This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
APItest: Remove use of macros about to be removed
[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
6a5bc5ac
KW
14#ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */
15#define PERL_HANDY_H_
9d745869 16
24792b8d
NC
17#ifndef PERL_CORE
18# define Null(type) ((type)NULL)
954c1994
GS
19
20/*
ccfc67b7 21=head1 Handy Values
954c1994 22
78342678 23=for apidoc AmnU||Nullch
72d33970
FC
24Null character pointer. (No longer available when C<PERL_CORE> is
25defined.)
2307c6d0 26
78342678 27=for apidoc AmnU||Nullsv
72d33970 28Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
29
30=cut
31*/
32
24792b8d
NC
33# define Nullch Null(char*)
34# define Nullfp Null(PerlIO*)
35# define Nullsv Null(SV*)
36#endif
8d063cd8 37
641d3f0b
PP
38#ifdef TRUE
39#undef TRUE
40#endif
41#ifdef FALSE
42#undef FALSE
43#endif
44#define TRUE (1)
45#define FALSE (0)
46
cf3f0ffb
DM
47/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
48 * (compiler permitting) that casting away const-ness will give a warning;
49 * e.g.:
50 *
51 * const SV *sv = ...;
52 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
53 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
54 */
55
b1bc3f34 56#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
f065df04 57# define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
b1bc3f34
NC
58#else
59# define MUTABLE_PTR(p) ((void *) (p))
60#endif
61
a062e10d 62#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 63#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 64#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 65#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 66#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 67#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 68
f789f6a4 69#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
70# include <stdbool.h>
71# ifndef HAS_BOOL
72# define HAS_BOOL 1
73# endif
74#endif
75
8e84507e 76/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
77 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
78 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 79 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
80 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
81 g++ can be identified by __GNUG__.
82 Andy Dougherty February 2000
5d94fbed 83*/
3609ea0d 84#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 85# ifndef PERL_BOOL_AS_CHAR
5d94fbed 86# ifndef HAS_BOOL
c1d22f6b 87# define HAS_BOOL 1
5d94fbed 88# endif
f789f6a4 89# endif
5d94fbed
AD
90#endif
91
92#ifndef HAS_BOOL
f789f6a4
FC
93# ifdef bool
94# undef bool
95# endif
70d5cb32 96# define bool char
c1d22f6b 97# define HAS_BOOL 1
a687059c 98#endif
0d3e774c 99
25ba28ce
KW
100/*
101=for apidoc Am|bool|cBOOL|bool expr
102
103Cast-to-bool. A simple S<C<(bool) I<expr>>> cast may not do the right thing:
104if C<bool> is defined as C<char>, for example, then the cast from C<int> is
105implementation-defined.
106
107C<(bool)!!(cbool)> in a ternary triggers a bug in xlc on AIX
108
109=cut
110*/
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__
07798b17
AC
119#elif (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
120 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
121# define FUNCTION__ ""
46c6c7e2 122#else
07798b17 123# define FUNCTION__ __FUNCTION__ /* Common extension. */
46c6c7e2
JH
124#endif
125
27d4fb96
PP
126/* XXX A note on the perl source internal type system. The
127 original intent was that I32 be *exactly* 32 bits.
128
129 Currently, we only guarantee that I32 is *at least* 32 bits.
130 Specifically, if int is 64 bits, then so is I32. (This is the case
131 for the Cray.) This has the advantage of meshing nicely with
132 standard library calls (where we pass an I32 and the library is
133 expecting an int), but the disadvantage that an I32 is not 32 bits.
134 Andy Dougherty August 1996
24fef2a7 135
dc45a647
MB
136 There is no guarantee that there is *any* integral type with
137 exactly 32 bits. It is perfectly legal for a system to have
138 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 139
dc45a647
MB
140 Similarly, there is no guarantee that I16 and U16 have exactly 16
141 bits.
693762b4 142
8e84507e
NIS
143 For dealing with issues that may arise from various 32/64-bit
144 systems, we will ask Configure to check out
8175356b 145
3609ea0d
JH
146 SHORTSIZE == sizeof(short)
147 INTSIZE == sizeof(int)
148 LONGSIZE == sizeof(long)
dc45a647 149 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 150 PTRSIZE == sizeof(void *)
dc45a647
MB
151 DOUBLESIZE == sizeof(double)
152 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 153
27d4fb96
PP
154*/
155
69512466
JH
156#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
157# include <inttypes.h>
dd0eed91
JH
158# ifdef INT32_MIN_BROKEN
159# undef INT32_MIN
160# define INT32_MIN (-2147483647-1)
161# endif
162# ifdef INT64_MIN_BROKEN
163# undef INT64_MIN
164# define INT64_MIN (-9223372036854775807LL-1)
165# endif
69512466
JH
166#endif
167
8175356b
JH
168typedef I8TYPE I8;
169typedef U8TYPE U8;
170typedef I16TYPE I16;
171typedef U16TYPE U16;
172typedef I32TYPE I32;
173typedef U32TYPE U32;
16d89be8 174
74b807c7 175#ifdef QUADKIND
8175356b
JH
176typedef I64TYPE I64;
177typedef U64TYPE U64;
16d89be8 178#endif
8175356b 179
d8668976 180#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 181
5ff3f7a4
GS
182/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
183 Please search CHAR_MAX in perl.h for further details. */
184#define U8_MAX UINT8_MAX
185#define U8_MIN UINT8_MIN
186
5ff3f7a4
GS
187#define I16_MAX INT16_MAX
188#define I16_MIN INT16_MIN
189#define U16_MAX UINT16_MAX
190#define U16_MIN UINT16_MIN
191
5ff3f7a4
GS
192#define I32_MAX INT32_MAX
193#define I32_MIN INT32_MIN
0e983133
GS
194#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
195# define U32_MAX UINT32_MAX
196#else
197# define U32_MAX 4294967295U
198#endif
5ff3f7a4
GS
199#define U32_MIN UINT32_MIN
200
201#else
202
5c9fa16e
KA
203/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
204 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
205#define U8_MAX PERL_UCHAR_MAX
206#define U8_MIN PERL_UCHAR_MIN
79072805 207
27d4fb96
PP
208#define I16_MAX PERL_SHORT_MAX
209#define I16_MIN PERL_SHORT_MIN
210#define U16_MAX PERL_USHORT_MAX
211#define U16_MIN PERL_USHORT_MIN
79072805 212
c4f23d77 213#if LONGSIZE > 4
27d4fb96
PP
214# define I32_MAX PERL_INT_MAX
215# define I32_MIN PERL_INT_MIN
216# define U32_MAX PERL_UINT_MAX
217# define U32_MIN PERL_UINT_MIN
79072805 218#else
27d4fb96
PP
219# define I32_MAX PERL_LONG_MAX
220# define I32_MIN PERL_LONG_MIN
221# define U32_MAX PERL_ULONG_MAX
222# define U32_MIN PERL_ULONG_MIN
79072805
LW
223#endif
224
5ff3f7a4
GS
225#endif
226
247cee9f
KW
227/* These C99 typedefs are useful sometimes for, say, loop variables whose
228 * maximum values are small, but for which speed trumps size. If we have a C99
229 * compiler, use that. Otherwise, a plain 'int' should be good enough.
230 *
231 * Restrict these to core for now until we are more certain this is a good
232 * idea. */
233#if defined(PERL_CORE) || defined(PERL_EXT)
234# ifdef I_STDINT
235 typedef int_fast8_t PERL_INT_FAST8_T;
236 typedef uint_fast8_t PERL_UINT_FAST8_T;
237 typedef int_fast16_t PERL_INT_FAST16_T;
238 typedef uint_fast16_t PERL_UINT_FAST16_T;
239# else
240 typedef int PERL_INT_FAST8_T;
241 typedef unsigned int PERL_UINT_FAST8_T;
242 typedef int PERL_INT_FAST16_T;
243 typedef unsigned int PERL_UINT_FAST16_T;
244# endif
245#endif
246
464decb6
KW
247/* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
248 * anyone is grepping for it */
249#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
fc36a67e
PP
250#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
251#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
252
88794300 253/* Unused by core; should be deprecated */
ff68c719 254#define Ctl(ch) ((ch) & 037)
8d063cd8 255
98fce2a4
KW
256#if defined(PERL_CORE) || defined(PERL_EXT)
257# ifndef MIN
258# define MIN(a,b) ((a) < (b) ? (a) : (b))
259# endif
260# ifndef MAX
261# define MAX(a,b) ((a) > (b) ? (a) : (b))
262# endif
263#endif
264
84ff4fa9
KW
265/* Returns a boolean as to whether the input unsigned number is a power of 2
266 * (2**0, 2**1, etc). In other words if it has just a single bit set.
267 * If not, subtracting 1 would leave the uppermost bit set, so the & would
268 * yield non-zero */
269#if defined(PERL_CORE) || defined(PERL_EXT)
011b1419 270# define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
84ff4fa9
KW
271#endif
272
8d9433eb
KW
273/*
274=for apidoc Am|void|__ASSERT_|bool expr
275
276This is a helper macro to avoid preprocessor issues, replaced by nothing
277unless under DEBUGGING, where it expands to an assert of its argument,
278followed by a comma (hence the comma operator). If we just used a straight
279assert(), we would get a comma with nothing before it when not DEBUGGING.
280
281=cut
282
283We also use empty definition under Coverity since the __ASSERT__
284checks often check for things that Really Cannot Happen, and Coverity
285detects that and gets all excited. */
3e94db23
JH
286
287#if defined(DEBUGGING) && !defined(__COVERITY__)
0f092d08
KW
288# define __ASSERT_(statement) assert(statement),
289#else
290# define __ASSERT_(statement)
291#endif
292
3fe05580 293/*
a4ee4fb5 294=head1 SV Manipulation Functions
3fe05580 295
3bb9fd01 296=for apidoc Ama|SV*|newSVpvs|"literal string"
1568d13a 297Like C<newSVpvn>, but takes a literal string instead of a
30a15352 298string/length pair.
3fe05580 299
3bb9fd01 300=for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
1568d13a 301Like C<newSVpvn_flags>, but takes a literal string instead of
30a15352 302a string/length pair.
84bafc02 303
3bb9fd01 304=for apidoc Ama|SV*|newSVpvs_share|"literal string"
1568d13a 305Like C<newSVpvn_share>, but takes a literal string instead of
30a15352 306a string/length pair and omits the hash parameter.
3fe05580 307
3bb9fd01 308=for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
1568d13a 309Like C<sv_catpvn_flags>, but takes a literal string instead
30a15352 310of a string/length pair.
9dcc53ea 311
3bb9fd01 312=for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
1568d13a 313Like C<sv_catpvn_nomg>, but takes a literal string instead of
0c395ea5 314a string/length pair.
9dcc53ea 315
3bb9fd01 316=for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
1568d13a 317Like C<sv_catpvn>, but takes a literal string instead of a
0c395ea5 318string/length pair.
3fe05580 319
3bb9fd01 320=for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
1568d13a 321Like C<sv_catpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
322string/length pair.
323
3bb9fd01 324=for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
1568d13a 325Like C<sv_setpvn>, but takes a literal string instead of a
0c395ea5 326string/length pair.
3fe05580 327
3bb9fd01 328=for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
1568d13a 329Like C<sv_setpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
330string/length pair.
331
3bb9fd01 332=for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
1568d13a 333Like C<sv_setref_pvn>, but takes a literal string instead of
0c395ea5 334a string/length pair.
9dcc53ea 335
3fe05580
MHM
336=head1 Memory Management
337
3bb9fd01 338=for apidoc Ama|char*|savepvs|"literal string"
1568d13a 339Like C<savepvn>, but takes a literal string instead of a
30a15352 340string/length pair.
3fe05580 341
3bb9fd01 342=for apidoc Ama|char*|savesharedpvs|"literal string"
9dcc53ea
Z
343A version of C<savepvs()> which allocates the duplicate string in memory
344which is shared between threads.
345
3fe05580
MHM
346=head1 GV Functions
347
3bb9fd01 348=for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
1568d13a 349Like C<gv_stashpvn>, but takes a literal string instead of a
0c395ea5 350string/length pair.
3fe05580
MHM
351
352=head1 Hash Manipulation Functions
353
3bb9fd01 354=for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
1568d13a 355Like C<hv_fetch>, but takes a literal string instead of a
0c395ea5 356string/length pair.
3fe05580 357
3bb9fd01 358=for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
1568d13a 359Like C<hv_store>, but takes a literal string instead of a
0c395ea5 360string/length pair
3fe05580
MHM
361and omits the hash parameter.
362
510966aa
Z
363=head1 Lexer interface
364
3bb9fd01 365=for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
510966aa 366
1568d13a 367Like L</lex_stuff_pvn>, but takes a literal string instead of
0c395ea5 368a string/length pair.
510966aa 369
3fe05580
MHM
370=cut
371*/
372
a34e53fc
KW
373/*
374=head1 Handy Values
2efa8cc7 375
a34e53fc
KW
376=for apidoc Amu|pair|STR_WITH_LEN|"literal string"
377
378Returns two comma separated tokens of the input literal string, and its length.
379This is convenience macro which helps out in some API calls.
380Note that it can't be used as an argument to macros or functions that under
381some configurations might be macros, which means that it requires the full
382Perl_xxx(aTHX_ ...) form for any API calls where it's used.
