This is declared static in each of the few files that this is #defined for
1081 * to keep them from being publicly accessible. Hence there is a small amount
1082 * of wasted space */
1083
922e8cb4
KW
1084static const char* const swash_property_names[] = {
1085 "XPosixWord",
1086 "XPosixDigit",
1087 "XPosixAlpha",
1088 "XPosixLower",
1089 "XPosixUpper",
1090 "XPosixPunct",
1091 "XPosixPrint",
1092 "XPosixAlnum",
b0d691b2
KW
1093 "XPosixGraph",
1094 "Cased"
922e8cb4
KW
1095};
1096#endif
1097
6635f04f 1098START_EXTERN_C
96ac0975
NC
1099# ifdef DOINIT
1100EXTCONST U32 PL_charclass[] = {
1101# include "l1_char_class_tab.h"
1102};
1103
1104# else /* ! DOINIT */
1105EXTCONST U32 PL_charclass[];
1106# endif
6635f04f 1107END_EXTERN_C
96ac0975 1108
265c1f46 1109 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1110# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1111
1112 /* For internal core Perl use only: the base macro for defining macros like
1113 * isALPHA */
ff7ecfc3 1114# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1115 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1116
f4cdb42c
KW
1117 /* The mask for the _A versions of the macros; it just adds in the bit for
1118 * ASCII. */
1119# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1120
4650c663
KW
1121 /* For internal core Perl use only: the base macro for defining macros like
1122 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1123 * the ASCII bit are present */
b7d90381
KW
1124# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1125 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1126 == _CC_mask_A(classnum)))
f4cdb42c 1127
f4cdb42c 1128# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1129# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1130# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1131# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1132# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1133# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1134# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1135# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1136# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1137# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1138# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1139# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1140# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1141 */
d95f8b6a 1142# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1143# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1144# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1145# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1146
1147 /* continuation character for legal NAME in \N{NAME} */
1148# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1149
1150# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1151# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1152# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1153# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1154# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1155# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1156# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1157# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1158# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1159# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1160
0852beac
KW
1161# ifdef EBCDIC
1162# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1163# endif
1164
f12c0118
KW
1165 /* Participates in a single-character fold with a character above 255 */
1166# 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)))
1167
1168 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1169# 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 1170
4eeeb416 1171# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1172# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1173 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1174# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1175 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1176# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1177 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1178#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1179
1180 /* If we don't have perl.h, we are compiling a utility program. Below we
1181 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1182 * to it. Most are coded based on first principles. These are written to
74665a89 1183 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
0852beac
KW
1184# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1185# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1186# define isSPACE_A(c) (isBLANK_A(c) \
1187 || (c) == '\n' \
1188 || (c) == '\r' \
1189 || (c) == '\v' \
0852beac 1190 || (c) == '\f')
74665a89
KW
1191 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1192 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1193 * performance (if optimization isn't on), and allow the same code to be
1194 * used for both platform types */
1195# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1196 && ( (c) <= 'i' \
1197 || ((c) >= 'j' && (c) <= 'r') \
1198 || (c) >= 's'))
1199# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1200 && ( (c) <= 'I' \
1201 || ((c) >= 'J' && (c) <= 'R') \
1202 || (c) >= 'S'))
a4d7a999
KW
1203# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1204# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1205# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1206# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
74665a89
KW
1207# define isXDIGIT_A(c) (isDIGIT_A(c) \
1208 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1209 || ((c) <= 'F' && (c) >= 'A'))
74665a89
KW
1210# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1211 || (c) == '#' || (c) == '$' || (c) == '%' \
1212 || (c) == '&' || (c) == '\'' || (c) == '(' \
1213 || (c) == ')' || (c) == '*' || (c) == '+' \
1214 || (c) == ',' || (c) == '.' || (c) == '/' \
1215 || (c) == ':' || (c) == ';' || (c) == '<' \
1216 || (c) == '=' || (c) == '>' || (c) == '?' \
1217 || (c) == '@' || (c) == '[' || (c) == '\\' \
1218 || (c) == ']' || (c) == '^' || (c) == '_' \
1219 || (c) == '`' || (c) == '{' || (c) == '|' \
1220 || (c) == '}' || (c) == '~')
1221# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1222# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1223
0852beac 1224# ifdef EBCDIC
74665a89
KW
