This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
t/TEST: Unset PERL_UNICODE
[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)
6c2255e0 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__
7adf2470 119#elif (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
07798b17 120# define FUNCTION__ ""
46c6c7e2 121#else
07798b17 122# define FUNCTION__ __FUNCTION__ /* Common extension. */
46c6c7e2
JH
123#endif
124
27d4fb96
PP
125/* XXX A note on the perl source internal type system. The
126 original intent was that I32 be *exactly* 32 bits.
127
128 Currently, we only guarantee that I32 is *at least* 32 bits.
129 Specifically, if int is 64 bits, then so is I32. (This is the case
130 for the Cray.) This has the advantage of meshing nicely with
131 standard library calls (where we pass an I32 and the library is
132 expecting an int), but the disadvantage that an I32 is not 32 bits.
133 Andy Dougherty August 1996
24fef2a7 134
dc45a647
MB
135 There is no guarantee that there is *any* integral type with
136 exactly 32 bits. It is perfectly legal for a system to have
137 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 138
dc45a647
MB
139 Similarly, there is no guarantee that I16 and U16 have exactly 16
140 bits.
693762b4 141
8e84507e
NIS
142 For dealing with issues that may arise from various 32/64-bit
143 systems, we will ask Configure to check out
8175356b 144
3609ea0d
JH
145 SHORTSIZE == sizeof(short)
146 INTSIZE == sizeof(int)
147 LONGSIZE == sizeof(long)
dc45a647 148 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 149 PTRSIZE == sizeof(void *)
dc45a647
MB
150 DOUBLESIZE == sizeof(double)
151 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 152
27d4fb96
PP
153*/
154
69512466
JH
155#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
156# include <inttypes.h>
dd0eed91
JH
157# ifdef INT32_MIN_BROKEN
158# undef INT32_MIN
159# define INT32_MIN (-2147483647-1)
160# endif
161# ifdef INT64_MIN_BROKEN
162# undef INT64_MIN
163# define INT64_MIN (-9223372036854775807LL-1)
164# endif
69512466
JH
165#endif
166
8175356b
JH
167typedef I8TYPE I8;
168typedef U8TYPE U8;
169typedef I16TYPE I16;
170typedef U16TYPE U16;
171typedef I32TYPE I32;
172typedef U32TYPE U32;
16d89be8 173
74b807c7 174#ifdef QUADKIND
8175356b
JH
175typedef I64TYPE I64;
176typedef U64TYPE U64;
16d89be8 177#endif
8175356b 178
d8668976 179#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 180
5ff3f7a4
GS
181/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
182 Please search CHAR_MAX in perl.h for further details. */
183#define U8_MAX UINT8_MAX
184#define U8_MIN UINT8_MIN
185
5ff3f7a4
GS
186#define I16_MAX INT16_MAX
187#define I16_MIN INT16_MIN
188#define U16_MAX UINT16_MAX
189#define U16_MIN UINT16_MIN
190
5ff3f7a4
GS
191#define I32_MAX INT32_MAX
192#define I32_MIN INT32_MIN
0e983133
GS
193#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
194# define U32_MAX UINT32_MAX
195#else
196# define U32_MAX 4294967295U
197#endif
5ff3f7a4
GS
198#define U32_MIN UINT32_MIN
199
200#else
201
5c9fa16e
KA
202/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
203 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
204#define U8_MAX PERL_UCHAR_MAX
205#define U8_MIN PERL_UCHAR_MIN
79072805 206
27d4fb96
PP
207#define I16_MAX PERL_SHORT_MAX
208#define I16_MIN PERL_SHORT_MIN
209#define U16_MAX PERL_USHORT_MAX
210#define U16_MIN PERL_USHORT_MIN
79072805 211
c4f23d77 212#if LONGSIZE > 4
27d4fb96
PP
213# define I32_MAX PERL_INT_MAX
214# define I32_MIN PERL_INT_MIN
215# define U32_MAX PERL_UINT_MAX
216# define U32_MIN PERL_UINT_MIN
79072805 217#else
27d4fb96
PP
218# define I32_MAX PERL_LONG_MAX
219# define I32_MIN PERL_LONG_MIN
220# define U32_MAX PERL_ULONG_MAX
221# define U32_MIN PERL_ULONG_MIN
79072805
LW
222#endif
223
5ff3f7a4
GS
224#endif
225
247cee9f
KW
226/* These C99 typedefs are useful sometimes for, say, loop variables whose
227 * maximum values are small, but for which speed trumps size. If we have a C99
228 * compiler, use that. Otherwise, a plain 'int' should be good enough.
229 *
230 * Restrict these to core for now until we are more certain this is a good
231 * idea. */
232#if defined(PERL_CORE) || defined(PERL_EXT)
233# ifdef I_STDINT
234 typedef int_fast8_t PERL_INT_FAST8_T;
235 typedef uint_fast8_t PERL_UINT_FAST8_T;
236 typedef int_fast16_t PERL_INT_FAST16_T;
237 typedef uint_fast16_t PERL_UINT_FAST16_T;
238# else
239 typedef int PERL_INT_FAST8_T;
240 typedef unsigned int PERL_UINT_FAST8_T;
241 typedef int PERL_INT_FAST16_T;
242 typedef unsigned int PERL_UINT_FAST16_T;
243# endif
244#endif
245
464decb6 246/* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
ab350cbd
KW
247 * anyone is grepping for it. So BIT_DIGITS gives the number of decimal digits
248 * required to represent any possible unsigned number containing N bits.
249 * TYPE_DIGITS gives the number of decimal digits required to represent any
250 * possible unsigned number of type T. */
464decb6 251#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
fc36a67e
PP
252#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
253#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
254
88794300 255/* Unused by core; should be deprecated */
ff68c719 256#define Ctl(ch) ((ch) & 037)
8d063cd8 257
98fce2a4
KW
258#if defined(PERL_CORE) || defined(PERL_EXT)
259# ifndef MIN
260# define MIN(a,b) ((a) < (b) ? (a) : (b))
261# endif
262# ifndef MAX
263# define MAX(a,b) ((a) > (b) ? (a) : (b))
264# endif
265#endif
266
84ff4fa9
KW
267/* Returns a boolean as to whether the input unsigned number is a power of 2
268 * (2**0, 2**1, etc). In other words if it has just a single bit set.
269 * If not, subtracting 1 would leave the uppermost bit set, so the & would
270 * yield non-zero */
271#if defined(PERL_CORE) || defined(PERL_EXT)
011b1419 272# define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
84ff4fa9
KW
273#endif
274
d223e1ea
KW
275/* Returns a mask with the lowest n bits set */
276#ifdef HAS_LONG_LONG
277# define nBIT_MASK(n) ((1ULL << (n)) - 1)
278#else
279# define nBIT_MASK(n) ((1UL << (n)) - 1)
280#endif
281
1381ccb1
KW
282/* The largest unsigned number that will fit into n bits */
283#define nBIT_UMAX(n) nBIT_MASK(n)
284
8d9433eb
KW
285/*
286=for apidoc Am|void|__ASSERT_|bool expr
287
288This is a helper macro to avoid preprocessor issues, replaced by nothing
289unless under DEBUGGING, where it expands to an assert of its argument,
290followed by a comma (hence the comma operator). If we just used a straight
291assert(), we would get a comma with nothing before it when not DEBUGGING.
292
293=cut
294
295We also use empty definition under Coverity since the __ASSERT__
296checks often check for things that Really Cannot Happen, and Coverity
297detects that and gets all excited. */
3e94db23 298
e7ae132e
KW
299#if defined(DEBUGGING) && !defined(__COVERITY__) \
300 && ! defined(PERL_SMALL_MACRO_BUFFER)
0f092d08
KW
301# define __ASSERT_(statement) assert(statement),
302#else
303# define __ASSERT_(statement)
304#endif
305
3fe05580 306/*
a4ee4fb5 307=head1 SV Manipulation Functions
3fe05580 308
3bb9fd01 309=for apidoc Ama|SV*|newSVpvs|"literal string"
1568d13a 310Like C<newSVpvn>, but takes a literal string instead of a
30a15352 311string/length pair.
3fe05580 312
3bb9fd01 313=for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
1568d13a 314Like C<newSVpvn_flags>, but takes a literal string instead of
30a15352 315a string/length pair.
84bafc02 316
3bb9fd01 317=for apidoc Ama|SV*|newSVpvs_share|"literal string"
1568d13a 318Like C<newSVpvn_share>, but takes a literal string instead of
30a15352 319a string/length pair and omits the hash parameter.
3fe05580 320
3bb9fd01 321=for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
1568d13a 322Like C<sv_catpvn_flags>, but takes a literal string instead
30a15352 323of a string/length pair.
9dcc53ea 324
3bb9fd01 325=for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
1568d13a 326Like C<sv_catpvn_nomg>, but takes a literal string instead of
0c395ea5 327a string/length pair.
9dcc53ea 328
3bb9fd01 329=for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
1568d13a 330Like C<sv_catpvn>, but takes a literal string instead of a
0c395ea5 331string/length pair.
3fe05580 332
3bb9fd01 333=for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
1568d13a 334Like C<sv_catpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
335string/length pair.
336
3bb9fd01 337=for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
1568d13a 338Like C<sv_setpvn>, but takes a literal string instead of a
0c395ea5 339string/length pair.
3fe05580 340
3bb9fd01 341=for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
1568d13a 342Like C<sv_setpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
343string/length pair.
344
3bb9fd01 345=for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
1568d13a 346Like C<sv_setref_pvn>, but takes a literal string instead of
0c395ea5 347a string/length pair.
9dcc53ea 348
3fe05580
MHM
349=head1 Memory Management
350
3bb9fd01 351=for apidoc Ama|char*|savepvs|"literal string"
1568d13a 352Like C<savepvn>, but takes a literal string instead of a
30a15352 353string/length pair.
3fe05580 354
3bb9fd01 355=for apidoc Ama|char*|savesharedpvs|"literal string"
9dcc53ea
Z
356A version of C<savepvs()> which allocates the duplicate string in memory
357which is shared between threads.
358
3fe05580
MHM
359=head1 GV Functions
360
3bb9fd01 361=for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
1568d13a 362Like C<gv_stashpvn>, but takes a literal string instead of a
0c395ea5 363string/length pair.
3fe05580
MHM
364
365=head1 Hash Manipulation Functions
366
3bb9fd01 367=for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
1568d13a 368Like C<hv_fetch>, but takes a literal string instead of a
0c395ea5 369string/length pair.
3fe05580 370
3bb9fd01 371=for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
1568d13a 372Like C<hv_store>, but takes a literal string instead of a
0c395ea5 373string/length pair
3fe05580
MHM
374and omits the hash parameter.
375
510966aa
Z
376=head1 Lexer interface
377
3bb9fd01 378=for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
510966aa 379
1568d13a 380Like L</lex_stuff_pvn>, but takes a literal string instead of
0c395ea5 381a string/length pair.
510966aa 382
3fe05580
MHM
383=cut
384*/
385
a34e53fc
KW
386/*
387=head1 Handy Values
2efa8cc7 388
a34e53fc
KW
389=for apidoc Amu|pair|STR_WITH_LEN|"literal string"
390
391Returns two comma separated tokens of the input literal string, and its length.
392This is convenience macro which helps out in some API calls.
393Note that it can't be used as an argument to macros or functions that under
394some configurations might be macros, which means that it requires the full
395Perl_xxx(aTHX_ ...) form for any API calls where it's used.
396
397=cut
398*/
399
a34e53fc 400#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
ba3a79e7
GA
401
402/* STR_WITH_LEN() shortcuts */
403#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
404#define newSVpvs_flags(str,flags) \
405 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 406#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
407#define sv_catpvs_flags(sv, str, flags) \
408 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
409#define sv_catpvs_nomg(sv, str) \
410 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
411#define sv_catpvs(sv, str) \
412 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
413#define sv_catpvs_mg(sv, str) \
414 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 415#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
416#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
417#define sv_setref_pvs(rv, classname, str) \
418 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 419#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
420#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
421#define gv_stashpvs(str, create) \
422 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
423#define gv_fetchpvs(namebeg, add, sv_type) \
424 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
425#define gv_fetchpvn(namebeg, len, add, sv_type) \
426 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
427#define sv_catxmlpvs(dsv, str, utf8) \
428 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 429
ba3a79e7 430
510966aa
Z
431#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
432
b96d8cd9
NC
433#define get_cvs(str, flags) \
434 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 435
9b6e9510
KW
436/* internal helpers */
437#define PERL_RVS_TO_DECIMAL_(r,v,s) (((r)*1000000)+((v)*1000)+(s))
438#define PERL_DECIMAL_VERSION_ \
439 PERL_RVS_TO_DECIMAL_(PERL_REVISION, PERL_VERSION, PERL_SUBVERSION)
440
441/*
442=for apidoc AmR|bool|PERL_VERSION_EQ|const int r|const int v|const int s
443
444Returns whether or not the perl currently executing has the specified
445relationship to the perl given by the parameters. For example,
446
447 #if PERL_VERSION_GT(5,24,2)
448 code that will only be compiled on perls after v5.24.2
449 #else
450 fallback code
451 #endif
452
453Note that this is usable in making compile-time decisions
454
455The possible comparisons are C<PERL_VERSION_EQ>, C<PERL_VERSION_NE>,
456C<PERL_VERSION_GE>, C<PERL_VERSION_GT>, C<PERL_VERSION_LE>, and
457C<PERL_VERSION_LT>.