383
384=cut
385*/
386
387
388#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
ba3a79e7
GA
389
390/* STR_WITH_LEN() shortcuts */
391#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
392#define newSVpvs_flags(str,flags) \
393 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 394#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
395#define sv_catpvs_flags(sv, str, flags) \
396 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
397#define sv_catpvs_nomg(sv, str) \
398 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
399#define sv_catpvs(sv, str) \
400 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
401#define sv_catpvs_mg(sv, str) \
402 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 403#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
404#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
405#define sv_setref_pvs(rv, classname, str) \
406 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 407#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
408#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
409#define gv_stashpvs(str, create) \
410 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
411#define gv_fetchpvs(namebeg, add, sv_type) \
412 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
413#define gv_fetchpvn(namebeg, len, add, sv_type) \
414 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
415#define sv_catxmlpvs(dsv, str, utf8) \
416 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 417
ba3a79e7 418
510966aa
Z
419#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
420
b96d8cd9
NC
421#define get_cvs(str, flags) \
422 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 423
954c1994 424/*
ccfc67b7
JH
425=head1 Miscellaneous Functions
426
954c1994 427=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
428Test two C<NUL>-terminated strings to see if they are different. Returns true
429or false.
954c1994
GS
430
431=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
432Test two C<NUL>-terminated strings to see if they are equal. Returns true or
433false.
954c1994
GS
434
435=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
436Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
437second, C<s2>. Returns true or false.
954c1994
GS
438
439=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
440Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
441equal to the second, C<s2>. Returns true or false.
954c1994
GS
442
443=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
444Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
445the second, C<s2>. Returns true or false.
954c1994
GS
446
447=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
448Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
449or equal to the second, C<s2>. Returns true or false.
954c1994
GS
450
451=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
452Test two C<NUL>-terminated strings to see if they are different. The C<len>
453parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
454wrapper for C<strncmp>).
455
456=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
457Test two C<NUL>-terminated strings to see if they are equal. The C<len>
458parameter indicates the number of bytes to compare. Returns true or false. (A
459wrapper for C<strncmp>).
954c1994 460
bd18bd40
KW
461=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
462Test two buffers (which may contain embedded C<NUL> characters, to see if they
463are equal. The C<len> parameter indicates the number of bytes to compare.
464Returns zero if equal, or non-zero if non-equal.
465
3bb9fd01 466=for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
467Like L</memEQ>, but the second string is a literal enclosed in double quotes,
468C<l1> gives the number of bytes in C<s1>.
469Returns zero if equal, or non-zero if non-equal.
470
bd18bd40
KW
471=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
472Test two buffers (which may contain embedded C<NUL> characters, to see if they
473are not equal. The C<len> parameter indicates the number of bytes to compare.
474Returns zero if non-equal, or non-zero if equal.
475
3bb9fd01 476=for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
477Like L</memNE>, but the second string is a literal enclosed in double quotes,
478C<l1> gives the number of bytes in C<s1>.
479Returns zero if non-equal, or zero if non-equal.
480
954c1994 481=cut
fc169e00
KW
482
483New macros should use the following conventions for their names (which are
484based on the underlying C library functions):
485
486 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
487
488 Each has two main parameters, string-like operands that are compared
489 against each other, as specified by the macro name. Some macros may
490 additionally have one or potentially even two length parameters. If a length
491 parameter applies to both string parameters, it will be positioned third;
492 otherwise any length parameter immediately follows the string parameter it
493 applies to.
494
495 If the prefix to the name is 'str', the string parameter is a pointer to a C
496 language string. Such a string does not contain embedded NUL bytes; its
497 length may be unknown, but can be calculated by C<strlen()>, since it is
498 terminated by a NUL, which isn't included in its length.
499
500 The optional 'n' following 'str' means that that there is a third parameter,
501 giving the maximum number of bytes to look at in each string. Even if both
502 strings are longer than the length parameter, those extra bytes will be
503 unexamined.
504
505 The 's' suffix means that the 2nd byte string parameter is a literal C
506 double-quoted string. Its length will automatically be calculated by the
507 macro, so no length parameter will ever be needed for it.
508
509 If the prefix is 'mem', the string parameters don't have to be C strings;
510 they may contain embedded NUL bytes, do not necessarily have a terminating
511 NUL, and their lengths can be known only through other means, which in
512 practice are additional parameter(s) passed to the function. All 'mem'
513 functions have at least one length parameter. Barring any 'l' or 's' suffix,
514 there is a single length parameter, in position 3, which applies to both
515 string parameters. The 's' suffix means, as described above, that the 2nd
516 string is a literal double-quoted C string (hence its length is calculated by
517 the macro, and the length parameter to the function applies just to the first
518 string parameter, and hence is positioned just after it). An 'l' suffix
519 means that the 2nd string parameter has its own length parameter, and the
520 signature will look like memFOOl(s1, l1, s2, l2).
521
522 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
523 substring of the 1st string. 'P' if present indicates that the substring
524 must be a "proper" one in tha mathematical sense that the first one must be
525 strictly larger than the 2nd.
526
954c1994
GS
527*/
528
62946e08 529
75400963
KW
530#define strNE(s1,s2) (strcmp(s1,s2) != 0)
531#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
532#define strLT(s1,s2) (strcmp(s1,s2) < 0)
533#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
534#define strGT(s1,s2) (strcmp(s1,s2) > 0)
535#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 536
75400963
KW
537#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
538#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 539
9d3980bc
KW
540#define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
541#define memNE(s1,s2,l) (! memEQ(s1,s2,l))
36477c24 542
085b7534 543/* memEQ and memNE where second comparand is a string constant */
568a785a 544#define memEQs(s1, l, s2) \
777fa2cb 545 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 546#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 547
fdbb9a7c
KW
548/* Keep these private until we decide it was a good idea */
549#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
550
551#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
552
bdb7e3f0 553#define memBEGINs(s1, l, s2) \
30a6480c 554 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 555 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158 556#define memBEGINPs(s1, l, s2) \
30a6480c 557 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
de627158 558 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0 559#define memENDs(s1, l, s2) \
30a6480c 560 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 561 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424 562#define memENDPs(s1, l, s2) \
30a6480c 563 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
b80f8424 564 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 565#endif /* End of making macros private */
bdb7e3f0 566
062b6850
KW
567#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
568#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
569#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
570#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
571
bbce6d69
PP
572/*
573 * Character classes.
574 *
575 * Unfortunately, the introduction of locales means that we
576 * can't trust isupper(), etc. to tell the truth. And when
577 * it comes to /\w+/ with tainting enabled, we *must* be able
578 * to trust our character classes.
579 *
580 * Therefore, the default tests in the text of Perl will be
581 * independent of locale. Any code that wants to depend on
582 * the current locale will use the tests that begin with "lc".
583 */
584
2304df62
AD
585#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
586# ifndef CTYPE256
587# define CTYPE256
588# endif
589#endif
590
954c1994 591/*
ccfc67b7 592
dcccc8ff 593=head1 Character classification
243effed
KW
594This section is about functions (really macros) that classify characters
595into types, such as punctuation versus alphabetic, etc. Most of these are
596analogous to regular expression character classes. (See
597L<perlrecharclass/POSIX Character Classes>.) There are several variants for
598each class. (Not all macros have all variants; each item below lists the
599ones valid for it.) None are affected by C<use bytes>, and only the ones
600with C<LC> in the name are affected by the current locale.
601
d713f9d9
KW
602The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
603treating it as a code point, and returns a boolean as to whether or not the
604character represented by it is (or on non-ASCII platforms, corresponds to) an
6aff1f14
KW
605ASCII character in the named class based on platform, Unicode, and Perl rules.
606If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 607
c98722a4 608Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
609with no suffix C<"_A">. This variant is used to emphasize by its name that
610only ASCII-range characters can return TRUE.
4b9734bf 611
d60679e1 612Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
613onto the platform. That is, the code points that are ASCII are unaffected,
614since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
615as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
616true when called with the code point 0xDF, which is a word character in both
4650c663 617ASCII and EBCDIC (though it represents different characters in each).
d713f9d9
KW
618If the input is a number that doesn't fit in an octet, FALSE is returned.
619(Perl's documentation uses a colloquial definition of Latin-1, to include all
620code points below 256.)
243effed 621
d713f9d9
KW
622Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
623inputs below 256, but if the code point is larger than 255, Unicode rules are
624used to determine if it is in the character class. For example,
d0da05db 625C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 626WITH MACRON in Unicode, and is a word character.
243effed 627
c98722a4 628Variant C<isI<FOO>_utf8_safe> is like C<isI<FOO>_uvchr>, but is used for UTF-8
d713f9d9
KW
629encoded strings. Each call classifies the first character of the string. This
630variant takes two parameters. The first, C<p>, is a
da8c1a98
KW
631pointer to the first byte of the character to be classified. (Recall that it
632may take more than one byte to represent a character in UTF-8 strings.) The
633second parameter, C<e>, points to anywhere in the string beyond the first
634character, up to one byte past the end of the entire string. The suffix
635C<_safe> in the function's name indicates that it will not attempt to read
636beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
637is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
638character is malformed in some way, the program may croak, or the function may
639return FALSE, at the discretion of the implementation, and subject to change in
640future releases.
641
c98722a4 642Variant C<isI<FOO>_utf8> is like C<isI<FOO>_utf8_safe>, but takes just a single
da8c1a98 643parameter, C<p>, which has the same meaning as the corresponding parameter does
c98722a4 644in C<isI<FOO>_utf8_safe>. The function therefore can't check if it is reading
8b4b30c5 645beyond the end of the string. Starting in Perl v5.32, it will take a second
c98722a4 646parameter, becoming a synonym for C<isI<FOO>_utf8_safe>. At that time every
34aeb2e9 647program that uses it will have to be changed to successfully compile. In the
c98722a4 648meantime, the first runtime call to C<isI<FOO>_utf8> from each call point in the
34aeb2e9 649program will raise a deprecation warning, enabled by default. You can convert
c98722a4 650your program now to use C<isI<FOO>_utf8_safe>, and avoid the warnings, and get an
8b4b30c5 651extra measure of protection, or you can wait until v5.32, when you'll be forced
34aeb2e9 652to add the C<e> parameter.
243effed 653
d713f9d9
KW
654Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
655but the result is based on the current locale, which is what C<LC> in the name
656stands for. If Perl can determine that the current locale is a UTF-8 locale,
657it uses the published Unicode rules; otherwise, it uses the C library function
658that gives the named classification. For example, C<isDIGIT_LC()> when not in
659a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
660returned if the input won't fit into an octet. On some platforms where the C
661library function is known to be defective, Perl changes its result to follow
662the POSIX standard's rules.
243effed 663
d713f9d9
KW
664Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
665than 256, but for larger ones it returns the Unicode classification of the code
666point.
243effed 667
c98722a4 668Variant C<isI<FOO>_LC_utf8_safe> is like C<isI<FOO>_LC_uvchr>, but is used for UTF-8
d713f9d9
KW
669encoded strings. Each call classifies the first character of the string. This
670variant takes two parameters. The first, C<p>, is a pointer to the first byte
671of the character to be classified. (Recall that it may take more than one byte
672to represent a character in UTF-8 strings.) The second parameter, C<e>,
673points to anywhere in the string beyond the first character, up to one byte
674past the end of the entire string. The suffix C<_safe> in the function's name
675indicates that it will not attempt to read beyond S<C<e - 1>>, provided that
676the constraint S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING>
677builds). If the UTF-8 for the input character is malformed in some way, the
678program may croak, or the function may return FALSE, at the discretion of the
679implementation, and subject to change in future releases.
da8c1a98 680
c98722a4 681Variant C<isI<FOO>_LC_utf8> is like C<isI<FOO>_LC_utf8_safe>, but takes just a single
da8c1a98 682parameter, C<p>, which has the same meaning as the corresponding parameter does
c98722a4 683in C<isI<FOO>_LC_utf8_safe>. The function therefore can't check if it is reading
8b4b30c5 684beyond the end of the string. Starting in Perl v5.32, it will take a second
c98722a4 685parameter, becoming a synonym for C<isI<FOO>_LC_utf8_safe>. At that time every
34aeb2e9 686program that uses it will have to be changed to successfully compile. In the
c98722a4 687meantime, the first runtime call to C<isI<FOO>_LC_utf8> from each call point in
34aeb2e9 688the program will raise a deprecation warning, enabled by default. You can
c98722a4 689convert your program now to use C<isI<FOO>_LC_utf8_safe>, and avoid the warnings,
8b4b30c5 690and get an extra measure of protection, or you can wait until v5.32, when
34aeb2e9 691you'll be forced to add the C<e> parameter.
ccfc67b7 692
d713f9d9
KW
693=for apidoc Am|bool|isALPHA|int ch
694Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
695analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
696See the L<top of this section|/Character classification> for an explanation of
697variants
da8c1a98
KW
698C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8_safe>,
699C<isALPHA_LC>, C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8_safe>.