1225 /* The below is accurate for the 3 EBCDIC code pages traditionally
1226 * supported by perl. The only difference between them in the controls
1227 * is the position of \n, and that is represented symbolically below */
1228# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1229 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1230 || (c) == '\t' || (c) == '\v' \
1231 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1232 || (c) == 7 /* U+7F DEL */ \
1233 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1234 /* DLE, DC[1-3] */ \
1235 || (c) == 0x18 /* U+18 CAN */ \
1236 || (c) == 0x19 /* U+19 EOM */ \
1237 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1238 || (c) == 0x26 /* U+17 ETB */ \
1239 || (c) == 0x27 /* U+1B ESC */ \
1240 || (c) == 0x2D /* U+05 ENQ */ \
1241 || (c) == 0x2E /* U+06 ACK */ \
1242 || (c) == 0x32 /* U+16 SYN */ \
1243 || (c) == 0x37 /* U+04 EOT */ \
1244 || (c) == 0x3C /* U+14 DC4 */ \
1245 || (c) == 0x3D /* U+15 NAK */ \
1246 || (c) == 0x3F)/* U+1A SUB */
0852beac 1247# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1248# else /* isASCII is already defined for ASCII platforms, so can use that to
1249 define isCNTRL */
1250# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1251# endif
1252
3ffc8c70 1253 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1254 * during debugging, and it seems best to keep them. We may be called
1255 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1256 * do anything anyway, so make it not a problem */
1257# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1258# define NATIVE_TO_LATIN1(ch) (ch)
1259# endif
3ded5eb0
KW
1260# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1261# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1262# define isBLANK_L1(c) (isBLANK_A(c) \
1263 || (FITS_IN_8_BITS(c) \
1264 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1265# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1266# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1267# define isLOWER_L1(c) (isLOWER_A(c) \
1268 || (FITS_IN_8_BITS(c) \
1269 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1270 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1271 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1272 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1273 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1274# define isPRINT_L1(c) (isPRINT_A(c) \
1275 || (FITS_IN_8_BITS(c) \
1276 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1277# define isPUNCT_L1(c) (isPUNCT_A(c) \
1278 || (FITS_IN_8_BITS(c) \
1279 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1280 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1281 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1282 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1283 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1284 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1285 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1286# define isSPACE_L1(c) (isSPACE_A(c) \
1287 || (FITS_IN_8_BITS(c) \
1288 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1289 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1290# define isUPPER_L1(c) (isUPPER_A(c) \
1291 || (FITS_IN_8_BITS(c) \
1292 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1293 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1294 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1295# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1296# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1297# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1298 || isBLANK_L1(c) \
1299 || (c) == '-' \
1300 || (c) == '(' \
1301 || (c) == ')')
1302 /* The following are not fully accurate in the above-ASCII range. I (khw)
1303 * don't think it's necessary to be so for the purposes where this gets
1304 * compiled */
1305# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1306# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1307
1308 /* And these aren't accurate at all. They are useful only for above
1309 * Latin1, which utilities and bootstrapping don't deal with */
1310# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1311# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1312# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1313
1314 /* Many of the macros later in this file are defined in terms of these. By
1315 * implementing them with a function, which converts the class number into
1316 * a call to the desired macro, all of the later ones work. However, that
1317 * function won't be actually defined when building a utility program (no
1318 * perl.h), and so a compiler error will be generated if one is attempted
1319 * to be used. And the above-Latin1 code points require Unicode tables to
1320 * be present, something unlikely to be the case when bootstrapping */
1321# define _generic_isCC(c, classnum) \
1322 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1323# define _generic_isCC_A(c, classnum) \
1324 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1325#endif /* End of no perl.h H_PERL */
8a58bdcf 1326
e66b99e9
KW
1327#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1328#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1329#define isASCII_A(c) isASCII(c)
1330#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1331#define isBLANK(c) isBLANK_A(c)
1332#define isCNTRL(c) isCNTRL_A(c)
1333#define isDIGIT(c) isDIGIT_A(c)
1334#define isGRAPH(c) isGRAPH_A(c)
1335#define isIDFIRST(c) isIDFIRST_A(c)
1336#define isLOWER(c) isLOWER_A(c)
1337#define isPRINT(c) isPRINT_A(c)
779cf272 1338#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1339#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1340#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1341#define isPUNCT(c) isPUNCT_A(c)
1342#define isSPACE(c) isSPACE_A(c)
1343#define isUPPER(c) isUPPER_A(c)
1344#define isWORDCHAR(c) isWORDCHAR_A(c)
1345#define isXDIGIT(c) isXDIGIT_A(c)
1346
1347/* ASCII casing. These could also be written as
1348 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1349 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1350 which uses table lookup and mask instead of subtraction. (This would
1351 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1352#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1353#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1354
25200305
KW
1355/* In the ASCII range, these are equivalent to what they're here defined to be.