458
459=for apidoc AmRh|bool|PERL_VERSION_NE|const int r|const int v|const int s
460=for apidoc AmRh|bool|PERL_VERSION_GE|const int r|const int v|const int s
461=for apidoc AmRh|bool|PERL_VERSION_GT|const int r|const int v|const int s
462=for apidoc AmRh|bool|PERL_VERSION_LE|const int r|const int v|const int s
463=for apidoc AmRh|bool|PERL_VERSION_LT|const int r|const int v|const int s
464
465=cut
466*/
467
468# define PERL_VERSION_EQ(r,v,s) (PERL_DECIMAL_VERSION_ == PERL_RVS_TO_DECIMAL_(r,v,s))
469# define PERL_VERSION_NE(r,v,s) (! PERL_VERSION_EQ(r,v,s))
470# define PERL_VERSION_LT(r,v,s) (PERL_DECIMAL_VERSION_ < PERL_RVS_TO_DECIMAL_(r,v,s))
471# define PERL_VERSION_LE(r,v,s) (PERL_DECIMAL_VERSION_ <= PERL_RVS_TO_DECIMAL_(r,v,s))
472# define PERL_VERSION_GT(r,v,s) (! PERL_VERSION_LE(r,v,s))
473# define PERL_VERSION_GE(r,v,s) (! PERL_VERSION_LT(r,v,s))
474
954c1994 475/*
ccfc67b7
JH
476=head1 Miscellaneous Functions
477
954c1994 478=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
479Test two C<NUL>-terminated strings to see if they are different. Returns true
480or false.
954c1994
GS
481
482=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
483Test two C<NUL>-terminated strings to see if they are equal. Returns true or
484false.
954c1994
GS
485
486=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
487Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
488second, C<s2>. Returns true or false.
954c1994
GS
489
490=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
491Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
492equal to the second, C<s2>. Returns true or false.
954c1994
GS
493
494=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
495Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
496the second, C<s2>. Returns true or false.
954c1994
GS
497
498=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
499Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
500or equal to the second, C<s2>. Returns true or false.
954c1994
GS
501
502=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
503Test two C<NUL>-terminated strings to see if they are different. The C<len>
504parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
505wrapper for C<strncmp>).
506
507=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
508Test two C<NUL>-terminated strings to see if they are equal. The C<len>
509parameter indicates the number of bytes to compare. Returns true or false. (A
510wrapper for C<strncmp>).
954c1994 511
bd18bd40
KW
512=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
513Test two buffers (which may contain embedded C<NUL> characters, to see if they
514are equal. The C<len> parameter indicates the number of bytes to compare.
515Returns zero if equal, or non-zero if non-equal.
516
3bb9fd01 517=for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
518Like L</memEQ>, but the second string is a literal enclosed in double quotes,
519C<l1> gives the number of bytes in C<s1>.
520Returns zero if equal, or non-zero if non-equal.
521
bd18bd40
KW
522=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
523Test two buffers (which may contain embedded C<NUL> characters, to see if they
524are not equal. The C<len> parameter indicates the number of bytes to compare.
525Returns zero if non-equal, or non-zero if equal.
526
3bb9fd01 527=for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
528Like L</memNE>, but the second string is a literal enclosed in double quotes,
529C<l1> gives the number of bytes in C<s1>.
530Returns zero if non-equal, or zero if non-equal.
531
4aada8b9
KW
532=for apidoc Am|bool|memCHRs|"list"|char c
533Returns the position of the first occurence of the byte C<c> in the literal
534string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
535treated as unsigned char. Thus this macro can be used to determine if C<c> is
536in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
537is C<NUL> (and the set doesn't include C<NUL>).
538
954c1994 539=cut
fc169e00
KW
540
541New macros should use the following conventions for their names (which are
542based on the underlying C library functions):
543
544 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
545
546 Each has two main parameters, string-like operands that are compared
547 against each other, as specified by the macro name. Some macros may
548 additionally have one or potentially even two length parameters. If a length
549 parameter applies to both string parameters, it will be positioned third;
550 otherwise any length parameter immediately follows the string parameter it
551 applies to.
552
553 If the prefix to the name is 'str', the string parameter is a pointer to a C
554 language string. Such a string does not contain embedded NUL bytes; its
555 length may be unknown, but can be calculated by C<strlen()>, since it is
556 terminated by a NUL, which isn't included in its length.
557
a3815e44 558 The optional 'n' following 'str' means that there is a third parameter,
fc169e00
KW
559 giving the maximum number of bytes to look at in each string. Even if both
560 strings are longer than the length parameter, those extra bytes will be
561 unexamined.
562
563 The 's' suffix means that the 2nd byte string parameter is a literal C
564 double-quoted string. Its length will automatically be calculated by the
565 macro, so no length parameter will ever be needed for it.
566
567 If the prefix is 'mem', the string parameters don't have to be C strings;
568 they may contain embedded NUL bytes, do not necessarily have a terminating
569 NUL, and their lengths can be known only through other means, which in
570 practice are additional parameter(s) passed to the function. All 'mem'
571 functions have at least one length parameter. Barring any 'l' or 's' suffix,
572 there is a single length parameter, in position 3, which applies to both
573 string parameters. The 's' suffix means, as described above, that the 2nd
574 string is a literal double-quoted C string (hence its length is calculated by
575 the macro, and the length parameter to the function applies just to the first
576 string parameter, and hence is positioned just after it). An 'l' suffix
577 means that the 2nd string parameter has its own length parameter, and the
578 signature will look like memFOOl(s1, l1, s2, l2).
579
580 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
581 substring of the 1st string. 'P' if present indicates that the substring
582 must be a "proper" one in tha mathematical sense that the first one must be
583 strictly larger than the 2nd.
584
954c1994
GS
585*/
586
62946e08 587
75400963
KW
588#define strNE(s1,s2) (strcmp(s1,s2) != 0)
589#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
590#define strLT(s1,s2) (strcmp(s1,s2) < 0)
591#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
592#define strGT(s1,s2) (strcmp(s1,s2) > 0)
593#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 594
75400963
KW
595#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
596#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 597
9d3980bc
KW
598#define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
599#define memNE(s1,s2,l) (! memEQ(s1,s2,l))
36477c24 600
085b7534 601/* memEQ and memNE where second comparand is a string constant */
568a785a 602#define memEQs(s1, l, s2) \
777fa2cb 603 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 604#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 605
fdbb9a7c
KW
606/* Keep these private until we decide it was a good idea */
607#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
608
609#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
610
bdb7e3f0 611#define memBEGINs(s1, l, s2) \
30a6480c 612 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 613 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158 614#define memBEGINPs(s1, l, s2) \
30a6480c 615 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
de627158 616 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0 617#define memENDs(s1, l, s2) \
30a6480c 618 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 619 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424 620#define memENDPs(s1, l, s2) \
30a6480c 621 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
b80f8424 622 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 623#endif /* End of making macros private */
bdb7e3f0 624
062b6850
KW
625#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
626#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
627#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
628#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
629
4aada8b9
KW
630#define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
631
bbce6d69
PP
632/*
633 * Character classes.
634 *
635 * Unfortunately, the introduction of locales means that we
636 * can't trust isupper(), etc. to tell the truth. And when
637 * it comes to /\w+/ with tainting enabled, we *must* be able
638 * to trust our character classes.
639 *
640 * Therefore, the default tests in the text of Perl will be
641 * independent of locale. Any code that wants to depend on
642 * the current locale will use the tests that begin with "lc".
643 */
644
2304df62
AD
645#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
646# ifndef CTYPE256
647# define CTYPE256
648# endif
649#endif
650
954c1994 651/*
ccfc67b7 652
dcccc8ff 653=head1 Character classification
243effed
KW
654This section is about functions (really macros) that classify characters
655into types, such as punctuation versus alphabetic, etc. Most of these are
656analogous to regular expression character classes. (See
657L<perlrecharclass/POSIX Character Classes>.) There are several variants for
658each class. (Not all macros have all variants; each item below lists the
659ones valid for it.) None are affected by C<use bytes>, and only the ones
660with C<LC> in the name are affected by the current locale.
661
d713f9d9
KW
662The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
663treating it as a code point, and returns a boolean as to whether or not the
664character represented by it is (or on non-ASCII platforms, corresponds to) an
6aff1f14
KW
665ASCII character in the named class based on platform, Unicode, and Perl rules.
666If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 667
c98722a4 668Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
669with no suffix C<"_A">. This variant is used to emphasize by its name that
670only ASCII-range characters can return TRUE.
4b9734bf 671
d60679e1 672Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
673onto the platform. That is, the code points that are ASCII are unaffected,
674since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
675as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
676true when called with the code point 0xDF, which is a word character in both
4650c663 677ASCII and EBCDIC (though it represents different characters in each).
d713f9d9
KW
678If the input is a number that doesn't fit in an octet, FALSE is returned.
679(Perl's documentation uses a colloquial definition of Latin-1, to include all
680code points below 256.)
243effed 681
d713f9d9
KW
682Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
683inputs below 256, but if the code point is larger than 255, Unicode rules are
684used to determine if it is in the character class. For example,
d0da05db 685C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 686WITH MACRON in Unicode, and is a word character.
243effed 687
059703b0
KW
688Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
689but are used for UTF-8 encoded strings. The two forms are different names for
690the same thing. Each call to one of these classifies the first character of
691the string starting at C<p>. The second parameter, C<e>, points to anywhere in
692the string beyond the first character, up to one byte past the end of the
693entire string. Although both variants are identical, the suffix C<_safe> in
694one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
695provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
696C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
697some way, the program may croak, or the function may return FALSE, at the
698discretion of the implementation, and subject to change in future releases.
243effed 699
d713f9d9
KW
700Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
701but the result is based on the current locale, which is what C<LC> in the name
702stands for. If Perl can determine that the current locale is a UTF-8 locale,
703it uses the published Unicode rules; otherwise, it uses the C library function
704that gives the named classification. For example, C<isDIGIT_LC()> when not in
705a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
706returned if the input won't fit into an octet. On some platforms where the C
707library function is known to be defective, Perl changes its result to follow
708the POSIX standard's rules.
243effed 709
d713f9d9
KW
710Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
711than 256, but for larger ones it returns the Unicode classification of the code
712point.
243effed 713
059703b0
KW
714Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
715C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
716are different names for the same thing. Each call to one of these classifies
717the first character of the string starting at C<p>. The second parameter,
718C<e>, points to anywhere in the string beyond the first character, up to one
719byte past the end of the entire string. Although both variants are identical,
720the suffix C<_safe> in one name emphasizes that it will not attempt to read
721beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
722is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
723character is malformed in some way, the program may croak, or the function may
724return FALSE, at the discretion of the implementation, and subject to change in
725future releases.
ccfc67b7 726
d713f9d9
KW
727=for apidoc Am|bool|isALPHA|int ch
728Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
729analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
730See the L<top of this section|/Character classification> for an explanation of
731variants
059703b0
KW
732C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>,
733C<isALPHA_utf8_safe>, C<isALPHA_LC>, C<isALPHA_LC_uvchr>, C<isALPHA_LC_utf8>,
734and C<isALPHA_LC_utf8_safe>.