8a58bdcf 700
f16858ed
KW
701=cut
702
703Here and below, we add the protoypes of these macros for downstream programs
704that would be interested in them, such as Devel::PPPort
705
706=for apidoc Amh|bool|isALPHA_A|int ch
707=for apidoc Amh|bool|isALPHA_L1|int ch
708=for apidoc Amh|bool|isALPHA_uvchr|int ch
709=for apidoc Amh|bool|isALPHA_utf8_safe|U8 * s|U8 * end
710=for apidoc Amh|bool|isALPHA_utf8|U8 * s
711=for apidoc Amh|bool|isALPHA_LC|int ch
712=for apidoc Amh|bool|isALPHA_LC_uvchr|int ch
713=for apidoc Amh|bool|isALPHA_LC_utf8_safe|U8 * s| U8 *end
714
d713f9d9
KW
715=for apidoc Am|bool|isALPHANUMERIC|int ch
716Returns a boolean indicating whether the specified character is one of
717C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
718See the L<top of this section|/Character classification> for an explanation of
719variants
d0da05db 720C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
da8c1a98
KW
721C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>,
722and C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 723
255b632a
KW
724A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
725this corresponds to the C language alphanumeric definition). Also
726there are the variants
727C<isALNUMC_A>, C<isALNUMC_L1>
728C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
729
f16858ed
KW
730=for apidoc Amh|bool|isALPHANUMERIC_A|int ch
731=for apidoc Amh|bool|isALPHANUMERIC_L1|int ch
732=for apidoc Amh|bool|isALPHANUMERIC_uvchr|int ch
733=for apidoc Amh|bool|isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
734=for apidoc Amh|bool|isALPHANUMERIC_utf8|U8 * s
735=for apidoc Amh|bool|isALPHANUMERIC_LC|int ch
736=for apidoc Amh|bool|isALPHANUMERIC_LC_uvchr|int ch
737=for apidoc Amh|bool|isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
738=for apidoc Amh|bool|isALNUMC|int ch
739=for apidoc Amh|bool|isALNUMC_A|int ch
740=for apidoc Amh|bool|isALNUMC_L1|int ch
741=for apidoc Amh|bool|isALNUMC_LC|int ch
742=for apidoc Amh|bool|isALNUMC_LC_uvchr|int ch
743
d713f9d9 744=for apidoc Am|bool|isASCII|int ch
8a58bdcf 745Returns a boolean indicating whether the specified character is one of the 128
243effed 746characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 747On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
748character corresponds to an ASCII character. Variants C<isASCII_A()> and
749C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
750See the L<top of this section|/Character classification> for an explanation of
751variants
da8c1a98
KW
752C<isASCII_uvchr>, C<isASCII_utf8_safe>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
753C<isASCII_LC_utf8_safe>. Note, however, that some platforms do not have the C
243effed
KW
754library routine C<isascii()>. In these cases, the variants whose names contain
755C<LC> are the same as the corresponding ones without.
756
f16858ed
KW
757=for apidoc Amh|bool|isASCII_A|int ch
758=for apidoc Amh|bool|isASCII_L1|int ch
759=for apidoc Amh|bool|isASCII_uvchr|int ch
760=for apidoc Amh|bool|isASCII_utf8_safe|U8 * s|U8 * end
761=for apidoc Amh|bool|isASCII_utf8|U8 * s
762=for apidoc Amh|bool|isASCII_LC|int ch
763=for apidoc Amh|bool|isASCII_LC_uvchr|int ch
764=for apidoc Amh|bool|isASCII_LC_utf8_safe|U8 * s| U8 *end
765
d98532ea
KW
766Also note, that because all ASCII characters are UTF-8 invariant (meaning they
767have the exact same representation (always a single byte) whether encoded in
768UTF-8 or not), C<isASCII> will give the correct results when called with any
da8c1a98
KW
769byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8_safe>
770will work properly on any string encoded or not in UTF-8.
d98532ea 771
243effed
KW
772=for apidoc Am|bool|isBLANK|char ch
773Returns a boolean indicating whether the specified character is a
6aff1f14 774character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
775See the L<top of this section|/Character classification> for an explanation of
776variants
da8c1a98
KW
777C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8_safe>,
778C<isBLANK_LC>, C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8_safe>. Note,
779however, that some platforms do not have the C library routine
780C<isblank()>. In these cases, the variants whose names contain C<LC> are
781the same as the corresponding ones without.
243effed 782
f16858ed
KW
783=for apidoc Amh|bool|isBLANK_A|int ch
784=for apidoc Amh|bool|isBLANK_L1|int ch
785=for apidoc Amh|bool|isBLANK_uvchr|int ch
786=for apidoc Amh|bool|isBLANK_utf8_safe|U8 * s|U8 * end
787=for apidoc Amh|bool|isBLANK_utf8|U8 * s
788=for apidoc Amh|bool|isBLANK_LC|int ch
789=for apidoc Amh|bool|isBLANK_LC_uvchr|int ch
790=for apidoc Amh|bool|isBLANK_LC_utf8_safe|U8 * s| U8 *end
791
243effed
KW
792=for apidoc Am|bool|isCNTRL|char ch
793Returns a boolean indicating whether the specified character is a
6aff1f14 794control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
795See the L<top of this section|/Character classification> for an explanation of
796variants
da8c1a98
KW
797C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8_safe>,
798C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8_safe> On EBCDIC
799platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 800
f16858ed
KW
801=for apidoc Amh|bool|isCNTRL_A|int ch
802=for apidoc Amh|bool|isCNTRL_L1|int ch
803=for apidoc Amh|bool|isCNTRL_uvchr|int ch
804=for apidoc Amh|bool|isCNTRL_utf8_safe|U8 * s|U8 * end
805=for apidoc Amh|bool|isCNTRL_utf8|U8 * s
806=for apidoc Amh|bool|isCNTRL_LC|int ch
807=for apidoc Amh|bool|isCNTRL_LC_uvchr|int ch
808=for apidoc Amh|bool|isCNTRL_LC_utf8_safe|U8 * s| U8 *end
809
954c1994 810=for apidoc Am|bool|isDIGIT|char ch
2787a470 811Returns a boolean indicating whether the specified character is a
6aff1f14 812digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 813Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
814See the L<top of this section|/Character classification> for an explanation of
815variants
da8c1a98
KW
816C<isDIGIT_uvchr>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
817C<isDIGIT_LC_utf8_safe>.
243effed 818
f16858ed
KW
819=for apidoc Amh|bool|isDIGIT_A|int ch
820=for apidoc Amh|bool|isDIGIT_L1|int ch
821=for apidoc Amh|bool|isDIGIT_uvchr|int ch
822=for apidoc Amh|bool|isDIGIT_utf8_safe|U8 * s|U8 * end
823=for apidoc Amh|bool|isDIGIT_utf8|U8 * s
824=for apidoc Amh|bool|isDIGIT_LC|int ch
825=for apidoc Amh|bool|isDIGIT_LC_uvchr|int ch
826=for apidoc Amh|bool|isDIGIT_LC_utf8_safe|U8 * s| U8 *end
827
243effed
KW
828=for apidoc Am|bool|isGRAPH|char ch
829Returns a boolean indicating whether the specified character is a
6aff1f14 830graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 831See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
832variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8_safe>,
833C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8_safe>.
954c1994 834
f16858ed
KW
835=for apidoc Amh|bool|isGRAPH_A|int ch
836=for apidoc Amh|bool|isGRAPH_L1|int ch
837=for apidoc Amh|bool|isGRAPH_uvchr|int ch
838=for apidoc Amh|bool|isGRAPH_utf8_safe|U8 * s|U8 * end
839=for apidoc Amh|bool|isGRAPH_utf8|U8 * s
840=for apidoc Amh|bool|isGRAPH_LC|int ch
841=for apidoc Amh|bool|isGRAPH_LC_uvchr|int ch
842=for apidoc Amh|bool|isGRAPH_LC_utf8_safe|U8 * s| U8 *end
843
0c82b6df 844=for apidoc Am|bool|isLOWER|char ch
2787a470 845Returns a boolean indicating whether the specified character is a
6aff1f14 846lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
847See the L<top of this section|/Character classification> for an explanation of
848variants
da8c1a98
KW
849C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8_safe>,
850C<isLOWER_LC>, C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8_safe>.
0c82b6df 851
f16858ed
KW
852=for apidoc Amh|bool|isLOWER_A|int ch
853=for apidoc Amh|bool|isLOWER_L1|int ch
854=for apidoc Amh|bool|isLOWER_uvchr|int ch
855=for apidoc Amh|bool|isLOWER_utf8_safe|U8 * s|U8 * end
856=for apidoc Amh|bool|isLOWER_utf8|U8 * s
857=for apidoc Amh|bool|isLOWER_LC|int ch
858=for apidoc Amh|bool|isLOWER_LC_uvchr|int ch
859=for apidoc Amh|bool|isLOWER_LC_utf8_safe|U8 * s| U8 *end
860
c99e91e9 861=for apidoc Am|bool|isOCTAL|char ch
2787a470 862Returns a boolean indicating whether the specified character is an
6aff1f14 863octal digit, [0-7].
243effed
KW
864The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
865C<isOCTAL>.
866
f16858ed
KW
867=for apidoc Amh|bool|isOCTAL_A|int ch
868=for apidoc Amh|bool|isOCTAL_L1|int ch
869
243effed
KW
870=for apidoc Am|bool|isPUNCT|char ch
871Returns a boolean indicating whether the specified character is a
6aff1f14
KW
872punctuation character, analogous to C<m/[[:punct:]]/>.
873Note that the definition of what is punctuation isn't as
243effed
KW
874straightforward as one might desire. See L<perlrecharclass/POSIX Character
875Classes> for details.
dcccc8ff 876See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
877variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8_safe>,
878C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8_safe>.
c99e91e9 879
f16858ed
KW
880=for apidoc Amh|bool|isPUNCT_A|int ch
881=for apidoc Amh|bool|isPUNCT_L1|int ch
882=for apidoc Amh|bool|isPUNCT_uvchr|int ch
883=for apidoc Amh|bool|isPUNCT_utf8_safe|U8 * s|U8 * end
884=for apidoc Amh|bool|isPUNCT_utf8|U8 * s
885=for apidoc Amh|bool|isPUNCT_LC|int ch
886=for apidoc Amh|bool|isPUNCT_LC_uvchr|int ch
887=for apidoc Amh|bool|isPUNCT_LC_utf8_safe|U8 * s| U8 *end
888
0c82b6df 889=for apidoc Am|bool|isSPACE|char ch
2787a470 890Returns a boolean indicating whether the specified character is a
6aff1f14 891whitespace character. This is analogous
398d098a 892to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 893this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
894locale forms of this macro (the ones with C<LC> in their names) matched
895precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
896in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
897(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
898See the L<top of this section|/Character classification> for an explanation of
899variants
da8c1a98
KW
900C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8_safe>,
901C<isSPACE_LC>, C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8_safe>.
0c82b6df 902
f16858ed
KW
903=for apidoc Amh|bool|isSPACE_A|int ch
904=for apidoc Amh|bool|isSPACE_L1|int ch
905=for apidoc Amh|bool|isSPACE_uvchr|int ch
906=for apidoc Amh|bool|isSPACE_utf8_safe|U8 * s|U8 * end
907=for apidoc Amh|bool|isSPACE_utf8|U8 * s
908=for apidoc Amh|bool|isSPACE_LC|int ch
909=for apidoc Amh|bool|isSPACE_LC_uvchr|int ch
910=for apidoc Amh|bool|isSPACE_LC_utf8_safe|U8 * s| U8 *end
911
398d098a
KW
912=for apidoc Am|bool|isPSXSPC|char ch
913(short for Posix Space)
779cf272
KW
914Starting in 5.18, this is identical in all its forms to the
915corresponding C<isSPACE()> macros.
398d098a
KW
916The locale forms of this macro are identical to their corresponding
917C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
918non-locale forms differ from their C<isSPACE()> forms only in that the
919C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
920Otherwise they are identical. Thus this macro is analogous to what
921C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 922See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
923variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8_safe>,
924C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 925
f16858ed
KW
926=for apidoc Amh|bool|isPSXSPC_A|int ch
927=for apidoc Amh|bool|isPSXSPC_L1|int ch
928=for apidoc Amh|bool|isPSXSPC_uvchr|int ch
929=for apidoc Amh|bool|isPSXSPC_utf8_safe|U8 * s|U8 * end
930=for apidoc Amh|bool|isPSXSPC_utf8|U8 * s
931=for apidoc Amh|bool|isPSXSPC_LC|int ch
932=for apidoc Amh|bool|isPSXSPC_LC_uvchr|int ch
933=for apidoc Amh|bool|isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
934
954c1994 935=for apidoc Am|bool|isUPPER|char ch
2787a470 936Returns a boolean indicating whether the specified character is an
6aff1f14 937uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 938See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
939variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8_safe>,
940C<isUPPER_LC>, C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8_safe>.
954c1994 941
f16858ed
KW
942=for apidoc Amh|bool|isUPPER_A|int ch
943=for apidoc Amh|bool|isUPPER_L1|int ch
944=for apidoc Amh|bool|isUPPER_uvchr|int ch
945=for apidoc Amh|bool|isUPPER_utf8_safe|U8 * s|U8 * end
946=for apidoc Amh|bool|isUPPER_utf8|U8 * s
947=for apidoc Amh|bool|isUPPER_LC|int ch
948=for apidoc Amh|bool|isUPPER_LC_uvchr|int ch
949=for apidoc Amh|bool|isUPPER_LC_utf8_safe|U8 * s| U8 *end
950
243effed 951=for apidoc Am|bool|isPRINT|char ch
8eea39dd 952Returns a boolean indicating whether the specified character is a
6aff1f14 953printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
954See the L<top of this section|/Character classification> for an explanation of
955variants
da8c1a98
KW
956C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8_safe>,
957C<isPRINT_LC>, C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8_safe>.
