This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Use UINTMAX_C() in nBIT_MASK() definition
[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 275/* Returns a mask with the lowest n bits set */
fae1e72b 276#define nBIT_MASK(n) ((UINTMAX_C(1) << (n)) - 1)
d223e1ea 277
1381ccb1
KW
278/* The largest unsigned number that will fit into n bits */
279#define nBIT_UMAX(n) nBIT_MASK(n)
280
8d9433eb
KW
281/*
282=for apidoc Am|void|__ASSERT_|bool expr
283
284This is a helper macro to avoid preprocessor issues, replaced by nothing
285unless under DEBUGGING, where it expands to an assert of its argument,
286followed by a comma (hence the comma operator). If we just used a straight
287assert(), we would get a comma with nothing before it when not DEBUGGING.
288
289=cut
290
291We also use empty definition under Coverity since the __ASSERT__
292checks often check for things that Really Cannot Happen, and Coverity
293detects that and gets all excited. */
3e94db23 294
e7ae132e
KW
295#if defined(DEBUGGING) && !defined(__COVERITY__) \
296 && ! defined(PERL_SMALL_MACRO_BUFFER)
0f092d08
KW
297# define __ASSERT_(statement) assert(statement),
298#else
299# define __ASSERT_(statement)
300#endif
301
3fe05580 302/*
a4ee4fb5 303=head1 SV Manipulation Functions
3fe05580 304
3bb9fd01 305=for apidoc Ama|SV*|newSVpvs|"literal string"
1568d13a 306Like C<newSVpvn>, but takes a literal string instead of a
30a15352 307string/length pair.
3fe05580 308
3bb9fd01 309=for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
1568d13a 310Like C<newSVpvn_flags>, but takes a literal string instead of
30a15352 311a string/length pair.
84bafc02 312
3bb9fd01 313=for apidoc Ama|SV*|newSVpvs_share|"literal string"
1568d13a 314Like C<newSVpvn_share>, but takes a literal string instead of
30a15352 315a string/length pair and omits the hash parameter.
3fe05580 316
3bb9fd01 317=for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
1568d13a 318Like C<sv_catpvn_flags>, but takes a literal string instead
30a15352 319of a string/length pair.
9dcc53ea 320
3bb9fd01 321=for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
1568d13a 322Like C<sv_catpvn_nomg>, but takes a literal string instead of
0c395ea5 323a string/length pair.
9dcc53ea 324
3bb9fd01 325=for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
1568d13a 326Like C<sv_catpvn>, but takes a literal string instead of a
0c395ea5 327string/length pair.
3fe05580 328
3bb9fd01 329=for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
1568d13a 330Like C<sv_catpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
331string/length pair.
332
3bb9fd01 333=for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
1568d13a 334Like C<sv_setpvn>, but takes a literal string instead of a
0c395ea5 335string/length pair.
3fe05580 336
3bb9fd01 337=for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
1568d13a 338Like C<sv_setpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
339string/length pair.
340
3bb9fd01 341=for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
1568d13a 342Like C<sv_setref_pvn>, but takes a literal string instead of
0c395ea5 343a string/length pair.
9dcc53ea 344
3fe05580
MHM
345=head1 Memory Management
346
3bb9fd01 347=for apidoc Ama|char*|savepvs|"literal string"
1568d13a 348Like C<savepvn>, but takes a literal string instead of a
30a15352 349string/length pair.
3fe05580 350
3bb9fd01 351=for apidoc Ama|char*|savesharedpvs|"literal string"
9dcc53ea
Z
352A version of C<savepvs()> which allocates the duplicate string in memory
353which is shared between threads.
354
3fe05580
MHM
355=head1 GV Functions
356
3bb9fd01 357=for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
1568d13a 358Like C<gv_stashpvn>, but takes a literal string instead of a
0c395ea5 359string/length pair.
3fe05580
MHM
360
361=head1 Hash Manipulation Functions
362
3bb9fd01 363=for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
1568d13a 364Like C<hv_fetch>, but takes a literal string instead of a
0c395ea5 365string/length pair.
3fe05580 366
3bb9fd01 367=for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
1568d13a 368Like C<hv_store>, but takes a literal string instead of a
0c395ea5 369string/length pair
3fe05580
MHM
370and omits the hash parameter.
371
510966aa
Z
372=head1 Lexer interface
373
3bb9fd01 374=for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
510966aa 375
1568d13a 376Like L</lex_stuff_pvn>, but takes a literal string instead of
0c395ea5 377a string/length pair.
510966aa 378
3fe05580
MHM
379=cut
380*/
381
a34e53fc
KW
382/*
383=head1 Handy Values
2efa8cc7 384
a34e53fc
KW
385=for apidoc Amu|pair|STR_WITH_LEN|"literal string"
386
387Returns two comma separated tokens of the input literal string, and its length.
388This is convenience macro which helps out in some API calls.
389Note that it can't be used as an argument to macros or functions that under
390some configurations might be macros, which means that it requires the full
391Perl_xxx(aTHX_ ...) form for any API calls where it's used.
392
393=cut
394*/
395
a34e53fc 396#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
ba3a79e7
GA
397
398/* STR_WITH_LEN() shortcuts */
399#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
400#define newSVpvs_flags(str,flags) \
401 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 402#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
403#define sv_catpvs_flags(sv, str, flags) \
404 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
405#define sv_catpvs_nomg(sv, str) \
406 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
407#define sv_catpvs(sv, str) \
408 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
409#define sv_catpvs_mg(sv, str) \
410 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 411#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
412#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
413#define sv_setref_pvs(rv, classname, str) \
414 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 415#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
416#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
417#define gv_stashpvs(str, create) \
418 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
419#define gv_fetchpvs(namebeg, add, sv_type) \
420 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
421#define gv_fetchpvn(namebeg, len, add, sv_type) \
422 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
423#define sv_catxmlpvs(dsv, str, utf8) \
424 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 425
ba3a79e7 426
510966aa
Z
427#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
428
b96d8cd9
NC
429#define get_cvs(str, flags) \
430 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 431
9b6e9510
KW
432/* internal helpers */
433#define PERL_RVS_TO_DECIMAL_(r,v,s) (((r)*1000000)+((v)*1000)+(s))
434#define PERL_DECIMAL_VERSION_ \
435 PERL_RVS_TO_DECIMAL_(PERL_REVISION, PERL_VERSION, PERL_SUBVERSION)
436
437/*
438=for apidoc AmR|bool|PERL_VERSION_EQ|const int r|const int v|const int s
439
440Returns whether or not the perl currently executing has the specified
441relationship to the perl given by the parameters. For example,
442
443 #if PERL_VERSION_GT(5,24,2)
444 code that will only be compiled on perls after v5.24.2
445 #else
446 fallback code
447 #endif
448
449Note that this is usable in making compile-time decisions
450
451The possible comparisons are C<PERL_VERSION_EQ>, C<PERL_VERSION_NE>,
452C<PERL_VERSION_GE>, C<PERL_VERSION_GT>, C<PERL_VERSION_LE>, and
453C<PERL_VERSION_LT>.