8a58bdcf 735
f16858ed
KW
736=cut
737
738Here and below, we add the protoypes of these macros for downstream programs
739that would be interested in them, such as Devel::PPPort
740
741=for apidoc Amh|bool|isALPHA_A|int ch
742=for apidoc Amh|bool|isALPHA_L1|int ch
743=for apidoc Amh|bool|isALPHA_uvchr|int ch
744=for apidoc Amh|bool|isALPHA_utf8_safe|U8 * s|U8 * end
d23c7e08 745=for apidoc Amh|bool|isALPHA_utf8|U8 * s|U8 * end
f16858ed
KW
746=for apidoc Amh|bool|isALPHA_LC|int ch
747=for apidoc Amh|bool|isALPHA_LC_uvchr|int ch
748=for apidoc Amh|bool|isALPHA_LC_utf8_safe|U8 * s| U8 *end
749
d713f9d9
KW
750=for apidoc Am|bool|isALPHANUMERIC|int ch
751Returns a boolean indicating whether the specified character is one of
752C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
753See the L<top of this section|/Character classification> for an explanation of
754variants
d0da05db 755C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
059703b0
KW
756C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>,
757C<isALPHANUMERIC_LC_uvchr>, C<isALPHANUMERIC_LC_utf8>, and
758C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 759
255b632a
KW
760A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
761this corresponds to the C language alphanumeric definition). Also
762there are the variants
763C<isALNUMC_A>, C<isALNUMC_L1>
764C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
765
f16858ed
KW
766=for apidoc Amh|bool|isALPHANUMERIC_A|int ch
767=for apidoc Amh|bool|isALPHANUMERIC_L1|int ch
768=for apidoc Amh|bool|isALPHANUMERIC_uvchr|int ch
769=for apidoc Amh|bool|isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
d23c7e08 770=for apidoc Amh|bool|isALPHANUMERIC_utf8|U8 * s|U8 * end
f16858ed
KW
771=for apidoc Amh|bool|isALPHANUMERIC_LC|int ch
772=for apidoc Amh|bool|isALPHANUMERIC_LC_uvchr|int ch
773=for apidoc Amh|bool|isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
774=for apidoc Amh|bool|isALNUMC|int ch
775=for apidoc Amh|bool|isALNUMC_A|int ch
776=for apidoc Amh|bool|isALNUMC_L1|int ch
777=for apidoc Amh|bool|isALNUMC_LC|int ch
778=for apidoc Amh|bool|isALNUMC_LC_uvchr|int ch
779
d713f9d9 780=for apidoc Am|bool|isASCII|int ch
8a58bdcf 781Returns a boolean indicating whether the specified character is one of the 128
243effed 782characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 783On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
784character corresponds to an ASCII character. Variants C<isASCII_A()> and
785C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
786See the L<top of this section|/Character classification> for an explanation of
787variants
059703b0
KW
788C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_utf8_safe>, C<isASCII_LC>,
789C<isASCII_LC_uvchr>, C<isASCII_LC_utf8>, and C<isASCII_LC_utf8_safe>.
790Note, however, that some platforms do not have the C library routine
791C<isascii()>. In these cases, the variants whose names contain C<LC> are the
792same as the corresponding ones without.
243effed 793
f16858ed
KW
794=for apidoc Amh|bool|isASCII_A|int ch
795=for apidoc Amh|bool|isASCII_L1|int ch
796=for apidoc Amh|bool|isASCII_uvchr|int ch
797=for apidoc Amh|bool|isASCII_utf8_safe|U8 * s|U8 * end
d23c7e08 798=for apidoc Amh|bool|isASCII_utf8|U8 * s|U8 * end
f16858ed
KW
799=for apidoc Amh|bool|isASCII_LC|int ch
800=for apidoc Amh|bool|isASCII_LC_uvchr|int ch
801=for apidoc Amh|bool|isASCII_LC_utf8_safe|U8 * s| U8 *end
802
d98532ea
KW
803Also note, that because all ASCII characters are UTF-8 invariant (meaning they
804have the exact same representation (always a single byte) whether encoded in
805UTF-8 or not), C<isASCII> will give the correct results when called with any
059703b0
KW
806byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
807C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
d98532ea 808
243effed
KW
809=for apidoc Am|bool|isBLANK|char ch
810Returns a boolean indicating whether the specified character is a
6aff1f14 811character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
812See the L<top of this section|/Character classification> for an explanation of
813variants
059703b0
KW
814C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>,
815C<isBLANK_utf8_safe>, C<isBLANK_LC>, C<isBLANK_LC_uvchr>, C<isBLANK_LC_utf8>,
816and C<isBLANK_LC_utf8_safe>. Note,
da8c1a98
KW
817however, that some platforms do not have the C library routine
818C<isblank()>. In these cases, the variants whose names contain C<LC> are
819the same as the corresponding ones without.
243effed 820
f16858ed
KW
821=for apidoc Amh|bool|isBLANK_A|int ch
822=for apidoc Amh|bool|isBLANK_L1|int ch
823=for apidoc Amh|bool|isBLANK_uvchr|int ch
824=for apidoc Amh|bool|isBLANK_utf8_safe|U8 * s|U8 * end
d23c7e08 825=for apidoc Amh|bool|isBLANK_utf8|U8 * s|U8 * end
f16858ed
KW
826=for apidoc Amh|bool|isBLANK_LC|int ch
827=for apidoc Amh|bool|isBLANK_LC_uvchr|int ch
828=for apidoc Amh|bool|isBLANK_LC_utf8_safe|U8 * s| U8 *end
829
243effed
KW
830=for apidoc Am|bool|isCNTRL|char ch
831Returns a boolean indicating whether the specified character is a
6aff1f14 832control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
833See the L<top of this section|/Character classification> for an explanation of
834variants
059703b0
KW
835C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>,
836C<isCNTRL_utf8_safe>, C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, C<isCNTRL_LC_utf8>
837and C<isCNTRL_LC_utf8_safe>. On EBCDIC
da8c1a98 838platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 839
f16858ed
KW
840=for apidoc Amh|bool|isCNTRL_A|int ch
841=for apidoc Amh|bool|isCNTRL_L1|int ch
842=for apidoc Amh|bool|isCNTRL_uvchr|int ch
843=for apidoc Amh|bool|isCNTRL_utf8_safe|U8 * s|U8 * end
d23c7e08 844=for apidoc Amh|bool|isCNTRL_utf8|U8 * s|U8 * end
f16858ed
KW
845=for apidoc Amh|bool|isCNTRL_LC|int ch
846=for apidoc Amh|bool|isCNTRL_LC_uvchr|int ch
847=for apidoc Amh|bool|isCNTRL_LC_utf8_safe|U8 * s| U8 *end
848
954c1994 849=for apidoc Am|bool|isDIGIT|char ch
2787a470 850Returns a boolean indicating whether the specified character is a
6aff1f14 851digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 852Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
853See the L<top of this section|/Character classification> for an explanation of
854variants
059703b0
KW
855C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>,
856C<isDIGIT_LC_uvchr>, C<isDIGIT_LC_utf8>, and C<isDIGIT_LC_utf8_safe>.
243effed 857
f16858ed
KW
858=for apidoc Amh|bool|isDIGIT_A|int ch
859=for apidoc Amh|bool|isDIGIT_L1|int ch
860=for apidoc Amh|bool|isDIGIT_uvchr|int ch
861=for apidoc Amh|bool|isDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 862=for apidoc Amh|bool|isDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
863=for apidoc Amh|bool|isDIGIT_LC|int ch
864=for apidoc Amh|bool|isDIGIT_LC_uvchr|int ch
865=for apidoc Amh|bool|isDIGIT_LC_utf8_safe|U8 * s| U8 *end
866
243effed
KW
867=for apidoc Am|bool|isGRAPH|char ch
868Returns a boolean indicating whether the specified character is a
6aff1f14 869graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 870See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
871variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>,
872C<isGRAPH_utf8_safe>, C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>,
873C<isGRAPH_LC_utf8_safe>, and C<isGRAPH_LC_utf8_safe>.
954c1994 874
f16858ed
KW
875=for apidoc Amh|bool|isGRAPH_A|int ch
876=for apidoc Amh|bool|isGRAPH_L1|int ch
877=for apidoc Amh|bool|isGRAPH_uvchr|int ch
878=for apidoc Amh|bool|isGRAPH_utf8_safe|U8 * s|U8 * end
d23c7e08 879=for apidoc Amh|bool|isGRAPH_utf8|U8 * s|U8 * end
f16858ed
KW
880=for apidoc Amh|bool|isGRAPH_LC|int ch
881=for apidoc Amh|bool|isGRAPH_LC_uvchr|int ch
882=for apidoc Amh|bool|isGRAPH_LC_utf8_safe|U8 * s| U8 *end
883
0c82b6df 884=for apidoc Am|bool|isLOWER|char ch
2787a470 885Returns a boolean indicating whether the specified character is a
6aff1f14 886lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
887See the L<top of this section|/Character classification> for an explanation of
888variants
059703b0
KW
889C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>,
890C<isLOWER_utf8_safe>, C<isLOWER_LC>, C<isLOWER_LC_uvchr>, C<isLOWER_LC_utf8>,
891and C<isLOWER_LC_utf8_safe>.
0c82b6df 892
f16858ed
KW
893=for apidoc Amh|bool|isLOWER_A|int ch
894=for apidoc Amh|bool|isLOWER_L1|int ch
895=for apidoc Amh|bool|isLOWER_uvchr|int ch
896=for apidoc Amh|bool|isLOWER_utf8_safe|U8 * s|U8 * end
d23c7e08 897=for apidoc Amh|bool|isLOWER_utf8|U8 * s|U8 * end
f16858ed
KW
898=for apidoc Amh|bool|isLOWER_LC|int ch
899=for apidoc Amh|bool|isLOWER_LC_uvchr|int ch
900=for apidoc Amh|bool|isLOWER_LC_utf8_safe|U8 * s| U8 *end
901
c99e91e9 902=for apidoc Am|bool|isOCTAL|char ch
2787a470 903Returns a boolean indicating whether the specified character is an
6aff1f14 904octal digit, [0-7].
243effed
KW
905The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
906C<isOCTAL>.
907
f16858ed
KW
908=for apidoc Amh|bool|isOCTAL_A|int ch
909=for apidoc Amh|bool|isOCTAL_L1|int ch
910
243effed
KW
911=for apidoc Am|bool|isPUNCT|char ch
912Returns a boolean indicating whether the specified character is a
6aff1f14
KW
913punctuation character, analogous to C<m/[[:punct:]]/>.
914Note that the definition of what is punctuation isn't as
243effed
KW
915straightforward as one might desire. See L<perlrecharclass/POSIX Character
916Classes> for details.
dcccc8ff 917See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
918variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>,
919C<isPUNCT_utf8_safe>, C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, C<isPUNCT_LC_utf8>,
920and C<isPUNCT_LC_utf8_safe>.
c99e91e9 921
f16858ed
KW
922=for apidoc Amh|bool|isPUNCT_A|int ch
923=for apidoc Amh|bool|isPUNCT_L1|int ch
924=for apidoc Amh|bool|isPUNCT_uvchr|int ch
925=for apidoc Amh|bool|isPUNCT_utf8_safe|U8 * s|U8 * end
d23c7e08 926=for apidoc Amh|bool|isPUNCT_utf8|U8 * s|U8 * end
f16858ed
KW
927=for apidoc Amh|bool|isPUNCT_LC|int ch
928=for apidoc Amh|bool|isPUNCT_LC_uvchr|int ch
929=for apidoc Amh|bool|isPUNCT_LC_utf8_safe|U8 * s| U8 *end
930
0c82b6df 931=for apidoc Am|bool|isSPACE|char ch
2787a470 932Returns a boolean indicating whether the specified character is a
6aff1f14 933whitespace character. This is analogous
398d098a 934to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 935this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
936locale forms of this macro (the ones with C<LC> in their names) matched
937precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
938in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
939(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
940See the L<top of this section|/Character classification> for an explanation of
941variants
059703b0
KW
942C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>,
943C<isSPACE_utf8_safe>, C<isSPACE_LC>, C<isSPACE_LC_uvchr>, C<isSPACE_LC_utf8>,
944and C<isSPACE_LC_utf8_safe>.
0c82b6df 945
f16858ed
KW
946=for apidoc Amh|bool|isSPACE_A|int ch
947=for apidoc Amh|bool|isSPACE_L1|int ch
948=for apidoc Amh|bool|isSPACE_uvchr|int ch
949=for apidoc Amh|bool|isSPACE_utf8_safe|U8 * s|U8 * end
d23c7e08 950=for apidoc Amh|bool|isSPACE_utf8|U8 * s|U8 * end
f16858ed
KW
951=for apidoc Amh|bool|isSPACE_LC|int ch
952=for apidoc Amh|bool|isSPACE_LC_uvchr|int ch
953=for apidoc Amh|bool|isSPACE_LC_utf8_safe|U8 * s| U8 *end
954
398d098a
KW
955=for apidoc Am|bool|isPSXSPC|char ch
956(short for Posix Space)
779cf272
KW
957Starting in 5.18, this is identical in all its forms to the
958corresponding C<isSPACE()> macros.