243effed 958
f16858ed
KW
959=for apidoc Amh|bool|isPRINT_A|int ch
960=for apidoc Amh|bool|isPRINT_L1|int ch
961=for apidoc Amh|bool|isPRINT_uvchr|int ch
962=for apidoc Amh|bool|isPRINT_utf8_safe|U8 * s|U8 * end
963=for apidoc Amh|bool|isPRINT_utf8|U8 * s
964=for apidoc Amh|bool|isPRINT_LC|int ch
965=for apidoc Amh|bool|isPRINT_LC_uvchr|int ch
966=for apidoc Amh|bool|isPRINT_LC_utf8_safe|U8 * s| U8 *end
967
243effed
KW
968=for apidoc Am|bool|isWORDCHAR|char ch
969Returns a boolean indicating whether the specified character is a character
970that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
971in a regular expression. A word character is an alphabetic character, a
972decimal digit, a connecting punctuation character (such as an underscore), or
973a "mark" character that attaches to one of those (like some sort of accent).
974C<isALNUM()> is a synonym provided for backward compatibility, even though a
975word character includes more than the standard C language meaning of
976alphanumeric.
dcccc8ff 977See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
978variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and
979C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and
980C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
981include the platform's native underscore.
8a58bdcf 982
f16858ed
KW
983=for apidoc Amh|bool|isWORDCHAR_A|int ch
984=for apidoc Amh|bool|isWORDCHAR_L1|int ch
985=for apidoc Amh|bool|isWORDCHAR_uvchr|int ch
986=for apidoc Amh|bool|isWORDCHAR_utf8_safe|U8 * s|U8 * end
987=for apidoc Amh|bool|isWORDCHAR_utf8|U8 * s
988=for apidoc Amh|bool|isWORDCHAR_LC|int ch
989=for apidoc Amh|bool|isWORDCHAR_LC_uvchr|int ch
990=for apidoc Amh|bool|isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
991=for apidoc Amh|bool|isALNUM|int ch
992=for apidoc Amh|bool|isALNUM_A|int ch
993=for apidoc Amh|bool|isALNUM_LC|int ch
994=for apidoc Amh|bool|isALNUM_LC_uvchr|int ch
995
8a58bdcf
KW
996=for apidoc Am|bool|isXDIGIT|char ch
997Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
998digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
999and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
1000See the L<top of this section|/Character classification> for an explanation of
1001variants
da8c1a98
KW
1002C<isXDIGIT_uvchr>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>,
1003and C<isXDIGIT_LC_utf8_safe>.
243effed 1004
f16858ed
KW
1005=for apidoc Amh|bool|isXDIGIT_A|int ch
1006=for apidoc Amh|bool|isXDIGIT_L1|int ch
1007=for apidoc Amh|bool|isXDIGIT_uvchr|int ch
1008=for apidoc Amh|bool|isXDIGIT_utf8_safe|U8 * s|U8 * end
1009=for apidoc Amh|bool|isXDIGIT_utf8|U8 * s
1010=for apidoc Amh|bool|isXDIGIT_LC|int ch
1011=for apidoc Amh|bool|isXDIGIT_LC_uvchr|int ch
1012=for apidoc Amh|bool|isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1013
3c3ecf18
KW
1014=for apidoc Am|bool|isIDFIRST|char ch
1015Returns a boolean indicating whether the specified character can be the first
1016character of an identifier. This is very close to, but not quite the same as
1017the official Unicode property C<XID_Start>. The difference is that this
1018returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
1019See the L<top of this section|/Character classification> for an explanation of
1020variants
da8c1a98
KW
1021C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8_safe>,
1022C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18 1023
f16858ed
KW
1024=for apidoc Amh|bool|isIDFIRST_A|int ch
1025=for apidoc Amh|bool|isIDFIRST_L1|int ch
1026=for apidoc Amh|bool|isIDFIRST_uvchr|int ch
1027=for apidoc Amh|bool|isIDFIRST_utf8_safe|U8 * s|U8 * end
1028=for apidoc Amh|bool|isIDFIRST_utf8|U8 * s
1029=for apidoc Amh|bool|isIDFIRST_LC|int ch
1030=for apidoc Amh|bool|isIDFIRST_LC_uvchr|int ch
1031=for apidoc Amh|bool|isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1032
3c3ecf18
KW
1033=for apidoc Am|bool|isIDCONT|char ch
1034Returns a boolean indicating whether the specified character can be the
1035second or succeeding character of an identifier. This is very close to, but
1036not quite the same as the official Unicode property C<XID_Continue>. The
1037difference is that this returns true only if the input character also matches
dcccc8ff
KW
1038L</isWORDCHAR>. See the L<top of this section|/Character classification> for
1039an
d0da05db 1040explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
da8c1a98
KW
1041C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
1042C<isIDCONT_LC_utf8_safe>.
3c3ecf18 1043
f16858ed
KW
1044=for apidoc Amh|bool|isIDCONT_A|int ch
1045=for apidoc Amh|bool|isIDCONT_L1|int ch
1046=for apidoc Amh|bool|isIDCONT_uvchr|int ch
1047=for apidoc Amh|bool|isIDCONT_utf8_safe|U8 * s|U8 * end
1048=for apidoc Amh|bool|isIDCONT_utf8|U8 * s
1049=for apidoc Amh|bool|isIDCONT_LC|int ch
1050=for apidoc Amh|bool|isIDCONT_LC_uvchr|int ch
1051=for apidoc Amh|bool|isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1052
243effed 1053=head1 Miscellaneous Functions
8eea39dd 1054
95a59cab 1055=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 1056Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
1057Behaviour is only well defined when isXDIGIT(*str) is true.
1058
e7c1e6c1 1059=head1 Character case changing
21da7284
KW
1060Perl uses "full" Unicode case mappings. This means that converting a single
1061character to another case may result in a sequence of more than one character.
1062For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1063character sequence C<SS>. This presents some complications The lowercase of
1064all characters in the range 0..255 is a single character, and thus
1065C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1066return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1067an API that does allow every possible legal result to be returned.) Likewise
1068no other function that is crippled by not being able to give the correct
1069results for the full range of possible inputs has been implemented here.
e7c1e6c1 1070
d713f9d9 1071=for apidoc Am|U8|toUPPER|int ch
1f607577
KW
1072Converts the specified character to uppercase. If the input is anything but an
1073ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 1074C<toUPPER_A> is equivalent.
954c1994 1075
d0da05db
KW
1076=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1077Converts the code point C<cp> to its uppercase version, and
1078stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1079point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1080that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1081bytes since the uppercase version may be longer than the original character.
1082
1083The first code point of the uppercased version is returned
21da7284
KW
1084(but note, as explained at L<the top of this section|/Character case
1085changing>, that there may be more.)
1f607577 1086
a239b1e2
KW
1087=for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1088Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1089extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
1090stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1091that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1092bytes since the uppercase version may be longer than the original character.
1093
1094The first code point of the uppercased version is returned
21da7284
KW
1095(but note, as explained at L<the top of this section|/Character case
1096changing>, that there may be more).
1f607577 1097
a239b1e2
KW
1098The suffix C<_safe> in the function's name indicates that it will not attempt
1099to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1100true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1101input character is malformed in some way, the program may croak, or the
1102function may return the REPLACEMENT CHARACTER, at the discretion of the
1103implementation, and subject to change in future releases.
1104
1105=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
1106This is like C<L</toUPPER_utf8_safe>>, but doesn't have the C<e>
1107parameter The function therefore can't check if it is reading
8b4b30c5 1108beyond the end of the string. Starting in Perl v5.32, it will take the C<e>
607313a1
KW
1109parameter, becoming a synonym for C<toUPPER_utf8_safe>. At that time every
1110program that uses it will have to be changed to successfully compile. In the
1111meantime, the first runtime call to C<toUPPER_utf8> from each call point in the
1112program will raise a deprecation warning, enabled by default. You can convert
1113your program now to use C<toUPPER_utf8_safe>, and avoid the warnings, and get an
8b4b30c5 1114extra measure of protection, or you can wait until v5.32, when you'll be forced
607313a1 1115to add the C<e> parameter.
1f607577 1116
25200305
KW
1117=for apidoc Am|U8|toFOLD|U8 ch
1118Converts the specified character to foldcase. If the input is anything but an
1119ASCII uppercase character, that input character itself is returned. Variant
1120C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 1121Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 1122
d0da05db
KW
1123=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1124Converts the code point C<cp> to its foldcase version, and
1125stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1126point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1127that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1128bytes since the foldcase version may be longer than the original character.
1129
1130The first code point of the foldcased version is returned
21da7284
KW
1131(but note, as explained at L<the top of this section|/Character case
1132changing>, that there may be more).
1f607577 1133
a239b1e2
KW
1134=for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1135Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1136extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
1137stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1138that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1139bytes since the foldcase version may be longer than the original character.
1140
1141The first code point of the foldcased version is returned
21da7284
KW
1142(but note, as explained at L<the top of this section|/Character case
1143changing>, that there may be more).
1f607577 1144
a239b1e2
KW
1145The suffix C<_safe> in the function's name indicates that it will not attempt
1146to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1147true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1148input character is malformed in some way, the program may croak, or the
1149function may return the REPLACEMENT CHARACTER, at the discretion of the
1150implementation, and subject to change in future releases.
1151
1152=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
1153This is like C<L</toFOLD_utf8_safe>>, but doesn't have the C<e>
1154parameter The function therefore can't check if it is reading
8b4b30c5 1155beyond the end of the string. Starting in Perl v5.32, it will take the C<e>
607313a1
KW
1156parameter, becoming a synonym for C<toFOLD_utf8_safe>. At that time every
1157program that uses it will have to be changed to successfully compile. In the
1158meantime, the first runtime call to C<toFOLD_utf8> from each call point in the
1159program will raise a deprecation warning, enabled by default. You can convert
1160your program now to use C<toFOLD_utf8_safe>, and avoid the warnings, and get an
8b4b30c5 1161extra measure of protection, or you can wait until v5.32, when you'll be forced
607313a1 1162to add the C<e> parameter.
1f607577
KW
1163
1164=for apidoc Am|U8|toLOWER|U8 ch
1165Converts the specified character to lowercase. If the input is anything but an
1166ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 1167C<toLOWER_A> is equivalent.
954c1994 1168
1f607577 1169=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
1170Converts the specified Latin1 character to lowercase. The results are
1171undefined if the input doesn't fit in a byte.
1f607577
KW
1172
1173=for apidoc Am|U8|toLOWER_LC|U8 ch
1174Converts the specified character to lowercase using the current locale's rules,
1175if possible; otherwise returns the input character itself.
1176
d0da05db
KW
1177=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1178Converts the code point C<cp> to its lowercase version, and
1179stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1180point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1181that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1182bytes since the lowercase version may be longer than the original character.
1183
1184The first code point of the lowercased version is returned
21da7284
KW
1185(but note, as explained at L<the top of this section|/Character case
1186changing>, that there may be more).
1f607577 1187
a239b1e2
KW
1188
1189=for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1190Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1191extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
1192stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1193that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1194bytes since the lowercase version may be longer than the original character.
1195
1196The first code point of the lowercased version is returned
21da7284
KW
1197(but note, as explained at L<the top of this section|/Character case
1198changing>, that there may be more).
1f607577 1199
a239b1e2
KW
1200The suffix C<_safe> in the function's name indicates that it will not attempt
1201to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1202true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1203input character is malformed in some way, the program may croak, or the
1204function may return the REPLACEMENT CHARACTER, at the discretion of the
1205implementation, and subject to change in future releases.
1206
1207=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
1208This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1209parameter The function therefore can't check if it is reading
8b4b30c5 1210beyond the end of the string. Starting in Perl v5.32, it will take the C<e>
607313a1
KW
1211parameter, becoming a synonym for C<toLOWER_utf8_safe>. At that time every
1212program that uses it will have to be changed to successfully compile. In the
1213meantime, the first runtime call to C<toLOWER_utf8> from each call point in the
1214program will raise a deprecation warning, enabled by default. You can convert
1215your program now to use C<toLOWER_utf8_safe>, and avoid the warnings, and get an
8b4b30c5 1216extra measure of protection, or you can wait until v5.32, when you'll be forced
607313a1 1217to add the C<e> parameter.
1f607577 1218
25200305
KW
1219=for apidoc Am|U8|toTITLE|U8 ch
1220Converts the specified character to titlecase. If the input is anything but an
1221ASCII lowercase character, that input character itself is returned. Variant
b7d90381 1222C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 1223range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 1224not a concept used in locale handling, so there is no functionality for that.)
25200305 1225
d0da05db
KW
1226=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1227Converts the code point C<cp> to its titlecase version, and
1228stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1229point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1230that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1231bytes since the titlecase version may be longer than the original character.
1232
1233The first code point of the titlecased version is returned
21da7284
KW
1234(but note, as explained at L<the top of this section|/Character case
1235changing>, that there may be more).
1f607577 1236
a239b1e2
KW
1237=for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1238Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1239extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1240stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1241that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1242bytes since the titlecase version may be longer than the original character.
1243
1244The first code point of the titlecased version is returned
21da7284
KW
1245(but note, as explained at L<the top of this section|/Character case
1246changing>, that there may be more).
1f607577 1247
a239b1e2
KW
1248The suffix C<_safe> in the function's name indicates that it will not attempt
1249to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1250true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1251input character is malformed in some way, the program may croak, or the
1252function may return the REPLACEMENT CHARACTER, at the discretion of the
1253implementation, and subject to change in future releases.