454
455=for apidoc AmRh|bool|PERL_VERSION_NE|const int r|const int v|const int s
456=for apidoc AmRh|bool|PERL_VERSION_GE|const int r|const int v|const int s
457=for apidoc AmRh|bool|PERL_VERSION_GT|const int r|const int v|const int s
458=for apidoc AmRh|bool|PERL_VERSION_LE|const int r|const int v|const int s
459=for apidoc AmRh|bool|PERL_VERSION_LT|const int r|const int v|const int s
460
461=cut
462*/
463
464# define PERL_VERSION_EQ(r,v,s) (PERL_DECIMAL_VERSION_ == PERL_RVS_TO_DECIMAL_(r,v,s))
465# define PERL_VERSION_NE(r,v,s) (! PERL_VERSION_EQ(r,v,s))
466# define PERL_VERSION_LT(r,v,s) (PERL_DECIMAL_VERSION_ < PERL_RVS_TO_DECIMAL_(r,v,s))
467# define PERL_VERSION_LE(r,v,s) (PERL_DECIMAL_VERSION_ <= PERL_RVS_TO_DECIMAL_(r,v,s))
468# define PERL_VERSION_GT(r,v,s) (! PERL_VERSION_LE(r,v,s))
469# define PERL_VERSION_GE(r,v,s) (! PERL_VERSION_LT(r,v,s))
470
954c1994 471/*
ccfc67b7
JH
472=head1 Miscellaneous Functions
473
954c1994 474=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
475Test two C<NUL>-terminated strings to see if they are different. Returns true
476or false.
954c1994
GS
477
478=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
479Test two C<NUL>-terminated strings to see if they are equal. Returns true or
480false.
954c1994
GS
481
482=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
483Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
484second, C<s2>. Returns true or false.
954c1994
GS
485
486=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
487Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
488equal to the second, C<s2>. Returns true or false.
954c1994
GS
489
490=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
491Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
492the second, C<s2>. Returns true or false.
954c1994
GS
493
494=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
495Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
496or equal to the second, C<s2>. Returns true or false.
954c1994
GS
497
498=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
499Test two C<NUL>-terminated strings to see if they are different. The C<len>
500parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
501wrapper for C<strncmp>).
502
503=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
504Test two C<NUL>-terminated strings to see if they are equal. The C<len>
505parameter indicates the number of bytes to compare. Returns true or false. (A
506wrapper for C<strncmp>).
954c1994 507
bd18bd40
KW
508=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
509Test two buffers (which may contain embedded C<NUL> characters, to see if they
510are equal. The C<len> parameter indicates the number of bytes to compare.
511Returns zero if equal, or non-zero if non-equal.
512
3bb9fd01 513=for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
514Like L</memEQ>, but the second string is a literal enclosed in double quotes,
515C<l1> gives the number of bytes in C<s1>.
516Returns zero if equal, or non-zero if non-equal.
517
bd18bd40
KW
518=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
519Test two buffers (which may contain embedded C<NUL> characters, to see if they
520are not equal. The C<len> parameter indicates the number of bytes to compare.
521Returns zero if non-equal, or non-zero if equal.
522
3bb9fd01 523=for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
2d8eeddb
KW
524Like L</memNE>, but the second string is a literal enclosed in double quotes,
525C<l1> gives the number of bytes in C<s1>.
526Returns zero if non-equal, or zero if non-equal.
527
4aada8b9
KW
528=for apidoc Am|bool|memCHRs|"list"|char c
529Returns the position of the first occurence of the byte C<c> in the literal
530string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
531treated as unsigned char. Thus this macro can be used to determine if C<c> is
532in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
533is C<NUL> (and the set doesn't include C<NUL>).
534
954c1994 535=cut
fc169e00
KW
536
537New macros should use the following conventions for their names (which are
538based on the underlying C library functions):
539
540 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
541
542 Each has two main parameters, string-like operands that are compared
543 against each other, as specified by the macro name. Some macros may
544 additionally have one or potentially even two length parameters. If a length
545 parameter applies to both string parameters, it will be positioned third;
546 otherwise any length parameter immediately follows the string parameter it
547 applies to.
548
549 If the prefix to the name is 'str', the string parameter is a pointer to a C
550 language string. Such a string does not contain embedded NUL bytes; its
551 length may be unknown, but can be calculated by C<strlen()>, since it is
552 terminated by a NUL, which isn't included in its length.
553
a3815e44 554 The optional 'n' following 'str' means that there is a third parameter,
fc169e00
KW
555 giving the maximum number of bytes to look at in each string. Even if both
556 strings are longer than the length parameter, those extra bytes will be
557 unexamined.
558
559 The 's' suffix means that the 2nd byte string parameter is a literal C
560 double-quoted string. Its length will automatically be calculated by the
561 macro, so no length parameter will ever be needed for it.
562
563 If the prefix is 'mem', the string parameters don't have to be C strings;
564 they may contain embedded NUL bytes, do not necessarily have a terminating
565 NUL, and their lengths can be known only through other means, which in
566 practice are additional parameter(s) passed to the function. All 'mem'
567 functions have at least one length parameter. Barring any 'l' or 's' suffix,
568 there is a single length parameter, in position 3, which applies to both
569 string parameters. The 's' suffix means, as described above, that the 2nd
570 string is a literal double-quoted C string (hence its length is calculated by
571 the macro, and the length parameter to the function applies just to the first
572 string parameter, and hence is positioned just after it). An 'l' suffix
573 means that the 2nd string parameter has its own length parameter, and the
574 signature will look like memFOOl(s1, l1, s2, l2).