398d098a
KW
959The locale forms of this macro are identical to their corresponding
960C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
961non-locale forms differ from their C<isSPACE()> forms only in that the
962C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
963Otherwise they are identical. Thus this macro is analogous to what
964C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 965See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
966variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
967C<isPSXSPC_utf8_safe>, C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>,
968C<isPSXSPC_LC_utf8>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 969
f16858ed
KW
970=for apidoc Amh|bool|isPSXSPC_A|int ch
971=for apidoc Amh|bool|isPSXSPC_L1|int ch
972=for apidoc Amh|bool|isPSXSPC_uvchr|int ch
973=for apidoc Amh|bool|isPSXSPC_utf8_safe|U8 * s|U8 * end
d23c7e08 974=for apidoc Amh|bool|isPSXSPC_utf8|U8 * s|U8 * end
f16858ed
KW
975=for apidoc Amh|bool|isPSXSPC_LC|int ch
976=for apidoc Amh|bool|isPSXSPC_LC_uvchr|int ch
977=for apidoc Amh|bool|isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
978
954c1994 979=for apidoc Am|bool|isUPPER|char ch
2787a470 980Returns a boolean indicating whether the specified character is an
6aff1f14 981uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 982See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
983variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>,
984C<isUPPER_utf8_safe>, C<isUPPER_LC>, C<isUPPER_LC_uvchr>, C<isUPPER_LC_utf8>,
985and C<isUPPER_LC_utf8_safe>.
954c1994 986
f16858ed
KW
987=for apidoc Amh|bool|isUPPER_A|int ch
988=for apidoc Amh|bool|isUPPER_L1|int ch
989=for apidoc Amh|bool|isUPPER_uvchr|int ch
990=for apidoc Amh|bool|isUPPER_utf8_safe|U8 * s|U8 * end
d23c7e08 991=for apidoc Amh|bool|isUPPER_utf8|U8 * s|U8 * end
f16858ed
KW
992=for apidoc Amh|bool|isUPPER_LC|int ch
993=for apidoc Amh|bool|isUPPER_LC_uvchr|int ch
994=for apidoc Amh|bool|isUPPER_LC_utf8_safe|U8 * s| U8 *end
995
243effed 996=for apidoc Am|bool|isPRINT|char ch
8eea39dd 997Returns a boolean indicating whether the specified character is a
6aff1f14 998printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
999See the L<top of this section|/Character classification> for an explanation of
1000variants
059703b0
KW
1001C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>,
1002C<isPRINT_utf8_safe>, C<isPRINT_LC>, C<isPRINT_LC_uvchr>, C<isPRINT_LC_utf8>,
1003and C<isPRINT_LC_utf8_safe>.
243effed 1004
f16858ed
KW
1005=for apidoc Amh|bool|isPRINT_A|int ch
1006=for apidoc Amh|bool|isPRINT_L1|int ch
1007=for apidoc Amh|bool|isPRINT_uvchr|int ch
1008=for apidoc Amh|bool|isPRINT_utf8_safe|U8 * s|U8 * end
d23c7e08 1009=for apidoc Amh|bool|isPRINT_utf8|U8 * s|U8 * end
f16858ed
KW
1010=for apidoc Amh|bool|isPRINT_LC|int ch
1011=for apidoc Amh|bool|isPRINT_LC_uvchr|int ch
1012=for apidoc Amh|bool|isPRINT_LC_utf8_safe|U8 * s| U8 *end
1013
243effed
KW
1014=for apidoc Am|bool|isWORDCHAR|char ch
1015Returns a boolean indicating whether the specified character is a character
1016that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1017in a regular expression. A word character is an alphabetic character, a
1018decimal digit, a connecting punctuation character (such as an underscore), or
1019a "mark" character that attaches to one of those (like some sort of accent).
1020C<isALNUM()> is a synonym provided for backward compatibility, even though a
1021word character includes more than the standard C language meaning of
1022alphanumeric.
dcccc8ff 1023See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
1024variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1025C<isWORDCHAR_utf8>, and C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>,
1026C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and C<isWORDCHAR_LC_utf8_safe>
1027are also as described there, but additionally include the platform's native
1028underscore.
8a58bdcf 1029
f16858ed
KW
1030=for apidoc Amh|bool|isWORDCHAR_A|int ch
1031=for apidoc Amh|bool|isWORDCHAR_L1|int ch
1032=for apidoc Amh|bool|isWORDCHAR_uvchr|int ch
1033=for apidoc Amh|bool|isWORDCHAR_utf8_safe|U8 * s|U8 * end
d23c7e08 1034=for apidoc Amh|bool|isWORDCHAR_utf8|U8 * s|U8 * end
f16858ed
KW
1035=for apidoc Amh|bool|isWORDCHAR_LC|int ch
1036=for apidoc Amh|bool|isWORDCHAR_LC_uvchr|int ch
1037=for apidoc Amh|bool|isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
1038=for apidoc Amh|bool|isALNUM|int ch
1039=for apidoc Amh|bool|isALNUM_A|int ch
1040=for apidoc Amh|bool|isALNUM_LC|int ch
1041=for apidoc Amh|bool|isALNUM_LC_uvchr|int ch
1042
8a58bdcf
KW
1043=for apidoc Am|bool|isXDIGIT|char ch
1044Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
1045digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1046and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
1047See the L<top of this section|/Character classification> for an explanation of
1048variants
059703b0
KW
1049C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>,
1050C<isXDIGIT_LC_uvchr>, C<isXDIGIT_LC_utf8>, and C<isXDIGIT_LC_utf8_safe>.
243effed 1051
f16858ed
KW
1052=for apidoc Amh|bool|isXDIGIT_A|int ch
1053=for apidoc Amh|bool|isXDIGIT_L1|int ch
1054=for apidoc Amh|bool|isXDIGIT_uvchr|int ch
1055=for apidoc Amh|bool|isXDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 1056=for apidoc Amh|bool|isXDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
1057=for apidoc Amh|bool|isXDIGIT_LC|int ch
1058=for apidoc Amh|bool|isXDIGIT_LC_uvchr|int ch
1059=for apidoc Amh|bool|isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1060
3c3ecf18
KW
1061=for apidoc Am|bool|isIDFIRST|char ch
1062Returns a boolean indicating whether the specified character can be the first
1063character of an identifier. This is very close to, but not quite the same as
1064the official Unicode property C<XID_Start>. The difference is that this
1065returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
1066See the L<top of this section|/Character classification> for an explanation of
1067variants
059703b0
KW
1068C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
1069C<isIDFIRST_utf8_safe>, C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>,
1070C<isIDFIRST_LC_utf8>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18 1071
f16858ed
KW
1072=for apidoc Amh|bool|isIDFIRST_A|int ch
1073=for apidoc Amh|bool|isIDFIRST_L1|int ch
1074=for apidoc Amh|bool|isIDFIRST_uvchr|int ch
1075=for apidoc Amh|bool|isIDFIRST_utf8_safe|U8 * s|U8 * end
d23c7e08 1076=for apidoc Amh|bool|isIDFIRST_utf8|U8 * s|U8 * end
f16858ed
KW
1077=for apidoc Amh|bool|isIDFIRST_LC|int ch
1078=for apidoc Amh|bool|isIDFIRST_LC_uvchr|int ch
1079=for apidoc Amh|bool|isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1080
3c3ecf18
KW
1081=for apidoc Am|bool|isIDCONT|char ch
1082Returns a boolean indicating whether the specified character can be the
1083second or succeeding character of an identifier. This is very close to, but
1084not quite the same as the official Unicode property C<XID_Continue>. The
1085difference is that this returns true only if the input character also matches
dcccc8ff 1086L</isWORDCHAR>. See the L<top of this section|/Character classification> for
059703b0
KW
1087an explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
1088C<isIDCONT_utf8>, C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>,
1089C<isIDCONT_LC_utf8>, and C<isIDCONT_LC_utf8_safe>.
3c3ecf18 1090
f16858ed
KW
1091=for apidoc Amh|bool|isIDCONT_A|int ch
1092=for apidoc Amh|bool|isIDCONT_L1|int ch
1093=for apidoc Amh|bool|isIDCONT_uvchr|int ch
1094=for apidoc Amh|bool|isIDCONT_utf8_safe|U8 * s|U8 * end
d23c7e08 1095=for apidoc Amh|bool|isIDCONT_utf8|U8 * s|U8 * end
f16858ed
KW
1096=for apidoc Amh|bool|isIDCONT_LC|int ch
1097=for apidoc Amh|bool|isIDCONT_LC_uvchr|int ch
1098=for apidoc Amh|bool|isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1099
243effed 1100=head1 Miscellaneous Functions
8eea39dd 1101
95a59cab 1102=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 1103Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
1104Behaviour is only well defined when isXDIGIT(*str) is true.
1105
e7c1e6c1 1106=head1 Character case changing
21da7284
KW
1107Perl uses "full" Unicode case mappings. This means that converting a single
1108character to another case may result in a sequence of more than one character.
1109For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1110character sequence C<SS>. This presents some complications The lowercase of
1111all characters in the range 0..255 is a single character, and thus
1112C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1113return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1114an API that does allow every possible legal result to be returned.) Likewise
1115no other function that is crippled by not being able to give the correct
1116results for the full range of possible inputs has been implemented here.
e7c1e6c1 1117
d713f9d9 1118=for apidoc Am|U8|toUPPER|int ch
1f607577
KW
1119Converts the specified character to uppercase. If the input is anything but an
1120ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 1121C<toUPPER_A> is equivalent.
954c1994 1122
d0da05db
KW
1123=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1124Converts the code point C<cp> to its uppercase 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 uppercase version may be longer than the original character.
1129
1130The first code point of the uppercased 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
059703b0 1134=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
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 uppercase 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 uppercase version may be longer than the original character.
1140
1141The first code point of the uppercased 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
059703b0
KW
1145It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1146S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1147the UTF-8 for the input character is malformed in some way, the program may
1148croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1149of the implementation, and subject to change in future releases.
a239b1e2 1150
059703b0
KW
1151=for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1152Same as L</toUPPER_utf8>.
1f607577 1153
25200305
KW
1154=for apidoc Am|U8|toFOLD|U8 ch
1155Converts the specified character to foldcase. If the input is anything but an
1156ASCII uppercase character, that input character itself is returned. Variant
1157C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 1158Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 1159
d0da05db
KW
1160=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1161Converts the code point C<cp> to its foldcase version, and
1162stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1163point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1164that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1165bytes since the foldcase version may be longer than the original character.
1166
1167The first code point of the foldcased version is returned
21da7284
KW
1168(but note, as explained at L<the top of this section|/Character case
1169changing>, that there may be more).
1f607577 1170
059703b0 1171=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1172Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1173extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
1174stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1175that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1176bytes since the foldcase version may be longer than the original character.
1177
1178The first code point of the foldcased version is returned
21da7284
KW
1179(but note, as explained at L<the top of this section|/Character case
1180changing>, that there may be more).
1f607577 1181
059703b0 1182It will not attempt
a239b1e2
KW
1183to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1184true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1185input character is malformed in some way, the program may croak, or the
1186function may return the REPLACEMENT CHARACTER, at the discretion of the
1187implementation, and subject to change in future releases.
1188
059703b0
KW
1189=for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1190Same as L</toFOLD_utf8>.
1f607577
KW
1191
1192=for apidoc Am|U8|toLOWER|U8 ch
1193Converts the specified character to lowercase. If the input is anything but an
1194ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 1195C<toLOWER_A> is equivalent.
954c1994 1196
1f607577 1197=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
1198Converts the specified Latin1 character to lowercase. The results are
1199undefined if the input doesn't fit in a byte.
1f607577
KW
1200
1201=for apidoc Am|U8|toLOWER_LC|U8 ch
1202Converts the specified character to lowercase using the current locale's rules,
1203if possible; otherwise returns the input character itself.
1204
d0da05db
KW
1205=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1206Converts the code point C<cp> to its lowercase version, and
1207stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1208point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1209that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1210bytes since the lowercase version may be longer than the original character.
1211
1212The first code point of the lowercased version is returned
21da7284
KW
1213(but note, as explained at L<the top of this section|/Character case
1214changing>, that there may be more).
1f607577 1215
bd350c85 1216=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1217Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1218extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
1219stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1220that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1221bytes since the lowercase version may be longer than the original character.
1222
1223The first code point of the lowercased version is returned
21da7284
KW
1224(but note, as explained at L<the top of this section|/Character case
1225changing>, that there may be more).
059703b0
KW
1226It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1227S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1228the UTF-8 for the input character is malformed in some way, the program may
1229croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1230of the implementation, and subject to change in future releases.
1f607577 1231
059703b0
KW
1232=for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1233Same as L</toLOWER_utf8>.
1f607577 1234
25200305
KW
1235=for apidoc Am|U8|toTITLE|U8 ch
1236Converts the specified character to titlecase. If the input is anything but an
1237ASCII lowercase character, that input character itself is returned. Variant
b7d90381 1238C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 1239range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 1240not a concept used in locale handling, so there is no functionality for that.)