1254
1255=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
1256This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1257parameter The function therefore can't check if it is reading
8b4b30c5 1258beyond the end of the string. Starting in Perl v5.32, it will take the C<e>
607313a1
KW
1259parameter, becoming a synonym for C<toTITLE_utf8_safe>. At that time every
1260program that uses it will have to be changed to successfully compile. In the
1261meantime, the first runtime call to C<toTITLE_utf8> from each call point in the
1262program will raise a deprecation warning, enabled by default. You can convert
1263your program now to use C<toTITLE_utf8_safe>, and avoid the warnings, and get an
8b4b30c5 1264extra measure of protection, or you can wait until v5.32, when you'll be forced
607313a1 1265to add the C<e> parameter.
1f607577 1266
954c1994 1267=cut
353c9b6f 1268
d0da05db 1269XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1270really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1271and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1272
8a58bdcf 1273Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1274patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1275
954c1994
GS
1276*/
1277
8f5283f4
KW
1278/*
1279 void below because that's the best fit, and works for Devel::PPPort
1280=for apidoc AmnU|void|WIDEST_UTYPE
1281
1282Yields the widest unsigned integer type on the platform, currently either
1283C<U32> or C<64>. This can be used in declarations such as
1284
1285 WIDEST_UTYPE my_uv;
1286
1287or casts
1288
1289 my_uv = (WIDEST_UTYPE) val;
1290
1291=cut
1292
1293*/
de9e2639
KW
1294#ifdef QUADKIND
1295# define WIDEST_UTYPE U64
7c062697
KW
1296#else
1297# define WIDEST_UTYPE U32
1298#endif
1299
3912bc88
KW
1300/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1301 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1302 * bits of information are lost even on a 64-bit machine, but to get the
1303 * compiler to optimize it out if possible. This is because Configure makes
1304 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1305 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1306 *
1307 * For Coverity, be always true, because otherwise Coverity thinks
1308 * it finds several expressions that are always true, independent
1309 * of operands. Well, they are, but that is kind of the point.
220c71bf 1310 */
7e75d1a1 1311#ifndef __COVERITY__
6c5b02ac
KW
1312 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1313 * pointer) */
1314#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1315 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1316#else
1317#define FITS_IN_8_BITS(c) (1)
1318#endif
cf301eb7 1319
45f4bb73 1320/* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
833b0f46 1321 * Written this way so that after optimization, only one conditional test is
76d3ad4c
KW
1322 * needed. (The NV casts stop any warnings about comparison always being true
1323 * if called with an unsigned. The cast preserves the sign, which is all we
1324 * care about.) */
1325#define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1326 __ASSERT_((NV) (n) >= 0) \
1327 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
833b0f46 1328
94250c4f
KW
1329/* Returns true if c is in the range l..u, where 'l' is non-negative
1330 * Written this way so that after optimization, only one conditional test is
4758c20d 1331 * needed. */
1eaefa6e 1332#define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
4758c20d
KW
1333 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
1334 : (sizeof(c) == sizeof(U16)) ? withinCOUNT(((U16) (c)), (l), ((u) - (l))) \
1335 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1336 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
45f4bb73 1337 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
305fe86e 1338
41f43cc2 1339#ifdef EBCDIC
b6340bd0 1340# ifndef _ALL_SOURCE
0852beac
KW
1341 /* The native libc isascii() et.al. functions return the wrong results
1342 * on at least z/OS unless this is defined. */
b6340bd0
KW
1343# error _ALL_SOURCE should probably be defined
1344# endif
41f43cc2 1345#else
0852beac
KW
1346 /* There is a simple definition of ASCII for ASCII platforms. But the
1347 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1348 * other macros below.
3f3c579d
KW
1349 *
1350 * The cast here is used instead of '(c) >= 0', because some compilers emit
1351 * a warning that that test is always true when the parameter is an
1352 * unsigned type. khw supposes that it could be written as
1353 * && ((c) == '\0' || (c) > 0)
1354 * to avoid the message, but the cast will likely avoid extra branches even
1355 * with stupid compilers.
1356 *
1357 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1358 * a pointer) */
9c903d59 1359# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1360#endif
1361
38694112
KW
1362/* Take the eight possible bit patterns of the lower 3 bits and you get the
1363 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1364 * can be ignored. If the rest match '0', we have an octal */
1365#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1366
9fb1bf9d 1367#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1368
a500dc72
KW
1369/* Character class numbers. For internal core Perl use only. The ones less
1370 * than 32 are used in PL_charclass[] and the ones up through the one that
1371 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1372 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1373 * their actual definitions are here. If that file has a name not used here,
1374 * it won't compile.
1709d539
KW
1375 *
1376 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561 1377 * frequency of their use. This gives a slight edge to exiting a loop earlier
58a3ba2c
KW
1378 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1379 * real Posix character class, and some (small) inefficiencies in regular
1380 * expression handling would be introduced by putting it in the middle of those
1381 * that are. Also, cntrl and ascii come after the others as it may be useful
1382 * to group these which have no members that match above Latin1, (or above
1383 * ASCII in the latter case) */
1384
1709d539
KW
1385# define _CC_WORDCHAR 0 /* \w and [:word:] */
1386# define _CC_DIGIT 1 /* \d and [:digit:] */
1387# define _CC_ALPHA 2 /* [:alpha:] */
1388# define _CC_LOWER 3 /* [:lower:] */
1389# define _CC_UPPER 4 /* [:upper:] */
1390# define _CC_PUNCT 5 /* [:punct:] */
1391# define _CC_PRINT 6 /* [:print:] */
15861f94 1392# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1393# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1394# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
779cf272 1395# define _CC_SPACE 10 /* \s, [:space:] */
34aeb2e9
KW
1396# define _CC_PSXSPC _CC_SPACE /* XXX Temporary, can be removed
1397 when the deprecated isFOO_utf8()
1398 functions are removed */
b0d691b2
KW
1399# define _CC_BLANK 11 /* [:blank:] */
1400# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1401# define _CC_CNTRL 13 /* [:cntrl:] */
1402# define _CC_ASCII 14 /* [:ascii:] */
1403# define _CC_VERTSPACE 15 /* \v */
1709d539 1404
a0947d7b
KW
1405# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1406
1709d539 1407/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1408 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1409# define _CC_IDFIRST 16
1410# define _CC_CHARNAME_CONT 17
1411# define _CC_NONLATIN1_FOLD 18
1412# define _CC_NONLATIN1_SIMPLE_FOLD 19
1413# define _CC_QUOTEMETA 20
1414# define _CC_NON_FINAL_FOLD 21
1415# define _CC_IS_IN_SOME_FOLD 22
1416# define _CC_MNEMONIC_CNTRL 23
073c22b3 1417
34aeb2e9
KW
1418# define _CC_IDCONT 24 /* XXX Temporary, can be removed when the deprecated
1419 isFOO_utf8() functions are removed */
1420
073c22b3
KW
1421/* This next group is only used on EBCDIC platforms, so theoretically could be
1422 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
1423# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1424# define _CC_UTF8_IS_START 29
1425# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1426# define _CC_UTF8_IS_CONTINUATION 31
1427/* Unused: 24-27
f4cdb42c
KW
1428 * If more bits are needed, one could add a second word for non-64bit
1429 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1430 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1431 * is used only for optimization (as of this writing), and differs in the
1432 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1433 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1434 * would just cause /i regexes which match them to run less efficiently.
1435 * Similarly the EBCDIC-only bits are used just for speed, and could be
1436 * replaced by other means */
96ac0975 1437
3a371f2f
KW
1438#if defined(PERL_CORE) || defined(PERL_EXT)
1439/* An enum version of the character class numbers, to help compilers
1440 * optimize */
1441typedef enum {
3a371f2f 1442 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1443 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1444 _CC_ENUM_ASCII = _CC_ASCII,
1445 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1446 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1447 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1448 _CC_ENUM_DIGIT = _CC_DIGIT,
1449 _CC_ENUM_GRAPH = _CC_GRAPH,
1450 _CC_ENUM_LOWER = _CC_LOWER,
1451 _CC_ENUM_PRINT = _CC_PRINT,
1452 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1453 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1454 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1455 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1456 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1457 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1458} _char_class_number;
1459#endif
1460
86f72d56 1461#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1462
6635f04f 1463START_EXTERN_C
96ac0975
NC
1464# ifdef DOINIT
1465EXTCONST U32 PL_charclass[] = {
1466# include "l1_char_class_tab.h"
1467};
1468
1469# else /* ! DOINIT */
1470EXTCONST U32 PL_charclass[];
1471# endif
6635f04f 1472END_EXTERN_C
96ac0975 1473
265c1f46 1474 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1475# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1476
1477 /* For internal core Perl use only: the base macro for defining macros like
1478 * isALPHA */
ff7ecfc3 1479# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1480 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1481
f4cdb42c
KW
1482 /* The mask for the _A versions of the macros; it just adds in the bit for
1483 * ASCII. */
1484# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1485
4650c663
KW
1486 /* For internal core Perl use only: the base macro for defining macros like
1487 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1488 * the ASCII bit are present */
b7d90381
KW
1489# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1490 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1491 == _CC_mask_A(classnum)))
f4cdb42c 1492
26c1d9d8
KW
1493/* On ASCII platforms certain classes form a single range. It's faster to
1494 * special case these. isDIGIT is a single range on all platforms */
b877c1ff
KW
1495# ifdef EBCDIC
1496# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1497# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1498# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1499# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1500# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1501# else
26c1d9d8 1502 /* By folding the upper and lowercase, we can use a single range */
b877c1ff 1503# define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
26c1d9d8 1504# define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
b877c1ff
KW
1505# define isLOWER_A(c) inRANGE(c, 'a', 'z')
1506# define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1507# define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1508# endif
15861f94 1509# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1510# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1511# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
b877c1ff 1512# define isDIGIT_A(c) inRANGE(c, '0', '9')
f4cdb42c
KW
1513# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1514# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
f4cdb42c 1515# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1516# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1517 */
d95f8b6a 1518# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1519# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1520# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1521# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1522
1523 /* continuation character for legal NAME in \N{NAME} */
1524# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1525
1526# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1527# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1528# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1529# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1530# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1531# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1532# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1533# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1534# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1535# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1536
0852beac
KW
1537# ifdef EBCDIC
1538# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1539# endif
1540
f12c0118
KW
1541 /* Participates in a single-character fold with a character above 255 */
1542# 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)))
1543
1544 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1545# 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 1546
4eeeb416 1547# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1548# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1549 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1550# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1551 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1552# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1553 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1554#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1555
1556 /* If we don't have perl.h, we are compiling a utility program. Below we
1557 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1558 * to it. Most are coded based on first principles. These are written to
74665a89 1559 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
182c4ace 1560# define isDIGIT_A(c) inRANGE(c, '0', '9')
0852beac 1561# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1562# define isSPACE_A(c) (isBLANK_A(c) \
1563 || (c) == '\n' \
1564 || (c) == '\r' \
1565 || (c) == '\v' \
0852beac 1566 || (c) == '\f')
74665a89
KW
1567 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1568 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1569 * performance (if optimization isn't on), and allow the same code to be
1570 * used for both platform types */
182c4ace
KW
1571# define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1572 || inRANGE((c), 'j', 'r') \
1573 || inRANGE((c), 's', 'z')
1574# define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1575 || inRANGE((c), 'J', 'R') \
1576 || inRANGE((c), 'S', 'Z')
a4d7a999
KW
1577# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1578# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1579# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1580# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
182c4ace
KW
1581# define isXDIGIT_A(c) ( isDIGIT_A(c) \
1582 || inRANGE((c), 'a', 'f') \
1583 || inRANGE((c), 'A', 'F')
74665a89
KW
1584# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1585 || (c) == '#' || (c) == '$' || (c) == '%' \
1586 || (c) == '&' || (c) == '\'' || (c) == '(' \
1587 || (c) == ')' || (c) == '*' || (c) == '+' \
1588 || (c) == ',' || (c) == '.' || (c) == '/' \
1589 || (c) == ':' || (c) == ';' || (c) == '<' \
1590 || (c) == '=' || (c) == '>' || (c) == '?' \
1591 || (c) == '@' || (c) == '[' || (c) == '\\' \
1592 || (c) == ']' || (c) == '^' || (c) == '_' \
1593 || (c) == '`' || (c) == '{' || (c) == '|' \
1594 || (c) == '}' || (c) == '~')
1595# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1596# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1597
0852beac 1598# ifdef EBCDIC
74665a89
KW
1599 /* The below is accurate for the 3 EBCDIC code pages traditionally
1600 * supported by perl. The only difference between them in the controls
1601 * is the position of \n, and that is represented symbolically below */
1602# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1603 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1604 || (c) == '\t' || (c) == '\v' \
182c4ace 1605 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
74665a89 1606 || (c) == 7 /* U+7F DEL */ \
182c4ace
KW
1607 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1608 DC[1-3] */ \
74665a89
KW
1609 || (c) == 0x18 /* U+18 CAN */ \
1610 || (c) == 0x19 /* U+19 EOM */ \
182c4ace 1611 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
74665a89
KW
1612 || (c) == 0x26 /* U+17 ETB */ \
1613 || (c) == 0x27 /* U+1B ESC */ \
1614 || (c) == 0x2D /* U+05 ENQ */ \
1615 || (c) == 0x2E /* U+06 ACK */ \
1616 || (c) == 0x32 /* U+16 SYN */ \
1617 || (c) == 0x37 /* U+04 EOT */ \
1618 || (c) == 0x3C /* U+14 DC4 */ \
1619 || (c) == 0x3D /* U+15 NAK */ \
1620 || (c) == 0x3F)/* U+1A SUB */
0852beac 1621# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1622# else /* isASCII is already defined for ASCII platforms, so can use that to
1623 define isCNTRL */
1624# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1625# endif
1626
3ffc8c70 1627 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1628 * during debugging, and it seems best to keep them. We may be called
1629 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1630 * do anything anyway, so make it not a problem */
1631# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1632# define NATIVE_TO_LATIN1(ch) (ch)
1633# endif
3ded5eb0
KW
1634# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1635# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1636# define isBLANK_L1(c) (isBLANK_A(c) \
1637 || (FITS_IN_8_BITS(c) \
1638 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1639# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1640# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1641# define isLOWER_L1(c) (isLOWER_A(c) \
1642 || (FITS_IN_8_BITS(c) \
ae683a5f 1643 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1644 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1645 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1646 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1647 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1648# define isPRINT_L1(c) (isPRINT_A(c) \
1649 || (FITS_IN_8_BITS(c) \
1650 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1651# define isPUNCT_L1(c) (isPUNCT_A(c) \
1652 || (FITS_IN_8_BITS(c) \
ae683a5f 1653 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1654 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1655 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1656 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1657 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1658 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1659 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1660# define isSPACE_L1(c) (isSPACE_A(c) \
1661 || (FITS_IN_8_BITS(c) \
ae683a5f 1662 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1663 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1664# define isUPPER_L1(c) (isUPPER_A(c) \
1665 || (FITS_IN_8_BITS(c) \
182c4ace
KW
1666 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1667 0xC0, 0xDE) \
3ded5eb0
KW
1668 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1669# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1670# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1671# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1672 || isBLANK_L1(c) \
1673 || (c) == '-' \
1674 || (c) == '(' \
1675 || (c) == ')')
1676 /* The following are not fully accurate in the above-ASCII range. I (khw)
1677 * don't think it's necessary to be so for the purposes where this gets
1678 * compiled */
1679# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1680# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1681
1682 /* And these aren't accurate at all. They are useful only for above
1683 * Latin1, which utilities and bootstrapping don't deal with */
1684# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1685# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1686# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1687
1688 /* Many of the macros later in this file are defined in terms of these. By
1689 * implementing them with a function, which converts the class number into
1690 * a call to the desired macro, all of the later ones work. However, that
1691 * function won't be actually defined when building a utility program (no
1692 * perl.h), and so a compiler error will be generated if one is attempted
1693 * to be used. And the above-Latin1 code points require Unicode tables to
1694 * be present, something unlikely to be the case when bootstrapping */
1695# define _generic_isCC(c, classnum) \
1696 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1697# define _generic_isCC_A(c, classnum) \
1698 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1699#endif /* End of no perl.h H_PERL */
8a58bdcf 1700
e66b99e9
KW
1701#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1702#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1703#define isASCII_A(c) isASCII(c)
1704#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1705#define isBLANK(c) isBLANK_A(c)
1706#define isCNTRL(c) isCNTRL_A(c)
1707#define isDIGIT(c) isDIGIT_A(c)
1708#define isGRAPH(c) isGRAPH_A(c)
1709#define isIDFIRST(c) isIDFIRST_A(c)
1710#define isLOWER(c) isLOWER_A(c)
1711#define isPRINT(c) isPRINT_A(c)
779cf272 1712#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1713#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1714#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1715#define isPUNCT(c) isPUNCT_A(c)
1716#define isSPACE(c) isSPACE_A(c)
1717#define isUPPER(c) isUPPER_A(c)
1718#define isWORDCHAR(c) isWORDCHAR_A(c)
1719#define isXDIGIT(c) isXDIGIT_A(c)
1720
1721/* ASCII casing. These could also be written as
1722 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1723 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1724 which uses table lookup and mask instead of subtraction. (This would
c5e9991e
KW
1725 work because the _MOD does not apply in the ASCII range).