575
576 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
577 substring of the 1st string. 'P' if present indicates that the substring
578 must be a "proper" one in tha mathematical sense that the first one must be
579 strictly larger than the 2nd.
580
954c1994
GS
581*/
582
62946e08 583
75400963
KW
584#define strNE(s1,s2) (strcmp(s1,s2) != 0)
585#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
586#define strLT(s1,s2) (strcmp(s1,s2) < 0)
587#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
588#define strGT(s1,s2) (strcmp(s1,s2) > 0)
589#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 590
75400963
KW
591#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
592#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 593
9d3980bc
KW
594#define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
595#define memNE(s1,s2,l) (! memEQ(s1,s2,l))
36477c24 596
085b7534 597/* memEQ and memNE where second comparand is a string constant */
568a785a 598#define memEQs(s1, l, s2) \
777fa2cb 599 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 600#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 601
fdbb9a7c
KW
602/* Keep these private until we decide it was a good idea */
603#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
604
605#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
606
bdb7e3f0 607#define memBEGINs(s1, l, s2) \
30a6480c 608 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 609 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158 610#define memBEGINPs(s1, l, s2) \
30a6480c 611 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
de627158 612 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0 613#define memENDs(s1, l, s2) \
30a6480c 614 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 615 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424 616#define memENDPs(s1, l, s2) \
30a6480c 617 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
b80f8424 618 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 619#endif /* End of making macros private */
bdb7e3f0 620
062b6850
KW
621#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
622#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
623#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
624#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
625
4aada8b9
KW
626#define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
627
bbce6d69
PP
628/*
629 * Character classes.
630 *
631 * Unfortunately, the introduction of locales means that we
632 * can't trust isupper(), etc. to tell the truth. And when
633 * it comes to /\w+/ with tainting enabled, we *must* be able
634 * to trust our character classes.
635 *
636 * Therefore, the default tests in the text of Perl will be
637 * independent of locale. Any code that wants to depend on
638 * the current locale will use the tests that begin with "lc".
639 */
640
2304df62
AD
641#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
642# ifndef CTYPE256
643# define CTYPE256
644# endif
645#endif
646
954c1994 647/*
ccfc67b7 648
dcccc8ff 649=head1 Character classification
243effed
KW
650This section is about functions (really macros) that classify characters
651into types, such as punctuation versus alphabetic, etc. Most of these are
652analogous to regular expression character classes. (See
653L<perlrecharclass/POSIX Character Classes>.) There are several variants for
654each class. (Not all macros have all variants; each item below lists the
655ones valid for it.) None are affected by C<use bytes>, and only the ones
656with C<LC> in the name are affected by the current locale.
657
d713f9d9
KW
658The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
659treating it as a code point, and returns a boolean as to whether or not the
660character represented by it is (or on non-ASCII platforms, corresponds to) an
6aff1f14
KW
661ASCII character in the named class based on platform, Unicode, and Perl rules.
662If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 663
c98722a4 664Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
665with no suffix C<"_A">. This variant is used to emphasize by its name that
666only ASCII-range characters can return TRUE.
4b9734bf 667
d60679e1 668Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
669onto the platform. That is, the code points that are ASCII are unaffected,
670since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
671as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
672true when called with the code point 0xDF, which is a word character in both
4650c663 673ASCII and EBCDIC (though it represents different characters in each).
d713f9d9
KW
674If the input is a number that doesn't fit in an octet, FALSE is returned.
675(Perl's documentation uses a colloquial definition of Latin-1, to include all
676code points below 256.)
243effed 677
d713f9d9
KW
678Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
679inputs below 256, but if the code point is larger than 255, Unicode rules are
680used to determine if it is in the character class. For example,
d0da05db 681C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 682WITH MACRON in Unicode, and is a word character.
243effed 683
059703b0
KW
684Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
685but are used for UTF-8 encoded strings. The two forms are different names for
686the same thing. Each call to one of these classifies the first character of
687the string starting at C<p>. The second parameter, C<e>, points to anywhere in
688the string beyond the first character, up to one byte past the end of the
689entire string. Although both variants are identical, the suffix C<_safe> in
690one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
691provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
692C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
693some way, the program may croak, or the function may return FALSE, at the
694discretion of the implementation, and subject to change in future releases.
243effed 695
d713f9d9
KW
696Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
697but the result is based on the current locale, which is what C<LC> in the name
698stands for. If Perl can determine that the current locale is a UTF-8 locale,
699it uses the published Unicode rules; otherwise, it uses the C library function
700that gives the named classification. For example, C<isDIGIT_LC()> when not in
701a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
702returned if the input won't fit into an octet. On some platforms where the C
703library function is known to be defective, Perl changes its result to follow
704the POSIX standard's rules.
243effed 705
d713f9d9
KW
706Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
707than 256, but for larger ones it returns the Unicode classification of the code
708point.
243effed 709
059703b0
KW
710Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
711C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
712are different names for the same thing. Each call to one of these classifies
713the first character of the string starting at C<p>. The second parameter,
714C<e>, points to anywhere in the string beyond the first character, up to one
715byte past the end of the entire string. Although both variants are identical,
716the suffix C<_safe> in one name emphasizes that it will not attempt to read
717beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
718is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
719character is malformed in some way, the program may croak, or the function may
720return FALSE, at the discretion of the implementation, and subject to change in
721future releases.
ccfc67b7 722
d713f9d9
KW
723=for apidoc Am|bool|isALPHA|int ch
724Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
725analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
726See the L<top of this section|/Character classification> for an explanation of
727variants
059703b0
KW
728C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>,
729C<isALPHA_utf8_safe>, C<isALPHA_LC>, C<isALPHA_LC_uvchr>, C<isALPHA_LC_utf8>,
730and C<isALPHA_LC_utf8_safe>.