1356 * But by creating these definitions, other code doesn't have to be aware of
1357 * this detail */
1358#define toFOLD(c) toLOWER(c)
25200305
KW
1359#define toTITLE(c) toUPPER(c)
1360
c753c8d3
KW
1361#define toLOWER_A(c) toLOWER(c)
1362#define toUPPER_A(c) toUPPER(c)
25200305
KW
1363#define toFOLD_A(c) toFOLD(c)
1364#define toTITLE_A(c) toTITLE(c)
1a0901db 1365
4650c663 1366/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1367#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1368 ? (c) \
f4cd282c 1369 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1370#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1371
1a0901db 1372/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1373 * all mapped to one of them, and these need special handling; returns the
1374 * input itself if is out-of-range */
b2bf251f 1375#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1376 ? (c) \
f4cd282c 1377 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1378#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1379
beab9ebe
KW
1380/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1381
1382/* For internal core Perl use only: the base macro for defining macros like
1383 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1384 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1385 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1386 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1387 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1388 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1389 * code points under locale rules, use the macros based on _generic_LC_uvchr
1390 * instead of this. */
beab9ebe
KW
1391#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1392 (! FITS_IN_8_BITS(c) \
1393 ? 0 \
31f05a37
KW
1394 : IN_UTF8_CTYPE_LOCALE \
1395 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1396 : cBOOL(non_utf8))
1397
1398/* For internal core Perl use only: a helper macro for defining macros like
1399 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1400 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1401 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1402 * _generic_LC_base for more info */
1403#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1404 _generic_LC_base(c,utf8_locale_classnum, \
1405 non_utf8_func( (_LC_CAST) (c)))
1406
1407/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1408 * 'c' is the platform's native underscore character */
1409#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1410 _generic_LC_base(c, utf8_locale_classnum, \
1411 (non_utf8_func( (_LC_CAST) (c)) \
1412 || (char)(c) == '_'))
1413
1414/* These next three are also for internal core Perl use only: case-change
1415 * helper macros */
1416#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1417 ? (c) \
31f05a37
KW
1418 : (IN_UTF8_CTYPE_LOCALE) \
1419 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1420 : (cast)function((cast)(c)))
beab9ebe 1421
31f05a37
KW
1422/* Note that the result can be larger than a byte in a UTF-8 locale. It
1423 * returns a single value, so can't adequately return the upper case of LATIN
1424 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1425 * values "SS"); instead it asserts against that under DEBUGGING, and
1426 * otherwise returns its input */
beab9ebe
KW
1427#define _generic_toUPPER_LC(c, function, cast) \
1428 (! FITS_IN_8_BITS(c) \
1429 ? (c) \
31f05a37 1430 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1431 ? (cast)function((cast)(c)) \
31f05a37
KW
1432 : ((((U8)(c)) == MICRO_SIGN) \
1433 ? GREEK_CAPITAL_LETTER_MU \
1434 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1435 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1436 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1437 ? (__ASSERT_(0) (c)) \
1438 : PL_mod_latin1_uc[ (U8) (c) ])))))
1439
1440/* Note that the result can be larger than a byte in a UTF-8 locale. It
1441 * returns a single value, so can't adequately return the fold case of LATIN
1442 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1443 * values "ss"); instead it asserts against that under DEBUGGING, and
1444 * otherwise returns its input */
beab9ebe 1445#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1446 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1447 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1448 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1449 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1450 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1451
84061b6a 1452/* Use the libc versions for these if available. */
f05550c0 1453#if defined(HAS_ISASCII)
84061b6a
KW
1454# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1455#else
1456# define isASCII_LC(c) isASCII(c)
1457#endif
1458
f05550c0 1459#if defined(HAS_ISBLANK)
84061b6a 1460# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1461#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1462# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1463#endif
1464
f05550c0 1465#define _LC_CAST U8
bbce6d69 1466
f05550c0 1467#ifdef WIN32
375f5f06
KW
1468 /* The Windows functions don't bother to follow the POSIX standard, which
1469 * for example says that something can't both be a printable and a control.