1726
1727 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1728 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1729 there non-ASCII invariants, and all of them are controls.) */
68067e4e
DM
1730#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1731#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1732
25200305
KW
1733/* In the ASCII range, these are equivalent to what they're here defined to be.
1734 * But by creating these definitions, other code doesn't have to be aware of
c5e9991e
KW
1735 * this detail. Actually this works for all UTF-8 invariants, not just the
1736 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
25200305 1737#define toFOLD(c) toLOWER(c)
25200305
KW
1738#define toTITLE(c) toUPPER(c)
1739
c753c8d3
KW
1740#define toLOWER_A(c) toLOWER(c)
1741#define toUPPER_A(c) toUPPER(c)
25200305
KW
1742#define toFOLD_A(c) toFOLD(c)
1743#define toTITLE_A(c) toTITLE(c)
1a0901db 1744
4650c663 1745/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1746#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1747 ? (c) \
f4cd282c 1748 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1749#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1750
1a0901db 1751/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1752 * all mapped to one of them, and these need special handling; returns the
1753 * input itself if is out-of-range */
b2bf251f 1754#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1755 ? (c) \
f4cd282c 1756 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1757#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1758
beab9ebe
KW
1759/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1760
1761/* For internal core Perl use only: the base macro for defining macros like
1762 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1763 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1764 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1765 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1766 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1767 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1768 * code points under locale rules, use the macros based on _generic_LC_uvchr
1769 * instead of this. */
beab9ebe
KW
1770#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1771 (! FITS_IN_8_BITS(c) \
1772 ? 0 \
31f05a37
KW
1773 : IN_UTF8_CTYPE_LOCALE \
1774 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1775 : cBOOL(non_utf8))
1776
1777/* For internal core Perl use only: a helper macro for defining macros like
1778 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1779 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1780 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1781 * _generic_LC_base for more info */
1782#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1783 _generic_LC_base(c,utf8_locale_classnum, \
1784 non_utf8_func( (_LC_CAST) (c)))
1785
1786/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1787 * 'c' is the platform's native underscore character */
1788#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1789 _generic_LC_base(c, utf8_locale_classnum, \
1790 (non_utf8_func( (_LC_CAST) (c)) \
1791 || (char)(c) == '_'))
1792
1793/* These next three are also for internal core Perl use only: case-change
247985d4
KW
1794 * helper macros. The reason for using the PL_latin arrays is in case the
1795 * system function is defective; it ensures uniform results that conform to the
b257a28c 1796 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
beab9ebe
KW
1797#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1798 ? (c) \
31f05a37
KW
1799 : (IN_UTF8_CTYPE_LOCALE) \
1800 ? PL_latin1_lc[ (U8) (c) ] \
5a10328c 1801 : (cast)function((cast)(c)))
beab9ebe 1802
31f05a37
KW
1803/* Note that the result can be larger than a byte in a UTF-8 locale. It
1804 * returns a single value, so can't adequately return the upper case of LATIN
1805 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1806 * values "SS"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1807 * otherwise returns its input. It does not handle the anomalies in UTF-8
1808 * Turkic locales. */
beab9ebe
KW
1809#define _generic_toUPPER_LC(c, function, cast) \
1810 (! FITS_IN_8_BITS(c) \
1811 ? (c) \
31f05a37 1812 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1813 ? (cast)function((cast)(c)) \
31f05a37
KW
1814 : ((((U8)(c)) == MICRO_SIGN) \
1815 ? GREEK_CAPITAL_LETTER_MU \
1816 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1817 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1818 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1819 ? (__ASSERT_(0) (c)) \
1820 : PL_mod_latin1_uc[ (U8) (c) ])))))
1821
1822/* Note that the result can be larger than a byte in a UTF-8 locale. It
1823 * returns a single value, so can't adequately return the fold case of LATIN
1824 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1825 * values "ss"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1826 * otherwise returns its input. It does not handle the anomalies in UTF-8
1827 * Turkic locales */
beab9ebe 1828#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1829 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1830 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1831 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1832 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1833 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1834
84061b6a 1835/* Use the libc versions for these if available. */
f05550c0 1836#if defined(HAS_ISASCII)
84061b6a
KW
1837# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1838#else
1839# define isASCII_LC(c) isASCII(c)
1840#endif
1841
f05550c0 1842#if defined(HAS_ISBLANK)
84061b6a 1843# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1844#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1845# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1846#endif
1847
f05550c0 1848#define _LC_CAST U8
bbce6d69 1849
f05550c0 1850#ifdef WIN32
375f5f06
KW
1851 /* The Windows functions don't bother to follow the POSIX standard, which
1852 * for example says that something can't both be a printable and a control.
1853 * But Windows treats the \t control as a printable, and does such things
1854 * as making superscripts into both digits and punctuation. This tames
1855 * these flaws by assuming that the definitions of both controls and space
1856 * are correct, and then making sure that other definitions don't have
1857 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1858 * Not all possible weirdnesses are checked for, just the ones that were
1859 * detected on actual Microsoft code pages */
1860
b7d90381
KW
1861# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1862# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1863
1864# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1865 && isALPHANUMERIC_LC(c))
1866# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1867 ! isPUNCT_LC(c))
1868# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1869 isALPHANUMERIC_LC(c))
1870# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1871# define isIDFIRST_LC(c) (((c) == '_') \
1872 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1873# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1874# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1875# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1876# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1877# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1878# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1879 && isALPHANUMERIC_LC(c))
f05550c0
BF
1880
1881# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1882# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1883# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1884
1885#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1886 /* For most other platforms */
beab9ebe 1887
f05550c0
BF
1888# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1889# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1890# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1891# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1892# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1893# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1894# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1895# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1896# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1897# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1898# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1899# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1900# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1901
1902
1903# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1904# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1905# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1906
1907#else /* The final fallback position */
1908
b7d90381
KW
1909# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1910# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1911# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1912# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1913# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1914# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1915# define isLOWER_LC(c) (isascii(c) && islower(c))
1916# define isPRINT_LC(c) (isascii(c) && isprint(c))
1917# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1918# define isSPACE_LC(c) (isascii(c) && isspace(c))
1919# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1920# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1921# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1922
1923# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1924# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1925# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1926
f05550c0 1927#endif
55204971 1928
eba68aa0
KW
1929#define isIDCONT(c) isWORDCHAR(c)
1930#define isIDCONT_A(c) isWORDCHAR_A(c)
1931#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1932#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1933#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1934
4650c663 1935/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1936 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1937 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1938 * classes where there is a macro or function 'above_latin1' that takes the
1939 * single argument 'c' and returns the desired value. These exist for those
1940 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1941 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1942 * for classes where that overhead is faster than a direct lookup.
d0da05db 1943 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1944 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1945 * have duplicate checks here, so could create versions of the macros that
1946 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1947
1948/* Note that all ignore 'use bytes' */
1e222e4f
KW
1949#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1950 ? _generic_isCC(c, classnum) \
cd500f2f 1951 : above_latin1(c))
1e222e4f
KW
1952#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
1953 ? _generic_isCC(c, classnum) \
922e8cb4 1954 : _is_uni_FOO(classnum, c))
d0da05db
KW
1955#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1956#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1957#define isASCII_uvchr(c) isASCII(c)
1958#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1959#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1960#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1961#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1962#define isIDCONT_uvchr(c) \
1963 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1964#define isIDFIRST_uvchr(c) \
1965 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
d0da05db
KW
1966#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1967#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1968
1969#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1970#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1971#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1972
1973#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1974#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1975#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1976#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1977
1978#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1979#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1980#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1981#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1982
1983/* For backwards compatibility, even though '_uni' should mean official Unicode
1984 * code points, in Perl it means native for those below 256 */
1985#define isALPHA_uni(c) isALPHA_uvchr(c)
1986#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1987#define isASCII_uni(c) isASCII_uvchr(c)
1988#define isBLANK_uni(c) isBLANK_uvchr(c)
1989#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1990#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1991#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1992#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1993#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1994#define isLOWER_uni(c) isLOWER_uvchr(c)
1995#define isPRINT_uni(c) isPRINT_uvchr(c)
1996#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1997#define isSPACE_uni(c) isSPACE_uvchr(c)
1998#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1999#define isUPPER_uni(c) isUPPER_uvchr(c)
2000#define isVERTWS_uni(c) isVERTWS_uvchr(c)
2001#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2002#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2003#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2004#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2005#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2006#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 2007
4650c663
KW
2008/* For internal core Perl use only: the base macros for defining macros like
2009 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 2010 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 2011 * simple class definitions; the other for more complex. These are like
d0da05db 2012 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
2013#define _generic_LC_uvchr(latin1, above_latin1, c) \
2014 (c < 256 ? latin1(c) : above_latin1(c))
2015#define _generic_LC_swash_uvchr(latin1, classnum, c) \
2016 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2017
2018#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
2019#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
2020 _CC_ALPHANUMERIC, c)
b7d90381
KW
2021#define isASCII_LC_uvchr(c) isASCII_LC(c)
2022#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2023 is_HORIZWS_cp_high, c)
feeab5a9 2024#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
2025#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2026#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 2027#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 2028 _is_uni_perl_idcont, c)
b7d90381 2029#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
2030 _is_uni_perl_idstart, c)
2031#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
2032#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 2033#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 2034#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 2035#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 2036 is_XPERLSPACE_cp_high, c)
cd500f2f 2037#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 2038#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 2039 _CC_WORDCHAR, c)
b7d90381
KW
2040#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2041 is_XDIGIT_cp_high, c)
e712593e 2042
b7d90381 2043#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 2044
4650c663
KW
2045/* For internal core Perl use only: the base macros for defining macros like
2046 * isALPHA_utf8. These are like the earlier defined macros, but take an input
2047 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
2048 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
2049 * 'utf8' parameter. This relies on the fact that ASCII characters have the
2050 * same representation whether utf8 or not. Note that it assumes that the utf8
2051 * has been validated, and ignores 'use bytes' */
34aeb2e9
KW
2052#define _base_generic_utf8(enum_name, name, p, use_locale ) \
2053 _is_utf8_FOO(CAT2(_CC_, enum_name), \
2054 (const U8 *) p, \
2055 "is" STRINGIFY(name) "_utf8", \
2056 "is" STRINGIFY(name) "_utf8_safe", \
2057 1, use_locale, __FILE__,__LINE__)
2058
2059#define _generic_utf8(name, p) _base_generic_utf8(name, name, p, 0)
da8c1a98
KW
2060
2061/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2062 * they don't otherwise go out of their way to look for malformed UTF-8. If
2063 * they can return accurate results without knowing if the input is otherwise
2064 * malformed, they do so. For example isASCII is accurate in spite of any
2065 * non-length malformations because it looks only at a single byte. Likewise
2066 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2067 * variant ones return FALSE. But, if the input has to be well-formed in order
2068 * for the results to be accurate, the macros will test and if malformed will
2069 * call a routine to die
2070 *
2071 * Except for toke.c, the macros do assume that e > p, asserting that on
2072 * DEBUGGING builds. Much code that calls these depends on this being true,
2073 * for other reasons. toke.c is treated specially as using the regular
2074 * assertion breaks it in many ways. All strings that these operate on there
2075 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2076 * bunch of code in toke.c assumes that this is true, so the assertion allows
2077 * for that */
2078#ifdef PERL_IN_TOKE_C
2079# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2080#else
2081# define _utf8_safe_assert(p,e) ((e) > (p))
2082#endif
2083
2084#define _generic_utf8_safe(classnum, p, e, above_latin1) \
c81b3562
KW
2085 ((! _utf8_safe_assert(p, e)) \
2086 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2087 : (UTF8_IS_INVARIANT(*(p))) \
da8c1a98
KW
2088 ? _generic_isCC(*(p), classnum) \
2089 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2090 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2091 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2092 classnum) \
2093 : (_force_out_malformed_utf8_message( \
2094 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2095 : above_latin1))
b7d90381
KW
2096/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2097 * 'above_latin1' can be a macro */
da8c1a98
KW
2098#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2099 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2100#define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \
2101 _generic_utf8_safe(classnum, p, e, \
2102 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2103 ? (_force_out_malformed_utf8_message( \
2104 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2105 : above_latin1(p)))
4650c663 2106/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 2107 * 'above_latin1' parameter */
da8c1a98
KW
2108#define _generic_swash_utf8_safe(classnum, p, e) \
2109_generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e))
922e8cb4 2110
cc8ab7c0 2111/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
2112 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2113 * class is TRUE for. Hence it can skip the tests for this range.