8a58bdcf 731
f16858ed
KW
732=cut
733
734Here and below, we add the protoypes of these macros for downstream programs
735that would be interested in them, such as Devel::PPPort
736
737=for apidoc Amh|bool|isALPHA_A|int ch
738=for apidoc Amh|bool|isALPHA_L1|int ch
739=for apidoc Amh|bool|isALPHA_uvchr|int ch
740=for apidoc Amh|bool|isALPHA_utf8_safe|U8 * s|U8 * end
d23c7e08 741=for apidoc Amh|bool|isALPHA_utf8|U8 * s|U8 * end
f16858ed
KW
742=for apidoc Amh|bool|isALPHA_LC|int ch
743=for apidoc Amh|bool|isALPHA_LC_uvchr|int ch
744=for apidoc Amh|bool|isALPHA_LC_utf8_safe|U8 * s| U8 *end
745
d713f9d9
KW
746=for apidoc Am|bool|isALPHANUMERIC|int ch
747Returns a boolean indicating whether the specified character is one of
748C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
749See the L<top of this section|/Character classification> for an explanation of
750variants
d0da05db 751C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
059703b0
KW
752C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>,
753C<isALPHANUMERIC_LC_uvchr>, C<isALPHANUMERIC_LC_utf8>, and
754C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 755
255b632a
KW
756A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
757this corresponds to the C language alphanumeric definition). Also
758there are the variants
759C<isALNUMC_A>, C<isALNUMC_L1>
760C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
761
f16858ed
KW
762=for apidoc Amh|bool|isALPHANUMERIC_A|int ch
763=for apidoc Amh|bool|isALPHANUMERIC_L1|int ch
764=for apidoc Amh|bool|isALPHANUMERIC_uvchr|int ch
765=for apidoc Amh|bool|isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
d23c7e08 766=for apidoc Amh|bool|isALPHANUMERIC_utf8|U8 * s|U8 * end
f16858ed
KW
767=for apidoc Amh|bool|isALPHANUMERIC_LC|int ch
768=for apidoc Amh|bool|isALPHANUMERIC_LC_uvchr|int ch
769=for apidoc Amh|bool|isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
770=for apidoc Amh|bool|isALNUMC|int ch
771=for apidoc Amh|bool|isALNUMC_A|int ch
772=for apidoc Amh|bool|isALNUMC_L1|int ch
773=for apidoc Amh|bool|isALNUMC_LC|int ch
774=for apidoc Amh|bool|isALNUMC_LC_uvchr|int ch
775
d713f9d9 776=for apidoc Am|bool|isASCII|int ch
8a58bdcf 777Returns a boolean indicating whether the specified character is one of the 128
243effed 778characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 779On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
780character corresponds to an ASCII character. Variants C<isASCII_A()> and
781C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
782See the L<top of this section|/Character classification> for an explanation of
783variants
059703b0
KW
784C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_utf8_safe>, C<isASCII_LC>,
785C<isASCII_LC_uvchr>, C<isASCII_LC_utf8>, and C<isASCII_LC_utf8_safe>.
786Note, however, that some platforms do not have the C library routine
787C<isascii()>. In these cases, the variants whose names contain C<LC> are the
788same as the corresponding ones without.
243effed 789
f16858ed
KW
790=for apidoc Amh|bool|isASCII_A|int ch
791=for apidoc Amh|bool|isASCII_L1|int ch
792=for apidoc Amh|bool|isASCII_uvchr|int ch
793=for apidoc Amh|bool|isASCII_utf8_safe|U8 * s|U8 * end
d23c7e08 794=for apidoc Amh|bool|isASCII_utf8|U8 * s|U8 * end
f16858ed
KW
795=for apidoc Amh|bool|isASCII_LC|int ch
796=for apidoc Amh|bool|isASCII_LC_uvchr|int ch
797=for apidoc Amh|bool|isASCII_LC_utf8_safe|U8 * s| U8 *end
798
d98532ea
KW
799Also note, that because all ASCII characters are UTF-8 invariant (meaning they
800have the exact same representation (always a single byte) whether encoded in
801UTF-8 or not), C<isASCII> will give the correct results when called with any
059703b0
KW
802byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
803C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
d98532ea 804
243effed
KW
805=for apidoc Am|bool|isBLANK|char ch
806Returns a boolean indicating whether the specified character is a
6aff1f14 807character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
808See the L<top of this section|/Character classification> for an explanation of
809variants
059703b0
KW
810C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>,
811C<isBLANK_utf8_safe>, C<isBLANK_LC>, C<isBLANK_LC_uvchr>, C<isBLANK_LC_utf8>,
812and C<isBLANK_LC_utf8_safe>. Note,
da8c1a98
KW
813however, that some platforms do not have the C library routine
814C<isblank()>. In these cases, the variants whose names contain C<LC> are
815the same as the corresponding ones without.
243effed 816
f16858ed
KW
817=for apidoc Amh|bool|isBLANK_A|int ch
818=for apidoc Amh|bool|isBLANK_L1|int ch
819=for apidoc Amh|bool|isBLANK_uvchr|int ch
820=for apidoc Amh|bool|isBLANK_utf8_safe|U8 * s|U8 * end
d23c7e08 821=for apidoc Amh|bool|isBLANK_utf8|U8 * s|U8 * end
f16858ed
KW
822=for apidoc Amh|bool|isBLANK_LC|int ch
823=for apidoc Amh|bool|isBLANK_LC_uvchr|int ch
824=for apidoc Amh|bool|isBLANK_LC_utf8_safe|U8 * s| U8 *end
825
243effed
KW
826=for apidoc Am|bool|isCNTRL|char ch
827Returns a boolean indicating whether the specified character is a
6aff1f14 828control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
829See the L<top of this section|/Character classification> for an explanation of
830variants
059703b0
KW
831C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>,
832C<isCNTRL_utf8_safe>, C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, C<isCNTRL_LC_utf8>
833and C<isCNTRL_LC_utf8_safe>. On EBCDIC
da8c1a98 834platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 835
f16858ed
KW
836=for apidoc Amh|bool|isCNTRL_A|int ch
837=for apidoc Amh|bool|isCNTRL_L1|int ch
838=for apidoc Amh|bool|isCNTRL_uvchr|int ch
839=for apidoc Amh|bool|isCNTRL_utf8_safe|U8 * s|U8 * end
d23c7e08 840=for apidoc Amh|bool|isCNTRL_utf8|U8 * s|U8 * end
f16858ed
KW
841=for apidoc Amh|bool|isCNTRL_LC|int ch
842=for apidoc Amh|bool|isCNTRL_LC_uvchr|int ch
843=for apidoc Amh|bool|isCNTRL_LC_utf8_safe|U8 * s| U8 *end
844
954c1994 845=for apidoc Am|bool|isDIGIT|char ch
2787a470 846Returns a boolean indicating whether the specified character is a
6aff1f14 847digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 848Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
849See the L<top of this section|/Character classification> for an explanation of
850variants
059703b0
KW
851C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>,
852C<isDIGIT_LC_uvchr>, C<isDIGIT_LC_utf8>, and C<isDIGIT_LC_utf8_safe>.