1470 * But Windows treats the \t control as a printable, and does such things
1471 * as making superscripts into both digits and punctuation. This tames
1472 * these flaws by assuming that the definitions of both controls and space
1473 * are correct, and then making sure that other definitions don't have
1474 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1475 * Not all possible weirdnesses are checked for, just the ones that were
1476 * detected on actual Microsoft code pages */
1477
b7d90381
KW
1478# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1479# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1480
1481# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1482 && isALPHANUMERIC_LC(c))
1483# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1484 ! isPUNCT_LC(c))
1485# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1486 isALPHANUMERIC_LC(c))
1487# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1488# define isIDFIRST_LC(c) (((c) == '_') \
1489 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1490# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1491# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1492# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1493# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1494# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1495# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1496 && isALPHANUMERIC_LC(c))
f05550c0
BF
1497
1498# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1499# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1500# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1501
1502#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1503 /* For most other platforms */
beab9ebe 1504
f05550c0
BF
1505# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1506# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1507# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1508# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1509# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1510# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1511# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1512# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1513# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1514# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1515# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1516# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1517# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1518
1519
1520# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1521# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1522# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1523
1524#else /* The final fallback position */
1525
b7d90381
KW
1526# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1527# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1528# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1529# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1530# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1531# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1532# define isLOWER_LC(c) (isascii(c) && islower(c))
1533# define isPRINT_LC(c) (isascii(c) && isprint(c))
1534# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1535# define isSPACE_LC(c) (isascii(c) && isspace(c))
1536# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1537# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1538# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1539
1540# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1541# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1542# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1543
f05550c0 1544#endif
55204971 1545
eba68aa0
KW
1546#define isIDCONT(c) isWORDCHAR(c)
1547#define isIDCONT_A(c) isWORDCHAR_A(c)
1548#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1549#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1550#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1551
4650c663 1552/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1553 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1554 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1555 * classes where there is a macro or function 'above_latin1' that takes the
1556 * single argument 'c' and returns the desired value. These exist for those
1557 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1558 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1559 * for classes where that overhead is faster than a direct lookup.
d0da05db 1560 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1561 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1562 * have duplicate checks here, so could create versions of the macros that
1563 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1564
1565/* Note that all ignore 'use bytes' */
d0da05db 1566#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
cd500f2f
KW
1567 ? _generic_isCC(c, classnum) \
1568 : above_latin1(c))
d0da05db 1569#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
cd500f2f 1570 ? _generic_isCC(c, classnum) \
922e8cb4 1571 : _is_uni_FOO(classnum, c))
d0da05db
KW
1572#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1573#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1574#define isASCII_uvchr(c) isASCII(c)
1575#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1576#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1577#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1578#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1579#define isIDCONT_uvchr(c) _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1580#define isIDFIRST_uvchr(c) _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
1581#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1582#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1583
1584#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1585#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1586#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1587
1588#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1589#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1590#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1591#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1592
1593#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1594#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1595#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1596#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1597
1598/* For backwards compatibility, even though '_uni' should mean official Unicode
1599 * code points, in Perl it means native for those below 256 */
1600#define isALPHA_uni(c) isALPHA_uvchr(c)
1601#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1602#define isASCII_uni(c) isASCII_uvchr(c)
1603#define isBLANK_uni(c) isBLANK_uvchr(c)
1604#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1605#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1606#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1607#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1608#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1609#define isLOWER_uni(c) isLOWER_uvchr(c)
1610#define isPRINT_uni(c) isPRINT_uvchr(c)
1611#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1612#define isSPACE_uni(c) isSPACE_uvchr(c)
1613#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1614#define isUPPER_uni(c) isUPPER_uvchr(c)
1615#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1616#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1617#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1618#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1619#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1620#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1621#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1622
4650c663
KW
1623/* For internal core Perl use only: the base macros for defining macros like
1624 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1625 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1626 * simple class definitions; the other for more complex. These are like
d0da05db 1627 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1628#define _generic_LC_uvchr(latin1, above_latin1, c) \
1629 (c < 256 ? latin1(c) : above_latin1(c))
1630#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1631 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1632
1633#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1634#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1635 _CC_ALPHANUMERIC, c)
b7d90381
KW
1636#define isASCII_LC_uvchr(c) isASCII_LC(c)
1637#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1638 is_HORIZWS_cp_high, c)
feeab5a9 1639#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1640#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1641#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1642#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1643 _is_uni_perl_idcont, c)
b7d90381 1644#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1645 _is_uni_perl_idstart, c)
1646#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1647#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1648#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1649#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1650#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1651 is_XPERLSPACE_cp_high, c)
cd500f2f 1652#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1653#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1654 _CC_WORDCHAR, c)
b7d90381
KW
1655#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1656 is_XDIGIT_cp_high, c)
e712593e 1657
b7d90381 1658#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1659
4650c663
KW
1660/* For internal core Perl use only: the base macros for defining macros like
1661 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1662 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1663 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1664 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1665 * same representation whether utf8 or not. Note that it assumes that the utf8
1666 * has been validated, and ignores 'use bytes' */
cd500f2f 1667#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1668 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1669 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1670 ? _generic_isCC( \
a62b247b 1671 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
aea0c08c 1672 *((p)+1 )), \
94bb8c36 1673 classnum) \
252810d0 1674 : utf8)
b7d90381
KW
1675/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1676 * 'above_latin1' can be a macro */
cd500f2f
KW
1677#define _generic_func_utf8(classnum, above_latin1, p) \
1678 _generic_utf8(classnum, p, above_latin1(p))
4650c663 1679/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1680 * 'above_latin1' parameter */
922e8cb4 1681#define _generic_swash_utf8(classnum, p) \
cd500f2f 1682 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
922e8cb4 1683
cc8ab7c0 1684/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1685 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1686 * class is TRUE for. Hence it can skip the tests for this range.