25200305 1241
d0da05db
KW
1242=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1243Converts the code point C<cp> to its titlecase version, and
1244stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1245point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1246that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1247bytes since the titlecase version may be longer than the original character.
1248
1249The first code point of the titlecased version is returned
21da7284
KW
1250(but note, as explained at L<the top of this section|/Character case
1251changing>, that there may be more).
1f607577 1252
059703b0 1253=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1254Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1255extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1256stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1257that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1258bytes since the titlecase version may be longer than the original character.
1259
1260The first code point of the titlecased version is returned
21da7284
KW
1261(but note, as explained at L<the top of this section|/Character case
1262changing>, that there may be more).
1f607577 1263
059703b0 1264It will not attempt
a239b1e2
KW
1265to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1266true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1267input character is malformed in some way, the program may croak, or the
1268function may return the REPLACEMENT CHARACTER, at the discretion of the
1269implementation, and subject to change in future releases.
1270
059703b0
KW
1271=for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1272Same as L</toTITLE_utf8>.
1f607577 1273
954c1994 1274=cut
353c9b6f 1275
d0da05db 1276XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1277really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1278and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1279
8a58bdcf 1280Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1281patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1282
954c1994
GS
1283*/
1284
8f5283f4
KW
1285/*
1286 void below because that's the best fit, and works for Devel::PPPort
1287=for apidoc AmnU|void|WIDEST_UTYPE
1288
1289Yields the widest unsigned integer type on the platform, currently either
1290C<U32> or C<64>. This can be used in declarations such as
1291
1292 WIDEST_UTYPE my_uv;
1293
1294or casts
1295
1296 my_uv = (WIDEST_UTYPE) val;
1297
1298=cut
1299
1300*/
de9e2639
KW
1301#ifdef QUADKIND
1302# define WIDEST_UTYPE U64
7c062697
KW
1303#else
1304# define WIDEST_UTYPE U32
1305#endif
1306
3912bc88
KW
1307/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1308 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1309 * bits of information are lost even on a 64-bit machine, but to get the
1310 * compiler to optimize it out if possible. This is because Configure makes
1311 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1312 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1313 *
1314 * For Coverity, be always true, because otherwise Coverity thinks
1315 * it finds several expressions that are always true, independent
1316 * of operands. Well, they are, but that is kind of the point.
220c71bf 1317 */
7e75d1a1 1318#ifndef __COVERITY__
6c5b02ac
KW
1319 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1320 * pointer) */
1321#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1322 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1323#else
1324#define FITS_IN_8_BITS(c) (1)
1325#endif
cf301eb7 1326
45f4bb73 1327/* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
833b0f46 1328 * Written this way so that after optimization, only one conditional test is
76d3ad4c
KW
1329 * needed. (The NV casts stop any warnings about comparison always being true
1330 * if called with an unsigned. The cast preserves the sign, which is all we
1331 * care about.) */
1332#define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1333 __ASSERT_((NV) (n) >= 0) \
1334 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
833b0f46 1335
94250c4f
KW
1336/* Returns true if c is in the range l..u, where 'l' is non-negative
1337 * Written this way so that after optimization, only one conditional test is
4758c20d 1338 * needed. */
1eaefa6e 1339#define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
4758c20d 1340 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
4758c20d
KW
1341 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1342 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
45f4bb73 1343 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
305fe86e 1344
41f43cc2 1345#ifdef EBCDIC
b6340bd0 1346# ifndef _ALL_SOURCE
0852beac
KW
1347 /* The native libc isascii() et.al. functions return the wrong results
1348 * on at least z/OS unless this is defined. */
b6340bd0
KW
1349# error _ALL_SOURCE should probably be defined
1350# endif
41f43cc2 1351#else
0852beac
KW
1352 /* There is a simple definition of ASCII for ASCII platforms. But the
1353 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1354 * other macros below.
3f3c579d
KW
1355 *
1356 * The cast here is used instead of '(c) >= 0', because some compilers emit
1357 * a warning that that test is always true when the parameter is an
1358 * unsigned type. khw supposes that it could be written as
1359 * && ((c) == '\0' || (c) > 0)
1360 * to avoid the message, but the cast will likely avoid extra branches even
1361 * with stupid compilers.
1362 *
1363 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1364 * a pointer) */
9c903d59 1365# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1366#endif
1367
38694112
KW
1368/* Take the eight possible bit patterns of the lower 3 bits and you get the
1369 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1370 * can be ignored. If the rest match '0', we have an octal */
1371#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1372
9fb1bf9d 1373#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1374
a500dc72
KW
1375/* Character class numbers. For internal core Perl use only. The ones less
1376 * than 32 are used in PL_charclass[] and the ones up through the one that
1377 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1378 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1379 * their actual definitions are here. If that file has a name not used here,
1380 * it won't compile.
1709d539
KW
1381 *
1382 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561 1383 * frequency of their use. This gives a slight edge to exiting a loop earlier
58a3ba2c
KW
1384 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1385 * real Posix character class, and some (small) inefficiencies in regular
1386 * expression handling would be introduced by putting it in the middle of those
1387 * that are. Also, cntrl and ascii come after the others as it may be useful
1388 * to group these which have no members that match above Latin1, (or above
1389 * ASCII in the latter case) */
1390
1709d539
KW
1391# define _CC_WORDCHAR 0 /* \w and [:word:] */
1392# define _CC_DIGIT 1 /* \d and [:digit:] */
1393# define _CC_ALPHA 2 /* [:alpha:] */
1394# define _CC_LOWER 3 /* [:lower:] */
1395# define _CC_UPPER 4 /* [:upper:] */
1396# define _CC_PUNCT 5 /* [:punct:] */
1397# define _CC_PRINT 6 /* [:print:] */
15861f94 1398# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1399# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1400# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
779cf272 1401# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1402# define _CC_BLANK 11 /* [:blank:] */
1403# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1404# define _CC_CNTRL 13 /* [:cntrl:] */
1405# define _CC_ASCII 14 /* [:ascii:] */
1406# define _CC_VERTSPACE 15 /* \v */
1709d539 1407
a0947d7b
KW
1408# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1409
1709d539 1410/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1411 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1412# define _CC_IDFIRST 16
1413# define _CC_CHARNAME_CONT 17
1414# define _CC_NONLATIN1_FOLD 18
1415# define _CC_NONLATIN1_SIMPLE_FOLD 19
1416# define _CC_QUOTEMETA 20
1417# define _CC_NON_FINAL_FOLD 21
1418# define _CC_IS_IN_SOME_FOLD 22
2ae9030c
KW
1419# define _CC_BINDIGIT 23
1420# define _CC_OCTDIGIT 24
1421# define _CC_MNEMONIC_CNTRL 25
073c22b3
KW
1422
1423/* This next group is only used on EBCDIC platforms, so theoretically could be
1424 * shared with something entirely different that's only on ASCII platforms */
abb8abf6 1425# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
0654f0ab 1426/* Unused: 26-30
f4cdb42c
KW
1427 * If more bits are needed, one could add a second word for non-64bit
1428 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1429 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1430 * is used only for optimization (as of this writing), and differs in the
1431 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1432 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1433 * would just cause /i regexes which match them to run less efficiently.
1434 * Similarly the EBCDIC-only bits are used just for speed, and could be
1435 * replaced by other means */
96ac0975 1436
3a371f2f
KW
1437#if defined(PERL_CORE) || defined(PERL_EXT)
1438/* An enum version of the character class numbers, to help compilers
1439 * optimize */
1440typedef enum {
3a371f2f 1441 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1442 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1443 _CC_ENUM_ASCII = _CC_ASCII,
1444 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1445 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1446 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1447 _CC_ENUM_DIGIT = _CC_DIGIT,
1448 _CC_ENUM_GRAPH = _CC_GRAPH,
1449 _CC_ENUM_LOWER = _CC_LOWER,
1450 _CC_ENUM_PRINT = _CC_PRINT,
1451 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1452 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1453 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1454 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1455 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1456 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1457} _char_class_number;
1458#endif
1459
86f72d56 1460#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1461
6635f04f 1462START_EXTERN_C
96ac0975
NC
1463# ifdef DOINIT
1464EXTCONST U32 PL_charclass[] = {
1465# include "l1_char_class_tab.h"
1466};
1467
1468# else /* ! DOINIT */
1469EXTCONST U32 PL_charclass[];
1470# endif
6635f04f 1471END_EXTERN_C
96ac0975 1472
265c1f46 1473 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1474# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1475
1476 /* For internal core Perl use only: the base macro for defining macros like
1477 * isALPHA */
ff7ecfc3 1478# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1479 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1480
f4cdb42c
KW
1481 /* The mask for the _A versions of the macros; it just adds in the bit for
1482 * ASCII. */
1483# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1484
4650c663
KW
1485 /* For internal core Perl use only: the base macro for defining macros like
1486 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1487 * the ASCII bit are present */
b7d90381
KW
1488# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1489 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1490 == _CC_mask_A(classnum)))
f4cdb42c 1491
26c1d9d8
KW
1492/* On ASCII platforms certain classes form a single range. It's faster to
1493 * special case these. isDIGIT is a single range on all platforms */
b877c1ff
KW
1494# ifdef EBCDIC
1495# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1496# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1497# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1498# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1499# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1500# else
26c1d9d8 1501 /* By folding the upper and lowercase, we can use a single range */
b877c1ff 1502# define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
26c1d9d8 1503# define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
b877c1ff
KW
1504# define isLOWER_A(c) inRANGE(c, 'a', 'z')
1505# define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1506# define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1507# endif
15861f94 1508# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1509# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1510# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
b877c1ff 1511# define isDIGIT_A(c) inRANGE(c, '0', '9')
f4cdb42c
KW
1512# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1513# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
f4cdb42c 1514# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1515# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1516 */
d95f8b6a 1517# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1518# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1519# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1520# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1521
1522 /* continuation character for legal NAME in \N{NAME} */
1523# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1524
1525# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1526# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1527# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1528# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1529# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1530# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1531# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1532# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1533# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1534# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1535
0852beac
KW
1536# ifdef EBCDIC
1537# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1538# endif
1539
f12c0118
KW
1540 /* Participates in a single-character fold with a character above 255 */
1541# 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)))
1542
1543 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1544# 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 1545
4eeeb416 1546# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1547# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1548 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1549# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1550 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
5e6ebb12
KW
1551
1552/* is c a control character for which we have a mnemonic? */
1553# if defined(PERL_CORE) || defined(PERL_EXT)
1554# define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1555# endif
687c8d01 1556#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1557
1558 /* If we don't have perl.h, we are compiling a utility program. Below we
1559 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1560 * to it. Most are coded based on first principles. These are written to
74665a89 1561 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
182c4ace 1562# define isDIGIT_A(c) inRANGE(c, '0', '9')
0852beac 1563# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1564# define isSPACE_A(c) (isBLANK_A(c) \
1565 || (c) == '\n' \
1566 || (c) == '\r' \
1567 || (c) == '\v' \
0852beac 1568 || (c) == '\f')
74665a89
KW
1569 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1570 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1571 * performance (if optimization isn't on), and allow the same code to be
1572 * used for both platform types */
182c4ace
KW
1573# define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1574 || inRANGE((c), 'j', 'r') \
1575 || inRANGE((c), 's', 'z')
1576# define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1577 || inRANGE((c), 'J', 'R') \
1578 || inRANGE((c), 'S', 'Z')
a4d7a999
KW
1579# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1580# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1581# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1582# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
182c4ace
KW
1583# define isXDIGIT_A(c) ( isDIGIT_A(c) \
1584 || inRANGE((c), 'a', 'f') \
1585 || inRANGE((c), 'A', 'F')
74665a89
KW
1586# define isPUNCT_A(c) ((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) == '^' || (c) == '_' \
1595 || (c) == '`' || (c) == '{' || (c) == '|' \
1596 || (c) == '}' || (c) == '~')
1597# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1598# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1599
0852beac 1600# ifdef EBCDIC
74665a89
KW
1601 /* The below is accurate for the 3 EBCDIC code pages traditionally
1602 * supported by perl. The only difference between them in the controls
1603 * is the position of \n, and that is represented symbolically below */