243effed 853
f16858ed
KW
854=for apidoc Amh|bool|isDIGIT_A|int ch
855=for apidoc Amh|bool|isDIGIT_L1|int ch
856=for apidoc Amh|bool|isDIGIT_uvchr|int ch
857=for apidoc Amh|bool|isDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 858=for apidoc Amh|bool|isDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
859=for apidoc Amh|bool|isDIGIT_LC|int ch
860=for apidoc Amh|bool|isDIGIT_LC_uvchr|int ch
861=for apidoc Amh|bool|isDIGIT_LC_utf8_safe|U8 * s| U8 *end
862
243effed
KW
863=for apidoc Am|bool|isGRAPH|char ch
864Returns a boolean indicating whether the specified character is a
6aff1f14 865graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 866See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
867variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>,
868C<isGRAPH_utf8_safe>, C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>,
869C<isGRAPH_LC_utf8_safe>, and C<isGRAPH_LC_utf8_safe>.
954c1994 870
f16858ed
KW
871=for apidoc Amh|bool|isGRAPH_A|int ch
872=for apidoc Amh|bool|isGRAPH_L1|int ch
873=for apidoc Amh|bool|isGRAPH_uvchr|int ch
874=for apidoc Amh|bool|isGRAPH_utf8_safe|U8 * s|U8 * end
d23c7e08 875=for apidoc Amh|bool|isGRAPH_utf8|U8 * s|U8 * end
f16858ed
KW
876=for apidoc Amh|bool|isGRAPH_LC|int ch
877=for apidoc Amh|bool|isGRAPH_LC_uvchr|int ch
878=for apidoc Amh|bool|isGRAPH_LC_utf8_safe|U8 * s| U8 *end
879
0c82b6df 880=for apidoc Am|bool|isLOWER|char ch
2787a470 881Returns a boolean indicating whether the specified character is a
6aff1f14 882lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
883See the L<top of this section|/Character classification> for an explanation of
884variants
059703b0
KW
885C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>,
886C<isLOWER_utf8_safe>, C<isLOWER_LC>, C<isLOWER_LC_uvchr>, C<isLOWER_LC_utf8>,
887and C<isLOWER_LC_utf8_safe>.
0c82b6df 888
f16858ed
KW
889=for apidoc Amh|bool|isLOWER_A|int ch
890=for apidoc Amh|bool|isLOWER_L1|int ch
891=for apidoc Amh|bool|isLOWER_uvchr|int ch
892=for apidoc Amh|bool|isLOWER_utf8_safe|U8 * s|U8 * end
d23c7e08 893=for apidoc Amh|bool|isLOWER_utf8|U8 * s|U8 * end
f16858ed
KW
894=for apidoc Amh|bool|isLOWER_LC|int ch
895=for apidoc Amh|bool|isLOWER_LC_uvchr|int ch
896=for apidoc Amh|bool|isLOWER_LC_utf8_safe|U8 * s| U8 *end
897
c99e91e9 898=for apidoc Am|bool|isOCTAL|char ch
2787a470 899Returns a boolean indicating whether the specified character is an
6aff1f14 900octal digit, [0-7].
243effed
KW
901The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
902C<isOCTAL>.
903
f16858ed
KW
904=for apidoc Amh|bool|isOCTAL_A|int ch
905=for apidoc Amh|bool|isOCTAL_L1|int ch
906
243effed
KW
907=for apidoc Am|bool|isPUNCT|char ch
908Returns a boolean indicating whether the specified character is a
6aff1f14
KW
909punctuation character, analogous to C<m/[[:punct:]]/>.
910Note that the definition of what is punctuation isn't as
243effed
KW
911straightforward as one might desire. See L<perlrecharclass/POSIX Character
912Classes> for details.
dcccc8ff 913See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
914variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>,
915C<isPUNCT_utf8_safe>, C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, C<isPUNCT_LC_utf8>,
916and C<isPUNCT_LC_utf8_safe>.
c99e91e9 917
f16858ed
KW
918=for apidoc Amh|bool|isPUNCT_A|int ch
919=for apidoc Amh|bool|isPUNCT_L1|int ch
920=for apidoc Amh|bool|isPUNCT_uvchr|int ch
921=for apidoc Amh|bool|isPUNCT_utf8_safe|U8 * s|U8 * end
d23c7e08 922=for apidoc Amh|bool|isPUNCT_utf8|U8 * s|U8 * end
f16858ed
KW
923=for apidoc Amh|bool|isPUNCT_LC|int ch
924=for apidoc Amh|bool|isPUNCT_LC_uvchr|int ch
925=for apidoc Amh|bool|isPUNCT_LC_utf8_safe|U8 * s| U8 *end
926
0c82b6df 927=for apidoc Am|bool|isSPACE|char ch
2787a470 928Returns a boolean indicating whether the specified character is a
6aff1f14 929whitespace character. This is analogous
398d098a 930to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 931this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
932locale forms of this macro (the ones with C<LC> in their names) matched
933precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
934in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
935(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
936See the L<top of this section|/Character classification> for an explanation of
937variants
059703b0
KW
938C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>,
939C<isSPACE_utf8_safe>, C<isSPACE_LC>, C<isSPACE_LC_uvchr>, C<isSPACE_LC_utf8>,
940and C<isSPACE_LC_utf8_safe>.
0c82b6df 941
f16858ed
KW
942=for apidoc Amh|bool|isSPACE_A|int ch
943=for apidoc Amh|bool|isSPACE_L1|int ch
944=for apidoc Amh|bool|isSPACE_uvchr|int ch
945=for apidoc Amh|bool|isSPACE_utf8_safe|U8 * s|U8 * end
d23c7e08 946=for apidoc Amh|bool|isSPACE_utf8|U8 * s|U8 * end
f16858ed
KW
947=for apidoc Amh|bool|isSPACE_LC|int ch
948=for apidoc Amh|bool|isSPACE_LC_uvchr|int ch
949=for apidoc Amh|bool|isSPACE_LC_utf8_safe|U8 * s| U8 *end
950
398d098a
KW
951=for apidoc Am|bool|isPSXSPC|char ch
952(short for Posix Space)
779cf272
KW
953Starting in 5.18, this is identical in all its forms to the
954corresponding C<isSPACE()> macros.