1687 * 'above_latin1' should include its arguments */
3ded5eb0 1688#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1689 (UTF8_IS_INVARIANT(*(p)) \
1690 ? _generic_isCC(*(p), classnum) \
1691 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1692 ? above_latin1 \
cc8ab7c0
KW
1693 : 0)
1694
84238efa
KW
1695/* NOTE that some of these macros have very similar ones in regcharclass.h.
1696 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1697 * there, differing in name only by an underscore from the one here
1698 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1699 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1700 * points; the regcharclass.h ones are implemented as a series of
1701 * "if-else-if-else ..." */
1702
b7d90381
KW
1703#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1704#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1705#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
509fb054
KW
1706 utf8, the non-utf8 macro works
1707 */
b7d90381 1708#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2
KW
1709
1710#ifdef EBCDIC
1711 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1712 * more efficient macro instead of the more general one */
88794300 1713# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
e8fa43e2
KW
1714#else
1715# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1716#endif
1717
cd500f2f
KW
1718#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1719 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1720#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1721#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1722 _is_utf8_perl_idcont, p)
e5dcd934 1723
c11ff943
KW
1724/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1725 * IDFIRST is an alnum. See
c8362b00 1726 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1727 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1728 * This used to be not the XID version, but we decided to go with the more
1729 * modern Unicode definition */
b7d90381 1730#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1731 _is_utf8_perl_idstart, p)
e5dcd934 1732
b7d90381
KW
1733#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1734#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1735#define isPSXSPC_utf8(p) isSPACE_utf8(p)
1736#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1737#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1738#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1739#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1740#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1741#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
cd500f2f 1742 is_XDIGIT_high(p))
a0ed51b3 1743
7d2b61dd 1744#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1745#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1746#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1747#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1748
4650c663
KW
1749/* For internal core Perl use only: the base macros for defining macros like
1750 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1751 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1752 * 'macro' parameter */
94bb8c36
KW
1753#define _generic_LC_utf8(macro, p, utf8) \
1754 (UTF8_IS_INVARIANT(*(p)) \
1755 ? macro(*(p)) \
1756 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
a62b247b 1757 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
9201a572
KW
1758 : utf8)
1759
cd500f2f
KW
1760#define _generic_LC_swash_utf8(macro, classnum, p) \
1761 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1762#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1763 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1764
b7d90381 1765#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
cd500f2f 1766 _CC_ALPHANUMERIC, p)
b7d90381
KW
1767#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1768#define isASCII_LC_utf8(p) isASCII_LC(*p)
1769#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1770 is_HORIZWS_high, p)
1771#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1772#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1773#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1774#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1775 _is_utf8_perl_idcont, p)
1776#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1777 _is_utf8_perl_idstart, p)
1778#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1779#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1780#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1781#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1782#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1783 is_XPERLSPACE_high, p)
1784#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
cd500f2f
KW
1785#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1786 _CC_WORDCHAR, p)
b7d90381
KW
1787#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1788 is_XDIGIT_high, p)
aaa51d5e 1789
fbc19f27
KW
1790/* Macros for backwards compatibility and for completeness when the ASCII and
1791 * Latin1 values are identical */
b7d90381
KW
1792#define isALPHAU(c) isALPHA_L1(c)
1793#define isDIGIT_L1(c) isDIGIT_A(c)
1794#define isOCTAL(c) isOCTAL_A(c)
1795#define isOCTAL_L1(c) isOCTAL_A(c)
1796#define isXDIGIT_L1(c) isXDIGIT_A(c)
1797#define isALNUM(c) isWORDCHAR(c)
1798#define isALNUMU(c) isWORDCHAR_L1(c)
1799#define isALNUM_LC(c) isWORDCHAR_LC(c)
1800#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 1801#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
1802#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1803#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1804#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1805#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1806#define isALNUMC(c) isALPHANUMERIC(c)
1807#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1808#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 1809#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 1810#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 1811#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1812
2bd1cbf6
KW
1813/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1814 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1815 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1816 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1817 * the outlier from the block that contains the other controls, just like
1818 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1819 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
1820 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1821 * etc. */
2bd1cbf6 1822#ifndef EBCDIC