1604# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1605 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1606 || (c) == '\t' || (c) == '\v' \
182c4ace 1607 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
8ec0a736 1608 || (c) == 7F /* U+7F DEL */ \
182c4ace
KW
1609 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1610 DC[1-3] */ \
74665a89
KW
1611 || (c) == 0x18 /* U+18 CAN */ \
1612 || (c) == 0x19 /* U+19 EOM */ \
182c4ace 1613 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
74665a89
KW
1614 || (c) == 0x26 /* U+17 ETB */ \
1615 || (c) == 0x27 /* U+1B ESC */ \
1616 || (c) == 0x2D /* U+05 ENQ */ \
1617 || (c) == 0x2E /* U+06 ACK */ \
1618 || (c) == 0x32 /* U+16 SYN */ \
1619 || (c) == 0x37 /* U+04 EOT */ \
1620 || (c) == 0x3C /* U+14 DC4 */ \
1621 || (c) == 0x3D /* U+15 NAK */ \
1622 || (c) == 0x3F)/* U+1A SUB */
0852beac 1623# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1624# else /* isASCII is already defined for ASCII platforms, so can use that to
1625 define isCNTRL */
1626# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1627# endif
1628
3ffc8c70 1629 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1630 * during debugging, and it seems best to keep them. We may be called
1631 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1632 * do anything anyway, so make it not a problem */
1633# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1634# define NATIVE_TO_LATIN1(ch) (ch)
1635# endif
3ded5eb0
KW
1636# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1637# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1638# define isBLANK_L1(c) (isBLANK_A(c) \
1639 || (FITS_IN_8_BITS(c) \
1640 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1641# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1642# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1643# define isLOWER_L1(c) (isLOWER_A(c) \
1644 || (FITS_IN_8_BITS(c) \
ae683a5f 1645 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1646 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1647 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1648 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1649 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1650# define isPRINT_L1(c) (isPRINT_A(c) \
1651 || (FITS_IN_8_BITS(c) \
1652 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1653# define isPUNCT_L1(c) (isPUNCT_A(c) \
1654 || (FITS_IN_8_BITS(c) \
ae683a5f 1655 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1656 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1657 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1658 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1659 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1660 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1661 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1662# define isSPACE_L1(c) (isSPACE_A(c) \
1663 || (FITS_IN_8_BITS(c) \
ae683a5f 1664 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1665 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1666# define isUPPER_L1(c) (isUPPER_A(c) \
1667 || (FITS_IN_8_BITS(c) \
182c4ace
KW
1668 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1669 0xC0, 0xDE) \
3ded5eb0
KW
1670 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1671# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1672# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1673# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1674 || isBLANK_L1(c) \
1675 || (c) == '-' \
1676 || (c) == '(' \
1677 || (c) == ')')
1678 /* The following are not fully accurate in the above-ASCII range. I (khw)
1679 * don't think it's necessary to be so for the purposes where this gets
1680 * compiled */
1681# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1682# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1683
1684 /* And these aren't accurate at all. They are useful only for above
1685 * Latin1, which utilities and bootstrapping don't deal with */
1686# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1687# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1688# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1689
1690 /* Many of the macros later in this file are defined in terms of these. By
1691 * implementing them with a function, which converts the class number into
1692 * a call to the desired macro, all of the later ones work. However, that
1693 * function won't be actually defined when building a utility program (no
1694 * perl.h), and so a compiler error will be generated if one is attempted
1695 * to be used. And the above-Latin1 code points require Unicode tables to
1696 * be present, something unlikely to be the case when bootstrapping */
1697# define _generic_isCC(c, classnum) \
1698 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1699# define _generic_isCC_A(c, classnum) \
1700 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1701#endif /* End of no perl.h H_PERL */
8a58bdcf 1702
e66b99e9
KW
1703#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1704#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1705#define isASCII_A(c) isASCII(c)
1706#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1707#define isBLANK(c) isBLANK_A(c)
1708#define isCNTRL(c) isCNTRL_A(c)
1709#define isDIGIT(c) isDIGIT_A(c)
1710#define isGRAPH(c) isGRAPH_A(c)
1711#define isIDFIRST(c) isIDFIRST_A(c)
1712#define isLOWER(c) isLOWER_A(c)
1713#define isPRINT(c) isPRINT_A(c)
779cf272 1714#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1715#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1716#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1717#define isPUNCT(c) isPUNCT_A(c)
1718#define isSPACE(c) isSPACE_A(c)
1719#define isUPPER(c) isUPPER_A(c)
1720#define isWORDCHAR(c) isWORDCHAR_A(c)
1721#define isXDIGIT(c) isXDIGIT_A(c)
1722
1723/* ASCII casing. These could also be written as
1724 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1725 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1726 which uses table lookup and mask instead of subtraction. (This would
c5e9991e
KW
1727 work because the _MOD does not apply in the ASCII range).
1728
1729 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1730 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1731 there non-ASCII invariants, and all of them are controls.) */
68067e4e
DM
1732#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1733#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1734
25200305
KW
1735/* In the ASCII range, these are equivalent to what they're here defined to be.
1736 * But by creating these definitions, other code doesn't have to be aware of
c5e9991e
KW
1737 * this detail. Actually this works for all UTF-8 invariants, not just the
1738 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
25200305 1739#define toFOLD(c) toLOWER(c)
25200305
KW
1740#define toTITLE(c) toUPPER(c)
1741
c753c8d3
KW
1742#define toLOWER_A(c) toLOWER(c)
1743#define toUPPER_A(c) toUPPER(c)
25200305
KW
1744#define toFOLD_A(c) toFOLD(c)
1745#define toTITLE_A(c) toTITLE(c)
1a0901db 1746
4650c663 1747/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1748#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1749 ? (c) \
f4cd282c 1750 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1751#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1752
1a0901db 1753/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1754 * all mapped to one of them, and these need special handling; returns the
1755 * input itself if is out-of-range */
b2bf251f 1756#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1757 ? (c) \
f4cd282c 1758 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1759#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1760
beab9ebe
KW
1761/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1762
1763/* For internal core Perl use only: the base macro for defining macros like
1764 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1765 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1766 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1767 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1768 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1769 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1770 * code points under locale rules, use the macros based on _generic_LC_uvchr
1771 * instead of this. */
beab9ebe
KW
1772#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1773 (! FITS_IN_8_BITS(c) \
1774 ? 0 \
31f05a37
KW
1775 : IN_UTF8_CTYPE_LOCALE \
1776 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1777 : cBOOL(non_utf8))
1778
1779/* For internal core Perl use only: a helper macro for defining macros like
1780 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1781 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1782 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1783 * _generic_LC_base for more info */
1784#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1785 _generic_LC_base(c,utf8_locale_classnum, \
1786 non_utf8_func( (_LC_CAST) (c)))
1787
1788/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1789 * 'c' is the platform's native underscore character */
1790#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1791 _generic_LC_base(c, utf8_locale_classnum, \
1792 (non_utf8_func( (_LC_CAST) (c)) \
1793 || (char)(c) == '_'))
1794
1795/* These next three are also for internal core Perl use only: case-change
247985d4
KW
1796 * helper macros. The reason for using the PL_latin arrays is in case the
1797 * system function is defective; it ensures uniform results that conform to the
b257a28c 1798 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
beab9ebe
KW
1799#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1800 ? (c) \
31f05a37
KW
1801 : (IN_UTF8_CTYPE_LOCALE) \
1802 ? PL_latin1_lc[ (U8) (c) ] \
5a10328c 1803 : (cast)function((cast)(c)))
beab9ebe 1804
31f05a37
KW
1805/* Note that the result can be larger than a byte in a UTF-8 locale. It
1806 * returns a single value, so can't adequately return the upper case of LATIN
1807 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1808 * values "SS"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1809 * otherwise returns its input. It does not handle the anomalies in UTF-8
1810 * Turkic locales. */
beab9ebe
KW
1811#define _generic_toUPPER_LC(c, function, cast) \
1812 (! FITS_IN_8_BITS(c) \
1813 ? (c) \
31f05a37 1814 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1815 ? (cast)function((cast)(c)) \
31f05a37
KW
1816 : ((((U8)(c)) == MICRO_SIGN) \
1817 ? GREEK_CAPITAL_LETTER_MU \
1818 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1819 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1820 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1821 ? (__ASSERT_(0) (c)) \
1822 : PL_mod_latin1_uc[ (U8) (c) ])))))
1823
1824/* Note that the result can be larger than a byte in a UTF-8 locale. It
1825 * returns a single value, so can't adequately return the fold case of LATIN
1826 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1827 * values "ss"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1828 * otherwise returns its input. It does not handle the anomalies in UTF-8
1829 * Turkic locales */
beab9ebe 1830#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1831 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1832 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1833 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1834 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1835 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1836
84061b6a 1837/* Use the libc versions for these if available. */
f05550c0 1838#if defined(HAS_ISASCII)
84061b6a
KW
1839# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1840#else
1841# define isASCII_LC(c) isASCII(c)
1842#endif
1843
f05550c0 1844#if defined(HAS_ISBLANK)
84061b6a 1845# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1846#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1847# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1848#endif
1849
f05550c0 1850#define _LC_CAST U8
bbce6d69 1851
f05550c0 1852#ifdef WIN32
375f5f06
KW
1853 /* The Windows functions don't bother to follow the POSIX standard, which
1854 * for example says that something can't both be a printable and a control.
1855 * But Windows treats the \t control as a printable, and does such things
1856 * as making superscripts into both digits and punctuation. This tames
1857 * these flaws by assuming that the definitions of both controls and space
1858 * are correct, and then making sure that other definitions don't have
1859 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1860 * Not all possible weirdnesses are checked for, just the ones that were
1861 * detected on actual Microsoft code pages */
1862
b7d90381
KW
1863# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1864# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1865
1866# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1867 && isALPHANUMERIC_LC(c))
1868# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1869 ! isPUNCT_LC(c))
1870# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1871 isALPHANUMERIC_LC(c))
1872# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1873# define isIDFIRST_LC(c) (((c) == '_') \
1874 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1875# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1876# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1877# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1878# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1879# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1880# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1881 && isALPHANUMERIC_LC(c))
f05550c0
BF
1882
1883# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1884# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1885# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1886
1887#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1888 /* For most other platforms */
beab9ebe 1889
f05550c0
BF
1890# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1891# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1892# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1893# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1894# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1895# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1896# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1897# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1898# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1899# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1900# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1901# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1902# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1903
1904
1905# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1906# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1907# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1908
1909#else /* The final fallback position */
1910
b7d90381
KW
1911# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1912# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1913# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1914# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1915# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1916# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1917# define isLOWER_LC(c) (isascii(c) && islower(c))
1918# define isPRINT_LC(c) (isascii(c) && isprint(c))
1919# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1920# define isSPACE_LC(c) (isascii(c) && isspace(c))
1921# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1922# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1923# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1924
1925# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1926# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1927# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1928
f05550c0 1929#endif
55204971 1930
eba68aa0
KW
1931#define isIDCONT(c) isWORDCHAR(c)
1932#define isIDCONT_A(c) isWORDCHAR_A(c)
1933#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1934#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1935#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1936
4650c663 1937/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1938 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1939 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1940 * classes where there is a macro or function 'above_latin1' that takes the
1941 * single argument 'c' and returns the desired value. These exist for those
2366ba44
KW
1942 * classes which have simple definitions, avoiding the overhead of an inversion
1943 * list binary search. _generic_invlist_uvchr() can be used
4650c663 1944 * for classes where that overhead is faster than a direct lookup.