398d098a
KW
955The locale forms of this macro are identical to their corresponding
956C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
957non-locale forms differ from their C<isSPACE()> forms only in that the
958C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
959Otherwise they are identical. Thus this macro is analogous to what
960C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 961See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
962variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
963C<isPSXSPC_utf8_safe>, C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>,
964C<isPSXSPC_LC_utf8>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 965
f16858ed
KW
966=for apidoc Amh|bool|isPSXSPC_A|int ch
967=for apidoc Amh|bool|isPSXSPC_L1|int ch
968=for apidoc Amh|bool|isPSXSPC_uvchr|int ch
969=for apidoc Amh|bool|isPSXSPC_utf8_safe|U8 * s|U8 * end
d23c7e08 970=for apidoc Amh|bool|isPSXSPC_utf8|U8 * s|U8 * end
f16858ed
KW
971=for apidoc Amh|bool|isPSXSPC_LC|int ch
972=for apidoc Amh|bool|isPSXSPC_LC_uvchr|int ch
973=for apidoc Amh|bool|isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
974
954c1994 975=for apidoc Am|bool|isUPPER|char ch
2787a470 976Returns a boolean indicating whether the specified character is an
6aff1f14 977uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 978See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
979variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>,
980C<isUPPER_utf8_safe>, C<isUPPER_LC>, C<isUPPER_LC_uvchr>, C<isUPPER_LC_utf8>,
981and C<isUPPER_LC_utf8_safe>.
954c1994 982
f16858ed
KW
983=for apidoc Amh|bool|isUPPER_A|int ch
984=for apidoc Amh|bool|isUPPER_L1|int ch
985=for apidoc Amh|bool|isUPPER_uvchr|int ch
986=for apidoc Amh|bool|isUPPER_utf8_safe|U8 * s|U8 * end
d23c7e08 987=for apidoc Amh|bool|isUPPER_utf8|U8 * s|U8 * end
f16858ed
KW
988=for apidoc Amh|bool|isUPPER_LC|int ch
989=for apidoc Amh|bool|isUPPER_LC_uvchr|int ch
990=for apidoc Amh|bool|isUPPER_LC_utf8_safe|U8 * s| U8 *end
991
243effed 992=for apidoc Am|bool|isPRINT|char ch
8eea39dd 993Returns a boolean indicating whether the specified character is a
6aff1f14 994printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
995See the L<top of this section|/Character classification> for an explanation of
996variants
059703b0
KW
997C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>,
998C<isPRINT_utf8_safe>, C<isPRINT_LC>, C<isPRINT_LC_uvchr>, C<isPRINT_LC_utf8>,
999and C<isPRINT_LC_utf8_safe>.
243effed 1000
f16858ed
KW
1001=for apidoc Amh|bool|isPRINT_A|int ch
1002=for apidoc Amh|bool|isPRINT_L1|int ch
1003=for apidoc Amh|bool|isPRINT_uvchr|int ch
1004=for apidoc Amh|bool|isPRINT_utf8_safe|U8 * s|U8 * end
d23c7e08 1005=for apidoc Amh|bool|isPRINT_utf8|U8 * s|U8 * end
f16858ed
KW
1006=for apidoc Amh|bool|isPRINT_LC|int ch
1007=for apidoc Amh|bool|isPRINT_LC_uvchr|int ch
1008=for apidoc Amh|bool|isPRINT_LC_utf8_safe|U8 * s| U8 *end
1009
243effed
KW
1010=for apidoc Am|bool|isWORDCHAR|char ch
1011Returns a boolean indicating whether the specified character is a character
1012that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1013in a regular expression. A word character is an alphabetic character, a
1014decimal digit, a connecting punctuation character (such as an underscore), or
1015a "mark" character that attaches to one of those (like some sort of accent).
1016C<isALNUM()> is a synonym provided for backward compatibility, even though a
1017word character includes more than the standard C language meaning of
1018alphanumeric.
dcccc8ff 1019See the L<top of this section|/Character classification> for an explanation of
059703b0
KW
1020variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1021C<isWORDCHAR_utf8>, and C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>,
1022C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and C<isWORDCHAR_LC_utf8_safe>
1023are also as described there, but additionally include the platform's native
1024underscore.
8a58bdcf 1025
f16858ed
KW
1026=for apidoc Amh|bool|isWORDCHAR_A|int ch
1027=for apidoc Amh|bool|isWORDCHAR_L1|int ch
1028=for apidoc Amh|bool|isWORDCHAR_uvchr|int ch
1029=for apidoc Amh|bool|isWORDCHAR_utf8_safe|U8 * s|U8 * end
d23c7e08 1030=for apidoc Amh|bool|isWORDCHAR_utf8|U8 * s|U8 * end
f16858ed
KW
1031=for apidoc Amh|bool|isWORDCHAR_LC|int ch
1032=for apidoc Amh|bool|isWORDCHAR_LC_uvchr|int ch
1033=for apidoc Amh|bool|isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
1034=for apidoc Amh|bool|isALNUM|int ch
1035=for apidoc Amh|bool|isALNUM_A|int ch
1036=for apidoc Amh|bool|isALNUM_LC|int ch
1037=for apidoc Amh|bool|isALNUM_LC_uvchr|int ch
1038
8a58bdcf
KW
1039=for apidoc Am|bool|isXDIGIT|char ch
1040Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
1041digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1042and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
1043See the L<top of this section|/Character classification> for an explanation of
1044variants
059703b0
KW
1045C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>,
1046C<isXDIGIT_LC_uvchr>, C<isXDIGIT_LC_utf8>, and C<isXDIGIT_LC_utf8_safe>.