75763b3a 1823# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 1824#else
75763b3a
KW
1825# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1826 ((isPRINT_A(c)) \
1827 ? (UNLIKELY((c) == '?') \
1828 ? QUESTION_MARK_CTRL \
1829 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1830 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1831 ? '?' \
1832 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 1833#endif
bbce6d69 1834
dea28490
JJ
1835/* Line numbers are unsigned, 32 bits. */
1836typedef U32 line_t;
e5dcd934 1837#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 1838
91152fc1
DG
1839/* Helpful alias for version prescan */
1840#define is_LAX_VERSION(a,b) \
1841 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1842
1843#define is_STRICT_VERSION(a,b) \
1844 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1845
1846#define BADVERSION(a,b,c) \
1847 if (b) { \
1848 *b = c; \
1849 } \
1850 return a;
8c52afec 1851
cb27eebd
KW
1852/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1853 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1854 * which is advanced. The input is validated only by an assert() in DEBUGGING
1855 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1856 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1857#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1858 ? (c) \
1859 : ((c) + 9))))
902008b8
KW
1860#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1861 ? (*(s)++) \
1862 : (*(s)++ + 9))))
95a59cab 1863
cb27eebd
KW
1864/* Converts a character known to represent an octal digit (0-7) to its numeric
1865 * value. The input is validated only by an assert() in DEBUGGING builds. In
1866 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1867#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1868
305b8651
KW
1869/* Efficiently returns a boolean as to if two native characters are equivalent
1870 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1871 * the ALPHA in the name is to remind you of that. This is asserted() in
1872 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1873 * works (on valid input) for both non- and UTF-8-encoded bytes.
1874 *
1875 * When one of the inputs is a compile-time constant and gets folded by the
1876 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1877 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1878 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
1879 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1880 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1881 * just a single 0, in the bit position where the upper- and lowercase differ.
1882 * */
305b8651
KW
1883#define isALPHA_FOLD_EQ(c1, c2) \
1884 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1885 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1886#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1887
8e84507e 1888/*
ccfc67b7
JH
1889=head1 Memory Management
1890
a02a5408 1891=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
1892The XSUB-writer's interface to the C C<malloc> function.
1893
596f7718 1894Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1895
c5008215
JC
1896In 5.9.3, Newx() and friends replace the older New() API, and drops
1897the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 1898themselves. This aid has been superseded by a new build option,
d10b4965 1899PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
1900there for use in XS modules supporting older perls.
1901
a02a5408 1902=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 1903The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 1904cast. See also C<L</Newx>>.
954c1994 1905
596f7718 1906Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1907
a02a5408 1908=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 1909The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 1910memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 1911
596f7718 1912Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1913
954c1994
GS
1914=for apidoc Am|void|Renew|void* ptr|int nitems|type
1915The XSUB-writer's interface to the C C<realloc> function.
1916
596f7718 1917Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1918
954c1994
GS
1919=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1920The XSUB-writer's interface to the C C<realloc> function, with
1921cast.
1922
596f7718 1923Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1924
49b8b560 1925=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
1926The XSUB-writer's interface to the C C<free> function.
1927
596f7718 1928This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 1929
954c1994
GS
1930=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1931The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 1932source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1933C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 1934
e90e2364 1935=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 1936Like C<Move> but returns C<dest>. Useful
72d33970 1937for encouraging compilers to tail-call
e90e2364
NC
1938optimise.
1939
954c1994
GS
1940=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1941The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 1942source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1943C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 1944
e90e2364
NC
1945=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1946
796b6530 1947Like C<Copy> but returns C<dest>. Useful
72d33970 1948for encouraging compilers to tail-call
e90e2364
NC
1949optimise.
1950
954c1994
GS
1951=for apidoc Am|void|Zero|void* dest|int nitems|type
1952
1953The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1954destination, C<nitems> is the number of items, and C<type> is the type.
1955
e90e2364
NC
1956=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1957
72d33970
FC
1958Like C<Zero> but returns dest. Useful
1959for encouraging compilers to tail-call
e90e2364
NC
1960optimise.
1961
da5d8dbb 1962=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 1963This is an architecture-independent macro to copy one structure to another.
954c1994 1964
7e337ee0
JH
1965=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1966
1967Fill up memory with a byte pattern (a byte repeated over and over
1968again) that hopefully catches attempts to access uninitialized memory.
1969
1970=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1971
1972PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1973
1c12ffb4 1974=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
1975
1976PoisonWith(0xEF) for catching access to freed memory.