2114 * 'above_latin1' should include its arguments */
da8c1a98
KW
2115#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2116 (__ASSERT_(_utf8_safe_assert(p, e)) \
2117 (UTF8_IS_INVARIANT(*(p))) \
2118 ? _generic_isCC(*(p), classnum) \
2119 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2120 ? 0 /* Note that doesn't check validity for latin1 */ \
2121 : above_latin1)
2122
84238efa 2123
34aeb2e9
KW
2124#define isALPHA_utf8(p) _generic_utf8(ALPHA, p)
2125#define isALPHANUMERIC_utf8(p) _generic_utf8(ALPHANUMERIC, p)
2126#define isASCII_utf8(p) _generic_utf8(ASCII, p)
2127#define isBLANK_utf8(p) _generic_utf8(BLANK, p)
2128#define isCNTRL_utf8(p) _generic_utf8(CNTRL, p)
2129#define isDIGIT_utf8(p) _generic_utf8(DIGIT, p)
2130#define isGRAPH_utf8(p) _generic_utf8(GRAPH, p)
2131#define isIDCONT_utf8(p) _generic_utf8(IDCONT, p)
2132#define isIDFIRST_utf8(p) _generic_utf8(IDFIRST, p)
2133#define isLOWER_utf8(p) _generic_utf8(LOWER, p)
2134#define isPRINT_utf8(p) _generic_utf8(PRINT, p)
2135#define isPSXSPC_utf8(p) _generic_utf8(PSXSPC, p)
2136#define isPUNCT_utf8(p) _generic_utf8(PUNCT, p)
2137#define isSPACE_utf8(p) _generic_utf8(SPACE, p)
2138#define isUPPER_utf8(p) _generic_utf8(UPPER, p)
2139#define isVERTWS_utf8(p) _generic_utf8(VERTSPACE, p)
2140#define isWORDCHAR_utf8(p) _generic_utf8(WORDCHAR, p)
2141#define isXDIGIT_utf8(p) _generic_utf8(XDIGIT, p)
e8fa43e2 2142
da8c1a98
KW
2143#define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e)
2144#define isALPHANUMERIC_utf8_safe(p, e) \
2145 _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e)
2146#define isASCII_utf8_safe(p, e) \
2147 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2148 * works */ \
2149 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2150#define isBLANK_utf8_safe(p, e) \
2151 _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
2152
e8fa43e2
KW
2153#ifdef EBCDIC
2154 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2155 * more efficient macro instead of the more general one */
da8c1a98 2156# define isCNTRL_utf8_safe(p, e) \
56d02b8c 2157 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 2158#else
da8c1a98 2159# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
2160#endif
2161
da8c1a98
KW
2162#define isDIGIT_utf8_safe(p, e) \
2163 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2164 _is_utf8_FOO_with_len(_CC_DIGIT, p, e))
2165#define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e)
2166#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
2167 _is_utf8_perl_idcont_with_len, p, e)
e5dcd934 2168
c11ff943
KW
2169/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2170 * IDFIRST is an alnum. See
c8362b00 2171 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
2172 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2173 * This used to be not the XID version, but we decided to go with the more
2174 * modern Unicode definition */
da8c1a98
KW
2175#define isIDFIRST_utf8_safe(p, e) \
2176 _generic_func_utf8_safe(_CC_IDFIRST, \
2177 _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e))
2178
2179#define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e)
2180#define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e)
2181#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2182#define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e)
2183#define isSPACE_utf8_safe(p, e) \
2184 _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2185#define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e)
2186#define isVERTWS_utf8_safe(p, e) \
2187 _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
2188#define isWORDCHAR_utf8_safe(p, e) \
2189 _generic_swash_utf8_safe(_CC_WORDCHAR, p, e)
2190#define isXDIGIT_utf8_safe(p, e) \
2191 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2192 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2193 ? (_force_out_malformed_utf8_message( \
2194 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2195 : is_XDIGIT_high(p)))
a0ed51b3 2196
7d2b61dd 2197#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
2198#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
2199#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
2200#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 2201
567b353c 2202/* For internal core use only, subject to change */
607313a1
KW
2203#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f, "", 0)
2204#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f, "", 0)
2205#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f, "", 0)
2206#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f, "", 0)
a1a5ec35
KW
2207
2208#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2209#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2210#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2211#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 2212
4650c663
KW
2213/* For internal core Perl use only: the base macros for defining macros like
2214 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
2215 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
2216 * 'macro' parameter */
34aeb2e9
KW
2217#define _generic_LC_utf8(name, p) _base_generic_utf8(name, name, p, 1)
2218
2219#define isALPHA_LC_utf8(p) _generic_LC_utf8(ALPHA, p)
2220#define isALPHANUMERIC_LC_utf8(p) _generic_LC_utf8(ALPHANUMERIC, p)
2221#define isASCII_LC_utf8(p) _generic_LC_utf8(ASCII, p)
2222#define isBLANK_LC_utf8(p) _generic_LC_utf8(BLANK, p)
2223#define isCNTRL_LC_utf8(p) _generic_LC_utf8(CNTRL, p)
2224#define isDIGIT_LC_utf8(p) _generic_LC_utf8(DIGIT, p)
2225#define isGRAPH_LC_utf8(p) _generic_LC_utf8(GRAPH, p)
2226#define isIDCONT_LC_utf8(p) _generic_LC_utf8(IDCONT, p)
2227#define isIDFIRST_LC_utf8(p) _generic_LC_utf8(IDFIRST, p)
2228#define isLOWER_LC_utf8(p) _generic_LC_utf8(LOWER, p)
2229#define isPRINT_LC_utf8(p) _generic_LC_utf8(PRINT, p)
2230#define isPSXSPC_LC_utf8(p) _generic_LC_utf8(PSXSPC, p)
2231#define isPUNCT_LC_utf8(p) _generic_LC_utf8(PUNCT, p)
2232#define isSPACE_LC_utf8(p) _generic_LC_utf8(SPACE, p)
2233#define isUPPER_LC_utf8(p) _generic_LC_utf8(UPPER, p)
2234#define isWORDCHAR_LC_utf8(p) _generic_LC_utf8(WORDCHAR, p)
2235#define isXDIGIT_LC_utf8(p) _generic_LC_utf8(XDIGIT, p)
2236
da8c1a98
KW
2237/* For internal core Perl use only: the base macros for defining macros like
2238 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2239 * point in 'p' is within the 0-255 range, it uses locale rules from the
2240 * passed-in 'macro' parameter */
2241#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2242 (__ASSERT_(_utf8_safe_assert(p, e)) \
2243 (UTF8_IS_INVARIANT(*(p))) \
2244 ? macro(*(p)) \
2245 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2246 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2247 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2248 : (_force_out_malformed_utf8_message( \
2249 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2250 : above_latin1))
2251
2252#define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \
2253 _generic_LC_utf8_safe(macro, p, e, \
2254 _is_utf8_FOO_with_len(classnum, p, e))
2255
2256#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2257 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2258
2259#define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \
2260 _generic_LC_utf8_safe(classnum, p, e, \
2261 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2262 ? (_force_out_malformed_utf8_message( \
2263 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2264 : above_latin1(p)))
2265
2266#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2267 _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \
2268 _CC_ALPHANUMERIC, p, e)
2269#define isALPHA_LC_utf8_safe(p, e) \
2270 _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2271#define isASCII_LC_utf8_safe(p, e) \
2272 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2273#define isBLANK_LC_utf8_safe(p, e) \
2274 _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2275#define isCNTRL_LC_utf8_safe(p, e) \
2276 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2277#define isDIGIT_LC_utf8_safe(p, e) \
2278 _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2279#define isGRAPH_LC_utf8_safe(p, e) \
2280 _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2281#define isIDCONT_LC_utf8_safe(p, e) \
2282 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2283 _is_utf8_perl_idcont_with_len, p, e)
2284#define isIDFIRST_LC_utf8_safe(p, e) \
2285 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2286 _is_utf8_perl_idstart_with_len, p, e)
2287#define isLOWER_LC_utf8_safe(p, e) \
2288 _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2289#define isPRINT_LC_utf8_safe(p, e) \
2290 _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2291#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2292#define isPUNCT_LC_utf8_safe(p, e) \
2293 _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2294#define isSPACE_LC_utf8_safe(p, e) \
2295 _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2296#define isUPPER_LC_utf8_safe(p, e) \
2297 _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2298#define isWORDCHAR_LC_utf8_safe(p, e) \
2299 _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2300#define isXDIGIT_LC_utf8_safe(p, e) \
2301 _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2302
fbc19f27
KW
2303/* Macros for backwards compatibility and for completeness when the ASCII and
2304 * Latin1 values are identical */
b7d90381
KW
2305#define isALPHAU(c) isALPHA_L1(c)
2306#define isDIGIT_L1(c) isDIGIT_A(c)
2307#define isOCTAL(c) isOCTAL_A(c)
2308#define isOCTAL_L1(c) isOCTAL_A(c)
2309#define isXDIGIT_L1(c) isXDIGIT_A(c)
2310#define isALNUM(c) isWORDCHAR(c)
a377c856 2311#define isALNUM_A(c) isALNUM(c)
b7d90381
KW
2312#define isALNUMU(c) isWORDCHAR_L1(c)
2313#define isALNUM_LC(c) isWORDCHAR_LC(c)
2314#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2315#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
2316#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
2317#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
2318#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2319#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2320#define isALNUMC(c) isALPHANUMERIC(c)
2321#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2322#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2323#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 2324#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 2325#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 2326
2bd1cbf6
KW
2327/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2328 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2329 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2330 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2331 * the outlier from the block that contains the other controls, just like
2332 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2333 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2334 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2335 * etc. */
2bd1cbf6 2336#ifndef EBCDIC
75763b3a 2337# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2338#else
75763b3a
KW
2339# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2340 ((isPRINT_A(c)) \
2341 ? (UNLIKELY((c) == '?') \
2342 ? QUESTION_MARK_CTRL \
2343 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2344 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2345 ? '?' \
2346 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2347#endif
bbce6d69 2348
dea28490
JJ
2349/* Line numbers are unsigned, 32 bits. */
2350typedef U32 line_t;
e5dcd934 2351#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2352
91152fc1
DG
2353/* Helpful alias for version prescan */
2354#define is_LAX_VERSION(a,b) \
2355 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2356
2357#define is_STRICT_VERSION(a,b) \
2358 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2359
2360#define BADVERSION(a,b,c) \
2361 if (b) { \
2362 *b = c; \
2363 } \
2364 return a;
8c52afec 2365
cb27eebd
KW
2366/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2367 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2368 * which is advanced. The input is validated only by an assert() in DEBUGGING
2369 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2370 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2371#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2372 ? (c) \
2373 : ((c) + 9))))
902008b8
KW
2374#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2375 ? (*(s)++) \
2376 : (*(s)++ + 9))))
95a59cab 2377
cb27eebd
KW
2378/* Converts a character known to represent an octal digit (0-7) to its numeric
2379 * value. The input is validated only by an assert() in DEBUGGING builds. In
2380 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2381#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2382
305b8651
KW
2383/* Efficiently returns a boolean as to if two native characters are equivalent
2384 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2385 * the ALPHA in the name is to remind you of that. This is asserted() in
2386 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2387 * works (on valid input) for both non- and UTF-8-encoded bytes.