243effed 1047
f16858ed
KW
1048=for apidoc Amh|bool|isXDIGIT_A|int ch
1049=for apidoc Amh|bool|isXDIGIT_L1|int ch
1050=for apidoc Amh|bool|isXDIGIT_uvchr|int ch
1051=for apidoc Amh|bool|isXDIGIT_utf8_safe|U8 * s|U8 * end
d23c7e08 1052=for apidoc Amh|bool|isXDIGIT_utf8|U8 * s|U8 * end
f16858ed
KW
1053=for apidoc Amh|bool|isXDIGIT_LC|int ch
1054=for apidoc Amh|bool|isXDIGIT_LC_uvchr|int ch
1055=for apidoc Amh|bool|isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1056
3c3ecf18
KW
1057=for apidoc Am|bool|isIDFIRST|char ch
1058Returns a boolean indicating whether the specified character can be the first
1059character of an identifier. This is very close to, but not quite the same as
1060the official Unicode property C<XID_Start>. The difference is that this
1061returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
1062See the L<top of this section|/Character classification> for an explanation of
1063variants
059703b0
KW
1064C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
1065C<isIDFIRST_utf8_safe>, C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>,
1066C<isIDFIRST_LC_utf8>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18 1067
f16858ed
KW
1068=for apidoc Amh|bool|isIDFIRST_A|int ch
1069=for apidoc Amh|bool|isIDFIRST_L1|int ch
1070=for apidoc Amh|bool|isIDFIRST_uvchr|int ch
1071=for apidoc Amh|bool|isIDFIRST_utf8_safe|U8 * s|U8 * end
d23c7e08 1072=for apidoc Amh|bool|isIDFIRST_utf8|U8 * s|U8 * end
f16858ed
KW
1073=for apidoc Amh|bool|isIDFIRST_LC|int ch
1074=for apidoc Amh|bool|isIDFIRST_LC_uvchr|int ch
1075=for apidoc Amh|bool|isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1076
3c3ecf18
KW
1077=for apidoc Am|bool|isIDCONT|char ch
1078Returns a boolean indicating whether the specified character can be the
1079second or succeeding character of an identifier. This is very close to, but
1080not quite the same as the official Unicode property C<XID_Continue>. The
1081difference is that this returns true only if the input character also matches
dcccc8ff 1082L</isWORDCHAR>. See the L<top of this section|/Character classification> for
059703b0
KW
1083an explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
1084C<isIDCONT_utf8>, C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>,
1085C<isIDCONT_LC_utf8>, and C<isIDCONT_LC_utf8_safe>.
3c3ecf18 1086
f16858ed
KW
1087=for apidoc Amh|bool|isIDCONT_A|int ch
1088=for apidoc Amh|bool|isIDCONT_L1|int ch
1089=for apidoc Amh|bool|isIDCONT_uvchr|int ch
1090=for apidoc Amh|bool|isIDCONT_utf8_safe|U8 * s|U8 * end
d23c7e08 1091=for apidoc Amh|bool|isIDCONT_utf8|U8 * s|U8 * end
f16858ed
KW
1092=for apidoc Amh|bool|isIDCONT_LC|int ch
1093=for apidoc Amh|bool|isIDCONT_LC_uvchr|int ch
1094=for apidoc Amh|bool|isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1095
243effed 1096=head1 Miscellaneous Functions
8eea39dd 1097
95a59cab 1098=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 1099Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
1100Behaviour is only well defined when isXDIGIT(*str) is true.
1101
e7c1e6c1 1102=head1 Character case changing
21da7284
KW
1103Perl uses "full" Unicode case mappings. This means that converting a single
1104character to another case may result in a sequence of more than one character.
1105For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1106character sequence C<SS>. This presents some complications The lowercase of
1107all characters in the range 0..255 is a single character, and thus
1108C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1109return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1110an API that does allow every possible legal result to be returned.) Likewise
1111no other function that is crippled by not being able to give the correct
1112results for the full range of possible inputs has been implemented here.
e7c1e6c1 1113
d713f9d9 1114=for apidoc Am|U8|toUPPER|int ch
1f607577
KW
1115Converts the specified character to uppercase. If the input is anything but an
1116ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 1117C<toUPPER_A> is equivalent.
954c1994 1118
d0da05db
KW
1119=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1120Converts the code point C<cp> to its uppercase version, and
1121stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1122point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1123that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1124bytes since the uppercase version may be longer than the original character.
1125
1126The first code point of the uppercased version is returned
21da7284
KW
1127(but note, as explained at L<the top of this section|/Character case
1128changing>, that there may be more.)
1f607577 1129
059703b0 1130=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1131Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1132extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
1133stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1134that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1135bytes since the uppercase version may be longer than the original character.
1136
1137The first code point of the uppercased version is returned
21da7284
KW
1138(but note, as explained at L<the top of this section|/Character case
1139changing>, that there may be more).
1f607577 1140
059703b0
KW
1141It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1142S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1143the UTF-8 for the input character is malformed in some way, the program may
1144croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1145of the implementation, and subject to change in future releases.
a239b1e2 1146
059703b0
KW
1147=for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1148Same as L</toUPPER_utf8>.
1f607577 1149
25200305
KW
1150=for apidoc Am|U8|toFOLD|U8 ch
1151Converts the specified character to foldcase. If the input is anything but an
1152ASCII uppercase character, that input character itself is returned. Variant
1153C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 1154Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 1155
d0da05db
KW
1156=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1157Converts the code point C<cp> to its foldcase version, and
1158stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1159point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1160that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1161bytes since the foldcase version may be longer than the original character.
1162
1163The first code point of the foldcased version is returned
21da7284
KW
1164(but note, as explained at L<the top of this section|/Character case
1165changing>, that there may be more).
1f607577 1166
059703b0 1167=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1168Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1169extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
1170stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1171that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1172bytes since the foldcase version may be longer than the original character.
1173
1174The first code point of the foldcased version is returned
21da7284
KW
1175(but note, as explained at L<the top of this section|/Character case
1176changing>, that there may be more).
1f607577 1177
059703b0 1178It will not attempt
a239b1e2
KW
1179to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1180true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1181input character is malformed in some way, the program may croak, or the
1182function may return the REPLACEMENT CHARACTER, at the discretion of the
1183implementation, and subject to change in future releases.
1184
059703b0
KW
1185=for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1186Same as L</toFOLD_utf8>.
1f607577
KW
1187
1188=for apidoc Am|U8|toLOWER|U8 ch
1189Converts the specified character to lowercase. If the input is anything but an
1190ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 1191C<toLOWER_A> is equivalent.
954c1994 1192
1f607577 1193=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
1194Converts the specified Latin1 character to lowercase. The results are
1195undefined if the input doesn't fit in a byte.
1f607577
KW
1196
1197=for apidoc Am|U8|toLOWER_LC|U8 ch
1198Converts the specified character to lowercase using the current locale's rules,
1199if possible; otherwise returns the input character itself.
1200
d0da05db
KW
1201=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1202Converts the code point C<cp> to its lowercase version, and
1203stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1204point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1205that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1206bytes since the lowercase version may be longer than the original character.