1977
9965345d
JH
1978=for apidoc Am|void|Poison|void* dest|int nitems|type
1979
7e337ee0 1980PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
1981
1982=cut */
954c1994 1983
561b68a9
SH
1984/* Maintained for backwards-compatibility only. Use newSV() instead. */
1985#ifndef PERL_CORE
ff06c60c 1986#define NEWSV(x,len) newSV(len)
561b68a9 1987#endif
ff06c60c 1988
19a94d75
JH
1989#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1990
e6bdf523 1991
27d5b266 1992#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
1993
1994/* This expression will be constant-folded at compile time. It checks
1995 * whether or not the type of the count n is so small (e.g. U8 or U16, or
1996 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
1997 * As well as avoiding the need for a run-time check in some cases, it's
1998 * designed to avoid compiler warnings like:
1999 * comparison is always false due to limited range of data type
73e8ff00
DM
2000 * It's mathematically equivalent to
2001 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2002 */
2003
2004# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
004073ba 2005 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
e6bdf523 2006
88f9f128 2007/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2008 * compiler warnings. We *want* to write the expression as
2009 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2010 * (for some compile-time constant C), but even when the LHS
2011 * constant-folds to false at compile-time, g++ insists on emitting
2012 * warnings about the RHS (e.g. "comparison is always false"), so instead
2013 * we write it as
e6bdf523 2014 *
d98e5cde 2015 * (cond ? n : X) > C
88f9f128 2016 *
d98e5cde
DM
2017 * where X is a constant with X > C always false. Choosing a value for X
2018 * is tricky. If 0, some compilers will complain about 0 > C always being
2019 * false; if 1, Coverity complains when n happens to be the constant value
2020 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2021 * for X and hope that nothing else whines.
e6bdf523
DM
2022 */
2023
2024# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2025 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2026 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2027
2028# define MEM_WRAP_CHECK(n,t) \
2029 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2030 && (croak_memory_wrap(),0))
2031
2032# define MEM_WRAP_CHECK_1(n,t,a) \
2033 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2034 && (Perl_croak_nocontext("%s",(a)),0))
2035
8b44ba4c 2036#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2037
d1decf2b 2038#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
2039#else
2040
410319be
NC
2041#define MEM_WRAP_CHECK(n,t)
2042#define MEM_WRAP_CHECK_1(n,t,a)
2043#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
2044#define MEM_WRAP_CHECK_(n,t)
2045
6fff79ce 2046#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 2047
1936d2a7 2048#endif
8b44ba4c 2049
fe4f188c 2050#ifdef PERL_MEM_LOG
46c6c7e2 2051/*
9f653bb5 2052 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2053 * go through functions, which are handy for debugging breakpoints, but
2054 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2055 * line number, and C function name if available) passed in. This info can
2056 * then be used for logging the calls, for which one gets a sample
73d1d973 2057 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2058 *
46c6c7e2 2059 * Known problems:
94e892a6 2060 * - not all memory allocs get logged, only those
46c6c7e2 2061 * that go through Newx() and derivatives (while all
94e892a6 2062 * Safefrees do get logged)
46c6c7e2
JH
2063 * - __FILE__ and __LINE__ do not work everywhere
2064 * - __func__ or __FUNCTION__ even less so
2065 * - I think more goes on after the perlio frees but
2066 * the thing is that STDERR gets closed (as do all
2067 * the file descriptors)
2068 * - no deeper calling stack than the caller of the Newx()
2069 * or the kind, but do I look like a C reflection/introspection
2070 * utility to you?
2071 * - the function prototypes for the logging functions
2072 * probably should maybe be somewhere else than handy.h
2073 * - one could consider inlining (macrofying) the logging
2074 * for speed, but I am too lazy
2075 * - one could imagine recording the allocations in a hash,
2076 * (keyed by the allocation address?), and maintain that
2077 * through reallocs and frees, but how to do that without
2078 * any News() happening...?