d0da05db 1945 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1946 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1947 * have duplicate checks here, so could create versions of the macros that
1948 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1949
1950/* Note that all ignore 'use bytes' */
1e222e4f
KW
1951#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1952 ? _generic_isCC(c, classnum) \
cd500f2f 1953 : above_latin1(c))
2366ba44 1954#define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
1e222e4f 1955 ? _generic_isCC(c, classnum) \
922e8cb4 1956 : _is_uni_FOO(classnum, c))
2366ba44
KW
1957#define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
1958#define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
d0da05db
KW
1959#define isASCII_uvchr(c) isASCII(c)
1960#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1961#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2366ba44
KW
1962#define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
1963#define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1964#define isIDCONT_uvchr(c) \
1965 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1966#define isIDFIRST_uvchr(c) \
1967 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2366ba44
KW
1968#define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
1969#define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
d0da05db 1970
2366ba44 1971#define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
d0da05db
KW
1972#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1973#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1974
2366ba44 1975#define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
d0da05db 1976#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2366ba44 1977#define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
d0da05db
KW
1978#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1979
1980#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1981#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1982#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1983#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1984
1985/* For backwards compatibility, even though '_uni' should mean official Unicode
1986 * code points, in Perl it means native for those below 256 */
1987#define isALPHA_uni(c) isALPHA_uvchr(c)
1988#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1989#define isASCII_uni(c) isASCII_uvchr(c)
1990#define isBLANK_uni(c) isBLANK_uvchr(c)
1991#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1992#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1993#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1994#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1995#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1996#define isLOWER_uni(c) isLOWER_uvchr(c)
1997#define isPRINT_uni(c) isPRINT_uvchr(c)
1998#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1999#define isSPACE_uni(c) isSPACE_uvchr(c)
2000#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
2001#define isUPPER_uni(c) isUPPER_uvchr(c)
2002#define isVERTWS_uni(c) isVERTWS_uvchr(c)
2003#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2004#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2005#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2006#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2007#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2008#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 2009
4650c663
KW
2010/* For internal core Perl use only: the base macros for defining macros like
2011 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 2012 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 2013 * simple class definitions; the other for more complex. These are like
d0da05db 2014 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
2015#define _generic_LC_uvchr(latin1, above_latin1, c) \
2016 (c < 256 ? latin1(c) : above_latin1(c))
2366ba44 2017#define _generic_LC_invlist_uvchr(latin1, classnum, c) \
cd500f2f
KW
2018 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2019
2366ba44
KW
2020#define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2021#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
cd500f2f 2022 _CC_ALPHANUMERIC, c)
b7d90381
KW
2023#define isASCII_LC_uvchr(c) isASCII_LC(c)
2024#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2025 is_HORIZWS_cp_high, c)
feeab5a9 2026#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2366ba44
KW
2027#define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2028#define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 2029#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 2030 _is_uni_perl_idcont, c)
b7d90381 2031#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f 2032 _is_uni_perl_idstart, c)
2366ba44
KW
2033#define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2034#define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 2035#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2366ba44 2036#define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 2037#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 2038 is_XPERLSPACE_cp_high, c)
2366ba44
KW
2039#define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2040#define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
cd500f2f 2041 _CC_WORDCHAR, c)
b7d90381
KW
2042#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2043 is_XDIGIT_cp_high, c)
e712593e 2044
b7d90381 2045#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 2046
da8c1a98
KW
2047/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2048 * they don't otherwise go out of their way to look for malformed UTF-8. If
2049 * they can return accurate results without knowing if the input is otherwise
2050 * malformed, they do so. For example isASCII is accurate in spite of any
2051 * non-length malformations because it looks only at a single byte. Likewise
2052 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2053 * variant ones return FALSE. But, if the input has to be well-formed in order
2054 * for the results to be accurate, the macros will test and if malformed will
2055 * call a routine to die
2056 *
2057 * Except for toke.c, the macros do assume that e > p, asserting that on
2058 * DEBUGGING builds. Much code that calls these depends on this being true,
2059 * for other reasons. toke.c is treated specially as using the regular
2060 * assertion breaks it in many ways. All strings that these operate on there
2061 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2062 * bunch of code in toke.c assumes that this is true, so the assertion allows
2063 * for that */
2064#ifdef PERL_IN_TOKE_C
2065# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2066#else
2067# define _utf8_safe_assert(p,e) ((e) > (p))
2068#endif
2069
2070#define _generic_utf8_safe(classnum, p, e, above_latin1) \
c81b3562
KW
2071 ((! _utf8_safe_assert(p, e)) \
2072 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2073 : (UTF8_IS_INVARIANT(*(p))) \
da8c1a98
KW
2074 ? _generic_isCC(*(p), classnum) \
2075 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2076 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2077 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2078 classnum) \
2079 : (_force_out_malformed_utf8_message( \
2080 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2081 : above_latin1))
b7d90381
KW
2082/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2083 * 'above_latin1' can be a macro */
da8c1a98
KW
2084#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2085 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2366ba44 2086#define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2087 _generic_utf8_safe(classnum, p, e, \
2088 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2089 ? (_force_out_malformed_utf8_message( \
2090 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2091 : above_latin1(p)))
2366ba44
KW
2092/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2093 * 'above_latin1' parameter */
2094#define _generic_invlist_utf8_safe(classnum, p, e) \
2095 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
922e8cb4 2096
cc8ab7c0 2097/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
2098 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2099 * class is TRUE for. Hence it can skip the tests for this range.
2100 * 'above_latin1' should include its arguments */
da8c1a98
KW
2101#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2102 (__ASSERT_(_utf8_safe_assert(p, e)) \
2103 (UTF8_IS_INVARIANT(*(p))) \
2104 ? _generic_isCC(*(p), classnum) \
2105 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2106 ? 0 /* Note that doesn't check validity for latin1 */ \
2107 : above_latin1)
2108
84238efa 2109
059703b0
KW
2110#define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2111#define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2112#define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2113#define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2114#define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2115#define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2116#define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2117#define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2118#define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2119#define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2120#define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2121#define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2122#define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2123#define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2124#define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2125#define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2126#define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2127#define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
e8fa43e2 2128
2366ba44 2129#define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
da8c1a98 2130#define isALPHANUMERIC_utf8_safe(p, e) \
2366ba44 2131 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
da8c1a98
KW
2132#define isASCII_utf8_safe(p, e) \
2133 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2134 * works */ \
2135 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2136#define isBLANK_utf8_safe(p, e) \
2366ba44 2137 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
da8c1a98 2138
e8fa43e2
KW
2139#ifdef EBCDIC
2140 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2141 * more efficient macro instead of the more general one */
da8c1a98 2142# define isCNTRL_utf8_safe(p, e) \
56d02b8c 2143 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 2144#else
da8c1a98 2145# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
2146#endif
2147
da8c1a98
KW
2148#define isDIGIT_utf8_safe(p, e) \
2149 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2366ba44
KW
2150 _is_utf8_FOO(_CC_DIGIT, p, e))
2151#define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
da8c1a98 2152#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
dd1a3ba7 2153 _is_utf8_perl_idcont, p, e)
e5dcd934 2154
c11ff943
KW
2155/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2156 * IDFIRST is an alnum. See
8034715d 2157 * https://github.com/Perl/perl5/issues/10275 for more detail than you
f91dcd13
KW
2158 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2159 * This used to be not the XID version, but we decided to go with the more
2160 * modern Unicode definition */
da8c1a98
KW
2161#define isIDFIRST_utf8_safe(p, e) \
2162 _generic_func_utf8_safe(_CC_IDFIRST, \
dd1a3ba7 2163 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
da8c1a98 2164
2366ba44
KW
2165#define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2166#define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
da8c1a98 2167#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2366ba44 2168#define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
da8c1a98 2169#define isSPACE_utf8_safe(p, e) \
2366ba44
KW
2170 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2171#define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
da8c1a98 2172#define isVERTWS_utf8_safe(p, e) \
2366ba44 2173 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
da8c1a98 2174#define isWORDCHAR_utf8_safe(p, e) \
2366ba44 2175 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
da8c1a98
KW
2176#define isXDIGIT_utf8_safe(p, e) \
2177 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2178 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2179 ? (_force_out_malformed_utf8_message( \
2180 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2181 : is_XDIGIT_high(p)))
a0ed51b3 2182
059703b0
KW
2183#define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2184#define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2185#define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2186#define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2e8adce6 2187
567b353c 2188/* For internal core use only, subject to change */
059703b0
KW
2189#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2190#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2191#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2192#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
a1a5ec35
KW
2193
2194#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2195#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2196#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2197#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 2198
059703b0
KW
2199#define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2200#define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2201#define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2202#define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2203#define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2204#define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2205#define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2206#define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2207#define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2208#define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2209#define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2210#define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2211#define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2212#define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2213#define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2214#define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2215#define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
34aeb2e9 2216
da8c1a98
KW
2217/* For internal core Perl use only: the base macros for defining macros like
2218 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2219 * point in 'p' is within the 0-255 range, it uses locale rules from the
2220 * passed-in 'macro' parameter */
2221#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2222 (__ASSERT_(_utf8_safe_assert(p, e)) \
2223 (UTF8_IS_INVARIANT(*(p))) \
2224 ? macro(*(p)) \
2225 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2226 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2227 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2228 : (_force_out_malformed_utf8_message( \
2229 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2230 : above_latin1))
2231
2366ba44
KW
2232#define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2233 _generic_LC_utf8_safe(macro, p, e, \
2234 _is_utf8_FOO(classnum, p, e))
da8c1a98
KW
2235
2236#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2237 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2238
2366ba44 2239#define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2240 _generic_LC_utf8_safe(classnum, p, e, \
2241 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2242 ? (_force_out_malformed_utf8_message( \
2243 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2244 : above_latin1(p)))
2245
2246#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2366ba44 2247 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
da8c1a98
KW
2248 _CC_ALPHANUMERIC, p, e)
2249#define isALPHA_LC_utf8_safe(p, e) \
2366ba44 2250 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
da8c1a98
KW
2251#define isASCII_LC_utf8_safe(p, e) \
2252 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2253#define isBLANK_LC_utf8_safe(p, e) \
2366ba44 2254 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
da8c1a98
KW
2255#define isCNTRL_LC_utf8_safe(p, e) \
2256 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2257#define isDIGIT_LC_utf8_safe(p, e) \
2366ba44 2258 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
da8c1a98 2259#define isGRAPH_LC_utf8_safe(p, e) \
2366ba44 2260 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
da8c1a98
KW
2261#define isIDCONT_LC_utf8_safe(p, e) \
2262 _generic_LC_func_utf8_safe(isIDCONT_LC, \
dd1a3ba7 2263 _is_utf8_perl_idcont, p, e)
da8c1a98
KW
2264#define isIDFIRST_LC_utf8_safe(p, e) \
2265 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
dd1a3ba7 2266 _is_utf8_perl_idstart, p, e)
da8c1a98 2267#define isLOWER_LC_utf8_safe(p, e) \
2366ba44 2268 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
da8c1a98 2269#define isPRINT_LC_utf8_safe(p, e) \
2366ba44 2270 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
da8c1a98
KW
2271#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2272#define isPUNCT_LC_utf8_safe(p, e) \
2366ba44 2273 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
da8c1a98 2274#define isSPACE_LC_utf8_safe(p, e) \
2366ba44 2275 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
da8c1a98 2276#define isUPPER_LC_utf8_safe(p, e) \
2366ba44 2277 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
da8c1a98 2278#define isWORDCHAR_LC_utf8_safe(p, e) \
2366ba44 2279 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
da8c1a98 2280#define isXDIGIT_LC_utf8_safe(p, e) \
2366ba44 2281 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2282
fbc19f27
KW
2283/* Macros for backwards compatibility and for completeness when the ASCII and
2284 * Latin1 values are identical */
b7d90381
KW
2285#define isALPHAU(c) isALPHA_L1(c)
2286#define isDIGIT_L1(c) isDIGIT_A(c)
2287#define isOCTAL(c) isOCTAL_A(c)
2288#define isOCTAL_L1(c) isOCTAL_A(c)
2289#define isXDIGIT_L1(c) isXDIGIT_A(c)
2290#define isALNUM(c) isWORDCHAR(c)
a377c856 2291#define isALNUM_A(c) isALNUM(c)
b7d90381
KW
2292#define isALNUMU(c) isWORDCHAR_L1(c)
2293#define isALNUM_LC(c) isWORDCHAR_LC(c)
2294#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2295#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
059703b0 2296#define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
4c1d9526 2297#define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
059703b0 2298#define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
4c1d9526 2299#define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
b7d90381
KW
2300#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2301#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2302#define isALNUMC(c) isALPHANUMERIC(c)
2303#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2304#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2305#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
059703b0 2306#define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
4c1d9526
KW
2307#define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2308#define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
fbc19f27 2309
2bd1cbf6
KW
2310/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2311 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2312 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2313 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2314 * the outlier from the block that contains the other controls, just like
2315 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2316 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2317 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2318 * etc. */
2bd1cbf6 2319#ifndef EBCDIC
75763b3a 2320# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2321#else
75763b3a
KW
2322# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2323 ((isPRINT_A(c)) \
2324 ? (UNLIKELY((c) == '?') \
2325 ? QUESTION_MARK_CTRL \
2326 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2327 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2328 ? '?' \
2329 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2330#endif
bbce6d69 2331
dea28490
JJ
2332/* Line numbers are unsigned, 32 bits. */
2333typedef U32 line_t;
e5dcd934 2334#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2335
91152fc1
DG
2336/* Helpful alias for version prescan */
2337#define is_LAX_VERSION(a,b) \
2338 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2339
2340#define is_STRICT_VERSION(a,b) \
2341 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2342
2343#define BADVERSION(a,b,c) \
2344 if (b) { \
2345 *b = c; \
2346 } \
2347 return a;
8c52afec 2348
1ce77b7d
KW
2349/* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2350 * a-f) to its numeric value without using any branches. The input is
2351 * validated only by an assert() in DEBUGGING builds.
2352 *
2353 * It works by right shifting and isolating the bit that is 0 for the digits,
2354 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2355 * position, and then to the eights position. Both are added together to form
2356 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2357 * final four bits of the input to form the correct value. */
2358#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2359 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2360 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2361 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2362
2363/* The argument is a string pointer, which is advanced. */
2364#define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
95a59cab 2365
cb27eebd
KW
2366/* Converts a character known to represent an octal digit (0-7) to its numeric
2367 * value. The input is validated only by an assert() in DEBUGGING builds. In
2368 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2369#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2370
305b8651
KW
2371/* Efficiently returns a boolean as to if two native characters are equivalent
2372 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2373 * the ALPHA in the name is to remind you of that. This is asserted() in
2374 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2375 * works (on valid input) for both non- and UTF-8-encoded bytes.