2388 *
2389 * When one of the inputs is a compile-time constant and gets folded by the
2390 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2391 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2392 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2393 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2394 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2395 * just a single 0, in the bit position where the upper- and lowercase differ.
2396 * */
305b8651
KW
2397#define isALPHA_FOLD_EQ(c1, c2) \
2398 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2399 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2400#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2401
8e84507e 2402/*
ccfc67b7
JH
2403=head1 Memory Management
2404
a02a5408 2405=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2406The XSUB-writer's interface to the C C<malloc> function.
2407
596f7718 2408Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2409
c5008215
JC
2410In 5.9.3, Newx() and friends replace the older New() API, and drops
2411the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2412themselves. This aid has been superseded by a new build option,
d10b4965 2413PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2414there for use in XS modules supporting older perls.
2415
a02a5408 2416=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2417The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2418cast. See also C<L</Newx>>.
954c1994 2419
596f7718 2420Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2421
a02a5408 2422=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2423The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2424memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2425
596f7718 2426Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2427
954c1994
GS
2428=for apidoc Am|void|Renew|void* ptr|int nitems|type
2429The XSUB-writer's interface to the C C<realloc> function.
2430
596f7718 2431Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2432
954c1994
GS
2433=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2434The XSUB-writer's interface to the C C<realloc> function, with
2435cast.
2436
596f7718 2437Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2438
49b8b560 2439=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2440The XSUB-writer's interface to the C C<free> function.
2441
596f7718 2442This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2443
954c1994
GS
2444=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2445The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2446source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2447C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2448
e90e2364 2449=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2450Like C<Move> but returns C<dest>. Useful
72d33970 2451for encouraging compilers to tail-call
e90e2364
NC
2452optimise.
2453
954c1994
GS
2454=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2455The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2456source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2457C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2458
e90e2364
NC
2459=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2460
796b6530 2461Like C<Copy> but returns C<dest>. Useful
72d33970 2462for encouraging compilers to tail-call
e90e2364
NC
2463optimise.
2464
954c1994
GS
2465=for apidoc Am|void|Zero|void* dest|int nitems|type
2466
2467The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2468destination, C<nitems> is the number of items, and C<type> is the type.
2469
e90e2364
NC
2470=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2471
72d33970
FC
2472Like C<Zero> but returns dest. Useful
2473for encouraging compilers to tail-call
e90e2364
NC
2474optimise.
2475
da5d8dbb 2476=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2477This is an architecture-independent macro to copy one structure to another.
954c1994 2478
7e337ee0
JH
2479=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2480
2481Fill up memory with a byte pattern (a byte repeated over and over
2482again) that hopefully catches attempts to access uninitialized memory.
2483
2484=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2485
2486PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2487
1c12ffb4 2488=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2489
2490PoisonWith(0xEF) for catching access to freed memory.
2491
9965345d
JH
2492=for apidoc Am|void|Poison|void* dest|int nitems|type
2493
7e337ee0 2494PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2495
2496=cut */
954c1994 2497
561b68a9
SH
2498/* Maintained for backwards-compatibility only. Use newSV() instead. */
2499#ifndef PERL_CORE
ff06c60c 2500#define NEWSV(x,len) newSV(len)
561b68a9 2501#endif
ff06c60c 2502
b7112dce 2503#define MEM_SIZE_MAX ((MEM_SIZE)-1)
19a94d75 2504
a500027b 2505#define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
e6bdf523 2506
27d5b266 2507#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2508
2509/* This expression will be constant-folded at compile time. It checks
2510 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2511 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2512 * As well as avoiding the need for a run-time check in some cases, it's
2513 * designed to avoid compiler warnings like:
2514 * comparison is always false due to limited range of data type
73e8ff00
DM
2515 * It's mathematically equivalent to
2516 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2517 */
2518
2519# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
445198b9
LM
2520 ( sizeof(MEM_SIZE) < sizeof(n) \
2521 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 2522
88f9f128 2523/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2524 * compiler warnings. We *want* to write the expression as
2525 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2526 * (for some compile-time constant C), but even when the LHS
2527 * constant-folds to false at compile-time, g++ insists on emitting
2528 * warnings about the RHS (e.g. "comparison is always false"), so instead
2529 * we write it as
e6bdf523 2530 *
d98e5cde 2531 * (cond ? n : X) > C
88f9f128 2532 *
d98e5cde
DM
2533 * where X is a constant with X > C always false. Choosing a value for X
2534 * is tricky. If 0, some compilers will complain about 0 > C always being
2535 * false; if 1, Coverity complains when n happens to be the constant value
2536 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2537 * for X and hope that nothing else whines.
e6bdf523
DM
2538 */
2539
2540# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2541 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2542 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2543
2544# define MEM_WRAP_CHECK(n,t) \
2545 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2546 && (croak_memory_wrap(),0))
2547
2548# define MEM_WRAP_CHECK_1(n,t,a) \
2549 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2550 && (Perl_croak_nocontext("%s",(a)),0))
2551
814eedc8
DD
2552/* "a" arg must be a string literal */
2553# define MEM_WRAP_CHECK_s(n,t,a) \
2554 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2555 && (Perl_croak_nocontext("" a ""),0))
2556
8b44ba4c 2557#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2558
a500027b 2559#define PERL_STRLEN_ROUNDUP(n) ((void)(((n) > MEM_SIZE_MAX - 2 * PERL_STRLEN_ROUNDUP_QUANTUM) ? (croak_memory_wrap(),0) : 0), _PERL_STRLEN_ROUNDUP_UNCHECKED(n))
27d5b266
JH
2560#else
2561
410319be
NC
2562#define MEM_WRAP_CHECK(n,t)
2563#define MEM_WRAP_CHECK_1(n,t,a)
814eedc8 2564#define MEM_WRAP_CHECK_s(n,t,a)
8b44ba4c
NC
2565#define MEM_WRAP_CHECK_(n,t)
2566
a500027b 2567#define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
27d5b266 2568
1936d2a7 2569#endif
8b44ba4c 2570
fe4f188c 2571#ifdef PERL_MEM_LOG
46c6c7e2 2572/*
9f653bb5 2573 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2574 * go through functions, which are handy for debugging breakpoints, but
2575 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2576 * line number, and C function name if available) passed in. This info can
2577 * then be used for logging the calls, for which one gets a sample
73d1d973 2578 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2579 *
46c6c7e2 2580 * Known problems:
94e892a6 2581 * - not all memory allocs get logged, only those
46c6c7e2 2582 * that go through Newx() and derivatives (while all
94e892a6 2583 * Safefrees do get logged)
46c6c7e2
JH
2584 * - __FILE__ and __LINE__ do not work everywhere
2585 * - __func__ or __FUNCTION__ even less so
2586 * - I think more goes on after the perlio frees but
2587 * the thing is that STDERR gets closed (as do all
2588 * the file descriptors)
2589 * - no deeper calling stack than the caller of the Newx()
2590 * or the kind, but do I look like a C reflection/introspection
2591 * utility to you?
2592 * - the function prototypes for the logging functions
2593 * probably should maybe be somewhere else than handy.h
2594 * - one could consider inlining (macrofying) the logging
2595 * for speed, but I am too lazy
2596 * - one could imagine recording the allocations in a hash,
2597 * (keyed by the allocation address?), and maintain that
2598 * through reallocs and frees, but how to do that without
2599 * any News() happening...?
73d1d973 2600 * - lots of -Ddefines to get useful/controllable output
b953482e 2601 * - lots of ENV reads
46c6c7e2
JH
2602 */
2603
0b0ab801 2604# ifdef PERL_CORE
73d1d973 2605# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2606enum mem_log_type {
2607 MLT_ALLOC,
2608 MLT_REALLOC,
d7a2c63c
MHM
2609 MLT_FREE,
2610 MLT_NEW_SV,
2611 MLT_DEL_SV
0b0ab801
MHM
2612};
2613# endif
12754f92 2614# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2615void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2616void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2617# endif
0b0ab801
MHM
2618# endif
2619
fe4f188c
JH
2620#endif
2621
2622#ifdef PERL_MEM_LOG
d1401ee9
MHM
2623#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2624#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2625#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2626#endif
2627
2628#ifndef MEM_LOG_ALLOC
2629#define MEM_LOG_ALLOC(n,t,a) (a)
2630#endif
2631#ifndef MEM_LOG_REALLOC
2632#define MEM_LOG_REALLOC(n,t,v,a) (a)
2633#endif
2634#ifndef MEM_LOG_FREE
2635#define MEM_LOG_FREE(a) (a)
2636#endif
2637
d1401ee9
MHM
2638#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2639#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2640#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2641
2642#ifndef PERL_CORE
a02a5408
JC
2643/* pre 5.9.x compatibility */
2644#define New(x,v,n,t) Newx(v,n,t)
2645#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2646#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2647#endif
a02a5408 2648
ff68c719 2649#define Renew(v,n,t) \
d1401ee9 2650 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2651#define Renewc(v,n,t,c) \
d1401ee9 2652 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2653
2654#ifdef PERL_POISON
2655#define Safefree(d) \
06c0cc96 2656 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2657#else
fe4f188c 2658#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2659#endif
55497cff 2660
dbb57106
YO
2661/* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2662 * as it handles cases like constant string arguments without throwing warnings *
2663 * the cast is required, as is the inequality check, to avoid warnings */
45908e4d 2664#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2665
45908e4d
YO
2666
2667#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2668#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2669#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2670
bdd1531d 2671/* Like above, but returns a pointer to 'd' */
45908e4d
YO
2672#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2673#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
45908e4d 2674#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2675
7e337ee0
JH
2676#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2677#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2678#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2679#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2680
caa674f3
DD
2681#ifdef PERL_POISON
2682# define PERL_POISON_EXPR(x) x
2683#else
2684# define PERL_POISON_EXPR(x)
2685#endif
2686
ff68c719 2687#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2688
1b7e2294
KW
2689/*
2690=head1 Handy Values
2691
2692=for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2693
2694Returns the number of elements in the input C array (so you want your
2695zero-based indices to be less than but not equal to).
2696
2697=for apidoc Am|void *|C_ARRAY_END|void *a
2698
2699Returns a pointer to one element past the final element of the input C array.
2700
2701=cut
2702
2703C_ARRAY_END is one past the last: half-open/half-closed range, not
2704last-inclusive range.
2705*/
622913ab 2706#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2707#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2708
2cc61e15
DD
2709#ifdef NEED_VA_COPY
2710# ifdef va_copy
2711# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2712# elif defined(__va_copy)
2713# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2714# else
07798b17 2715# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2716# endif
2717#endif
2718
472d47bc
SB
2719/* convenience debug macros */
2720#ifdef USE_ITHREADS
2721#define pTHX_FORMAT "Perl interpreter: 0x%p"
2722#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2723#define pTHX_VALUE_ (void *)my_perl,
2724#define pTHX_VALUE (void *)my_perl
2725#define pTHX__VALUE_ ,(void *)my_perl,
2726#define pTHX__VALUE ,(void *)my_perl
472d47bc 2727#else
3609ea0d 2728#define pTHX_FORMAT
472d47bc 2729#define pTHX__FORMAT
3609ea0d 2730#define pTHX_VALUE_
472d47bc 2731#define pTHX_VALUE
3609ea0d 2732#define pTHX__VALUE_
472d47bc
SB
2733#define pTHX__VALUE
2734#endif /* USE_ITHREADS */
3609ea0d 2735
2acdbac1
NC
2736/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2737 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2738 CPAN::Unpack show any users outside the core. */
2739#ifdef PERL_CORE
dc6e8de0
A
2740# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2741 "Use of " s " is deprecated")
c9680906
A
2742# define deprecate_disappears_in(when,message) \
2743 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2744 message ", and will disappear in Perl " when)
ac641426
A
2745# define deprecate_fatal_in(when,message) \
2746 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2747 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2748#endif
2749
dfff4baf
BF
2750/* Internal macros to deal with gids and uids */
2751#ifdef PERL_CORE
2752
2753# if Uid_t_size > IVSIZE
2754# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2755# define SvUID(sv) SvNV(sv)
07798b17
AC
2756# elif Uid_t_sign <= 0
2757# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2758# define SvUID(sv) SvIV(sv)
dfff4baf 2759# else
07798b17
AC
2760# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2761# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2762# endif /* Uid_t_size */
2763
2764# if Gid_t_size > IVSIZE
2765# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2766# define SvGID(sv) SvNV(sv)
07798b17
AC
2767# elif Gid_t_sign <= 0
2768# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2769# define SvGID(sv) SvIV(sv)
dfff4baf 2770# else
07798b17
AC
2771# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2772# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2773# endif /* Gid_t_size */
2774
2775#endif
2776
6a5bc5ac 2777#endif /* PERL_HANDY_H_ */
9d745869 2778
e9a8c099 2779/*
14d04a33 2780 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2781 */