73d1d973 2079 * - lots of -Ddefines to get useful/controllable output
b953482e 2080 * - lots of ENV reads
46c6c7e2
JH
2081 */
2082
0b0ab801 2083# ifdef PERL_CORE
73d1d973 2084# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2085enum mem_log_type {
2086 MLT_ALLOC,
2087 MLT_REALLOC,
d7a2c63c
MHM
2088 MLT_FREE,
2089 MLT_NEW_SV,
2090 MLT_DEL_SV
0b0ab801
MHM
2091};
2092# endif
12754f92 2093# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2094void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2095void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2096# endif
0b0ab801
MHM
2097# endif
2098
fe4f188c
JH
2099#endif
2100
2101#ifdef PERL_MEM_LOG
d1401ee9
MHM
2102#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2103#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2104#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2105#endif
2106
2107#ifndef MEM_LOG_ALLOC
2108#define MEM_LOG_ALLOC(n,t,a) (a)
2109#endif
2110#ifndef MEM_LOG_REALLOC
2111#define MEM_LOG_REALLOC(n,t,v,a) (a)
2112#endif
2113#ifndef MEM_LOG_FREE
2114#define MEM_LOG_FREE(a) (a)
2115#endif
2116
d1401ee9
MHM
2117#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2118#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2119#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2120
2121#ifndef PERL_CORE
a02a5408
JC
2122/* pre 5.9.x compatibility */
2123#define New(x,v,n,t) Newx(v,n,t)
2124#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2125#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2126#endif
a02a5408 2127
ff68c719 2128#define Renew(v,n,t) \
d1401ee9 2129 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2130#define Renewc(v,n,t,c) \
d1401ee9 2131 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2132
2133#ifdef PERL_POISON
2134#define Safefree(d) \
06c0cc96 2135 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2136#else
fe4f188c 2137#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2138#endif
55497cff 2139
8b44ba4c
NC
2140#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2141#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2142#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2143
8b44ba4c
NC
2144#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2145#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2146#ifdef HAS_MEMSET
8b44ba4c 2147#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2148#else
8b44ba4c
NC
2149/* Using bzero(), which returns void. */
2150#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2151#endif
2152
7e337ee0
JH
2153#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2154#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2155#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2156#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2157
caa674f3
DD
2158#ifdef PERL_POISON
2159# define PERL_POISON_EXPR(x) x
2160#else
2161# define PERL_POISON_EXPR(x)
2162#endif
2163
2304df62 2164#ifdef USE_STRUCT_COPY
ff68c719 2165#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2166#else
2167#define StructCopy(s,d,t) Copy(s,d,1,t)
2168#endif
2cc61e15 2169
c3caa5c3
JH
2170/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2171 * want your zero-based indices to be less than but not equal to).
2172 *
2173 * C_ARRAY_END is one past the last: half-open/half-closed range,
2174 * not last-inclusive range. */
622913ab 2175#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2176#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2177
2cc61e15
DD
2178#ifdef NEED_VA_COPY
2179# ifdef va_copy
2180# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2181# else
a1866d1b
JH
2182# if defined(__va_copy)
2183# define Perl_va_copy(s, d) __va_copy(d, s)
2184# else
2185# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2186# endif
2cc61e15
DD
2187# endif
2188#endif
2189
472d47bc
SB
2190/* convenience debug macros */
2191#ifdef USE_ITHREADS
2192#define pTHX_FORMAT "Perl interpreter: 0x%p"
2193#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2194#define pTHX_VALUE_ (void *)my_perl,
2195#define pTHX_VALUE (void *)my_perl
2196#define pTHX__VALUE_ ,(void *)my_perl,
2197#define pTHX__VALUE ,(void *)my_perl
472d47bc 2198#else
3609ea0d 2199#define pTHX_FORMAT
472d47bc 2200#define pTHX__FORMAT
3609ea0d 2201#define pTHX_VALUE_
472d47bc 2202#define pTHX_VALUE
3609ea0d 2203#define pTHX__VALUE_
472d47bc
SB
2204#define pTHX__VALUE
2205#endif /* USE_ITHREADS */
3609ea0d 2206
2acdbac1
NC
2207/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2208 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2209 CPAN::Unpack show any users outside the core. */
2210#ifdef PERL_CORE
b7d90381
KW
2211# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2212 "Use of " s " is deprecated")
2acdbac1
NC
2213#endif
2214
dfff4baf
BF
2215/* Internal macros to deal with gids and uids */
2216#ifdef PERL_CORE
2217
2218# if Uid_t_size > IVSIZE
2219# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2220# define SvUID(sv) SvNV(sv)
2221# else
2222# if Uid_t_sign <= 0
2223# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2224# define SvUID(sv) SvIV(sv)
2225# else
2226# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2227# define SvUID(sv) SvUV(sv)
2228# endif
2229# endif /* Uid_t_size */
2230
2231# if Gid_t_size > IVSIZE
2232# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2233# define SvGID(sv) SvNV(sv)
2234# else
2235# if Gid_t_sign <= 0
2236# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2237# define SvGID(sv) SvIV(sv)
2238# else
2239# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2240# define SvGID(sv) SvUV(sv)
2241# endif
2242# endif /* Gid_t_size */
2243
2244#endif
2245
9d745869
KW
2246#endif /* HANDY_H */
2247
e9a8c099 2248/*
14d04a33 2249 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2250 */