2376 *
2377 * When one of the inputs is a compile-time constant and gets folded by the
2378 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2379 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2380 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2381 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2382 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2383 * just a single 0, in the bit position where the upper- and lowercase differ.
2384 * */
305b8651
KW
2385#define isALPHA_FOLD_EQ(c1, c2) \
2386 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2387 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2388#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2389
8e84507e 2390/*
ccfc67b7
JH
2391=head1 Memory Management
2392
a02a5408 2393=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2394The XSUB-writer's interface to the C C<malloc> function.
2395
596f7718 2396Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2397
c5008215
JC
2398In 5.9.3, Newx() and friends replace the older New() API, and drops
2399the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2400themselves. This aid has been superseded by a new build option,
d10b4965 2401PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2402there for use in XS modules supporting older perls.
2403
a02a5408 2404=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2405The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2406cast. See also C<L</Newx>>.
954c1994 2407
596f7718 2408Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2409
a02a5408 2410=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2411The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2412memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2413
596f7718 2414Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2415
954c1994
GS
2416=for apidoc Am|void|Renew|void* ptr|int nitems|type
2417The XSUB-writer's interface to the C C<realloc> function.
2418
596f7718 2419Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2420
954c1994
GS
2421=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2422The XSUB-writer's interface to the C C<realloc> function, with
2423cast.
2424
596f7718 2425Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2426
49b8b560 2427=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2428The XSUB-writer's interface to the C C<free> function.
2429
596f7718 2430This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2431
954c1994
GS
2432=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2433The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2434source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2435C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2436
e90e2364 2437=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2438Like C<Move> but returns C<dest>. Useful
72d33970 2439for encouraging compilers to tail-call
e90e2364
NC
2440optimise.
2441
954c1994
GS
2442=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2443The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2444source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2445C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2446
e90e2364
NC
2447=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2448
796b6530 2449Like C<Copy> but returns C<dest>. Useful
72d33970 2450for encouraging compilers to tail-call
e90e2364
NC
2451optimise.
2452
954c1994
GS
2453=for apidoc Am|void|Zero|void* dest|int nitems|type
2454
2455The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2456destination, C<nitems> is the number of items, and C<type> is the type.
2457
e90e2364
NC
2458=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2459
72d33970
FC
2460Like C<Zero> but returns dest. Useful
2461for encouraging compilers to tail-call
e90e2364
NC
2462optimise.
2463
da5d8dbb 2464=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2465This is an architecture-independent macro to copy one structure to another.
954c1994 2466
7e337ee0
JH
2467=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2468
2469Fill up memory with a byte pattern (a byte repeated over and over
2470again) that hopefully catches attempts to access uninitialized memory.
2471
2472=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2473
2474PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2475
1c12ffb4 2476=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2477
2478PoisonWith(0xEF) for catching access to freed memory.
2479
9965345d
JH
2480=for apidoc Am|void|Poison|void* dest|int nitems|type
2481
7e337ee0 2482PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2483
2484=cut */
954c1994 2485
561b68a9
SH
2486/* Maintained for backwards-compatibility only. Use newSV() instead. */
2487#ifndef PERL_CORE
ff06c60c 2488#define NEWSV(x,len) newSV(len)
561b68a9 2489#endif
ff06c60c 2490
b7112dce 2491#define MEM_SIZE_MAX ((MEM_SIZE)-1)
19a94d75 2492
a500027b 2493#define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
e6bdf523 2494
27d5b266 2495#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2496
2497/* This expression will be constant-folded at compile time. It checks
2498 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2499 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2500 * As well as avoiding the need for a run-time check in some cases, it's
2501 * designed to avoid compiler warnings like:
2502 * comparison is always false due to limited range of data type
73e8ff00
DM
2503 * It's mathematically equivalent to
2504 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2505 */
2506
2507# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
445198b9
LM
2508 ( sizeof(MEM_SIZE) < sizeof(n) \
2509 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 2510
88f9f128 2511/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2512 * compiler warnings. We *want* to write the expression as
2513 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2514 * (for some compile-time constant C), but even when the LHS
2515 * constant-folds to false at compile-time, g++ insists on emitting
2516 * warnings about the RHS (e.g. "comparison is always false"), so instead
2517 * we write it as
e6bdf523 2518 *
d98e5cde 2519 * (cond ? n : X) > C
88f9f128 2520 *
d98e5cde
DM
2521 * where X is a constant with X > C always false. Choosing a value for X
2522 * is tricky. If 0, some compilers will complain about 0 > C always being
2523 * false; if 1, Coverity complains when n happens to be the constant value
2524 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2525 * for X and hope that nothing else whines.
e6bdf523
DM
2526 */
2527
2528# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2529 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2530 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2531
2532# define MEM_WRAP_CHECK(n,t) \
2533 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2534 && (croak_memory_wrap(),0))
2535
2536# define MEM_WRAP_CHECK_1(n,t,a) \
2537 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2538 && (Perl_croak_nocontext("%s",(a)),0))
2539
814eedc8
DD
2540/* "a" arg must be a string literal */
2541# define MEM_WRAP_CHECK_s(n,t,a) \
2542 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2543 && (Perl_croak_nocontext("" a ""),0))
2544
8b44ba4c 2545#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2546
a500027b 2547#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
2548#else
2549
410319be
NC
2550#define MEM_WRAP_CHECK(n,t)
2551#define MEM_WRAP_CHECK_1(n,t,a)
814eedc8 2552#define MEM_WRAP_CHECK_s(n,t,a)
8b44ba4c
NC
2553#define MEM_WRAP_CHECK_(n,t)
2554
a500027b 2555#define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
27d5b266 2556
1936d2a7 2557#endif
8b44ba4c 2558
fe4f188c 2559#ifdef PERL_MEM_LOG
46c6c7e2 2560/*
9f653bb5 2561 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2562 * go through functions, which are handy for debugging breakpoints, but
2563 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2564 * line number, and C function name if available) passed in. This info can
2565 * then be used for logging the calls, for which one gets a sample
73d1d973 2566 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2567 *
46c6c7e2 2568 * Known problems:
94e892a6 2569 * - not all memory allocs get logged, only those
46c6c7e2 2570 * that go through Newx() and derivatives (while all
94e892a6 2571 * Safefrees do get logged)
46c6c7e2
JH
2572 * - __FILE__ and __LINE__ do not work everywhere
2573 * - __func__ or __FUNCTION__ even less so
2574 * - I think more goes on after the perlio frees but
2575 * the thing is that STDERR gets closed (as do all
2576 * the file descriptors)
2577 * - no deeper calling stack than the caller of the Newx()
2578 * or the kind, but do I look like a C reflection/introspection
2579 * utility to you?
2580 * - the function prototypes for the logging functions
2581 * probably should maybe be somewhere else than handy.h
2582 * - one could consider inlining (macrofying) the logging
2583 * for speed, but I am too lazy
2584 * - one could imagine recording the allocations in a hash,
2585 * (keyed by the allocation address?), and maintain that
2586 * through reallocs and frees, but how to do that without
2587 * any News() happening...?
73d1d973 2588 * - lots of -Ddefines to get useful/controllable output
b953482e 2589 * - lots of ENV reads
46c6c7e2
JH
2590 */
2591
0b0ab801 2592# ifdef PERL_CORE
73d1d973 2593# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2594enum mem_log_type {
2595 MLT_ALLOC,
2596 MLT_REALLOC,
d7a2c63c
MHM
2597 MLT_FREE,
2598 MLT_NEW_SV,
2599 MLT_DEL_SV
0b0ab801
MHM
2600};
2601# endif
12754f92 2602# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2603void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2604void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2605# endif
0b0ab801
MHM
2606# endif
2607
fe4f188c
JH
2608#endif
2609
2610#ifdef PERL_MEM_LOG
d1401ee9
MHM
2611#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2612#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2613#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2614#endif
2615
2616#ifndef MEM_LOG_ALLOC
2617#define MEM_LOG_ALLOC(n,t,a) (a)
2618#endif
2619#ifndef MEM_LOG_REALLOC
2620#define MEM_LOG_REALLOC(n,t,v,a) (a)
2621#endif
2622#ifndef MEM_LOG_FREE
2623#define MEM_LOG_FREE(a) (a)
2624#endif
2625
d1401ee9
MHM
2626#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2627#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2628#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2629
2630#ifndef PERL_CORE
a02a5408
JC
2631/* pre 5.9.x compatibility */
2632#define New(x,v,n,t) Newx(v,n,t)
2633#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2634#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2635#endif
a02a5408 2636
ff68c719 2637#define Renew(v,n,t) \
d1401ee9 2638 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2639#define Renewc(v,n,t,c) \
d1401ee9 2640 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2641
2642#ifdef PERL_POISON
2643#define Safefree(d) \
06c0cc96 2644 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2645#else
fe4f188c 2646#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2647#endif
55497cff 2648
dbb57106
YO
2649/* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2650 * as it handles cases like constant string arguments without throwing warnings *
2651 * the cast is required, as is the inequality check, to avoid warnings */
45908e4d 2652#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2653
45908e4d
YO
2654
2655#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)))
2656#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)))
2657#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2658
bdd1531d 2659/* Like above, but returns a pointer to 'd' */
45908e4d
YO
2660#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)))
2661#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 2662#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2663
7e337ee0
JH
2664#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2665#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2666#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2667#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2668
caa674f3
DD
2669#ifdef PERL_POISON
2670# define PERL_POISON_EXPR(x) x
2671#else
2672# define PERL_POISON_EXPR(x)
2673#endif
2674
ff68c719 2675#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2676
1b7e2294
KW
2677/*
2678=head1 Handy Values
2679
2680=for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2681
2682Returns the number of elements in the input C array (so you want your
2683zero-based indices to be less than but not equal to).
2684
2685=for apidoc Am|void *|C_ARRAY_END|void *a
2686
2687Returns a pointer to one element past the final element of the input C array.
2688
2689=cut
2690
2691C_ARRAY_END is one past the last: half-open/half-closed range, not
2692last-inclusive range.
2693*/
622913ab 2694#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2695#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2696
2cc61e15
DD
2697#ifdef NEED_VA_COPY
2698# ifdef va_copy
2699# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2700# elif defined(__va_copy)
2701# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2702# else
07798b17 2703# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2704# endif
2705#endif
2706
472d47bc
SB
2707/* convenience debug macros */
2708#ifdef USE_ITHREADS
2709#define pTHX_FORMAT "Perl interpreter: 0x%p"
2710#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2711#define pTHX_VALUE_ (void *)my_perl,
2712#define pTHX_VALUE (void *)my_perl
2713#define pTHX__VALUE_ ,(void *)my_perl,
2714#define pTHX__VALUE ,(void *)my_perl
472d47bc 2715#else
3609ea0d 2716#define pTHX_FORMAT
472d47bc 2717#define pTHX__FORMAT
3609ea0d 2718#define pTHX_VALUE_
472d47bc 2719#define pTHX_VALUE
3609ea0d 2720#define pTHX__VALUE_
472d47bc
SB
2721#define pTHX__VALUE
2722#endif /* USE_ITHREADS */
3609ea0d 2723
2acdbac1
NC
2724/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2725 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2726 CPAN::Unpack show any users outside the core. */
2727#ifdef PERL_CORE
dc6e8de0
A
2728# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2729 "Use of " s " is deprecated")
c9680906
A
2730# define deprecate_disappears_in(when,message) \
2731 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2732 message ", and will disappear in Perl " when)
ac641426
A
2733# define deprecate_fatal_in(when,message) \
2734 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2735 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2736#endif
2737
dfff4baf
BF
2738/* Internal macros to deal with gids and uids */
2739#ifdef PERL_CORE
2740
2741# if Uid_t_size > IVSIZE
2742# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2743# define SvUID(sv) SvNV(sv)
07798b17
AC
2744# elif Uid_t_sign <= 0
2745# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2746# define SvUID(sv) SvIV(sv)
dfff4baf 2747# else
07798b17
AC
2748# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2749# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2750# endif /* Uid_t_size */
2751
2752# if Gid_t_size > IVSIZE
2753# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2754# define SvGID(sv) SvNV(sv)
07798b17
AC
2755# elif Gid_t_sign <= 0
2756# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2757# define SvGID(sv) SvIV(sv)
dfff4baf 2758# else
07798b17
AC
2759# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2760# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2761# endif /* Gid_t_size */
2762
2763#endif
2764
6a5bc5ac 2765#endif /* PERL_HANDY_H_ */
9d745869 2766
e9a8c099 2767/*
14d04a33 2768 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2769 */