1207
1208The first code point of the lowercased version is returned
21da7284
KW
1209(but note, as explained at L<the top of this section|/Character case
1210changing>, that there may be more).
1f607577 1211
bd350c85 1212=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1213Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1214extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
1215stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1216that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1217bytes since the lowercase version may be longer than the original character.
1218
1219The first code point of the lowercased version is returned
21da7284
KW
1220(but note, as explained at L<the top of this section|/Character case
1221changing>, that there may be more).
059703b0
KW
1222It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1223S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1224the UTF-8 for the input character is malformed in some way, the program may
1225croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1226of the implementation, and subject to change in future releases.
1f607577 1227
059703b0
KW
1228=for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1229Same as L</toLOWER_utf8>.
1f607577 1230
25200305
KW
1231=for apidoc Am|U8|toTITLE|U8 ch
1232Converts the specified character to titlecase. If the input is anything but an
1233ASCII lowercase character, that input character itself is returned. Variant
b7d90381 1234C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 1235range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 1236not a concept used in locale handling, so there is no functionality for that.)
25200305 1237
d0da05db
KW
1238=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1239Converts the code point C<cp> to its titlecase version, and
1240stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1241point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1242that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1243bytes since the titlecase version may be longer than the original character.
1244
1245The first code point of the titlecased version is returned
21da7284
KW
1246(but note, as explained at L<the top of this section|/Character case
1247changing>, that there may be more).
1f607577 1248
059703b0 1249=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
a239b1e2
KW
1250Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1251extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1252stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1253that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1254bytes since the titlecase version may be longer than the original character.
1255
1256The first code point of the titlecased version is returned
21da7284
KW
1257(but note, as explained at L<the top of this section|/Character case
1258changing>, that there may be more).
1f607577 1259
059703b0 1260It will not attempt
a239b1e2
KW
1261to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1262true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1263input character is malformed in some way, the program may croak, or the
1264function may return the REPLACEMENT CHARACTER, at the discretion of the
1265implementation, and subject to change in future releases.
1266
059703b0
KW
1267=for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1268Same as L</toTITLE_utf8>.
1f607577 1269
954c1994 1270=cut
353c9b6f 1271
d0da05db 1272XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1273really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1274and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1275
8a58bdcf 1276Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1277patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1278
954c1994
GS
1279*/
1280
8f5283f4
KW
1281/*
1282 void below because that's the best fit, and works for Devel::PPPort
1283=for apidoc AmnU|void|WIDEST_UTYPE
1284
1285Yields the widest unsigned integer type on the platform, currently either
1286C<U32> or C<64>. This can be used in declarations such as
1287
1288 WIDEST_UTYPE my_uv;
1289
1290or casts
1291
1292 my_uv = (WIDEST_UTYPE) val;
1293
1294=cut
1295
1296*/
de9e2639
KW
1297#ifdef QUADKIND
1298# define WIDEST_UTYPE U64
7c062697
KW
1299#else
1300# define WIDEST_UTYPE U32
1301#endif
1302
3912bc88
KW
1303/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1304 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1305 * bits of information are lost even on a 64-bit machine, but to get the
1306 * compiler to optimize it out if possible. This is because Configure makes
1307 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1308 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1309 *
1310 * For Coverity, be always true, because otherwise Coverity thinks
1311 * it finds several expressions that are always true, independent
1312 * of operands. Well, they are, but that is kind of the point.
220c71bf 1313 */
7e75d1a1 1314#ifndef __COVERITY__
6c5b02ac
KW
1315 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1316 * pointer) */
1317#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1318 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1319#else
1320#define FITS_IN_8_BITS(c) (1)
1321#endif
cf301eb7 1322
45f4bb73 1323/* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
833b0f46 1324 * Written this way so that after optimization, only one conditional test is
76d3ad4c
KW
1325 * needed. (The NV casts stop any warnings about comparison always being true
1326 * if called with an unsigned. The cast preserves the sign, which is all we
1327 * care about.) */
1328#define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1329 __ASSERT_((NV) (n) >= 0) \
1330 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
833b0f46 1331
94250c4f
KW
1332/* Returns true if c is in the range l..u, where 'l' is non-negative
1333 * Written this way so that after optimization, only one conditional test is
4758c20d 1334 * needed. */
1eaefa6e 1335#define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
4758c20d 1336 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
4758c20d
KW
1337 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1338 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
45f4bb73 1339 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
305fe86e 1340
41f43cc2 1341#ifdef EBCDIC
b6340bd0 1342# ifndef _ALL_SOURCE
0852beac
KW
1343 /* The native libc isascii() et.al. functions return the wrong results
1344 * on at least z/OS unless this is defined. */
b6340bd0
KW
1345# error _ALL_SOURCE should probably be defined
1346# endif
41f43cc2 1347#else
0852beac
KW
1348 /* There is a simple definition of ASCII for ASCII platforms. But the
1349 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1350 * other macros below.
3f3c579d
KW
1351 *
1352 * The cast here is used instead of '(c) >= 0', because some compilers emit
1353 * a warning that that test is always true when the parameter is an
1354 * unsigned type. khw supposes that it could be written as
1355 * && ((c) == '\0' || (c) > 0)
1356 * to avoid the message, but the cast will likely avoid extra branches even
1357 * with stupid compilers.
1358 *
1359 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1360 * a pointer) */
9c903d59 1361# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1362#endif
1363
38694112
KW
1364/* Take the eight possible bit patterns of the lower 3 bits and you get the
1365 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1366 * can be ignored. If the rest match '0', we have an octal */
1367#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1368
9fb1bf9d 1369#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1370
a500dc72
KW
1371/* Character class numbers. For internal core Perl use only. The ones less
1372 * than 32 are used in PL_charclass[] and the ones up through the one that
1373 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1374 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1375 * their actual definitions are here. If that file has a name not used here,
1376 * it won't compile.
1709d539
KW
1377 *
1378 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561 1379 * frequency of their use. This gives a slight edge to exiting a loop earlier
58a3ba2c
KW
1380 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1381 * real Posix character class, and some (small) inefficiencies in regular
1382 * expression handling would be introduced by putting it in the middle of those
1383 * that are. Also, cntrl and ascii come after the others as it may be useful
1384 * to group these which have no members that match above Latin1, (or above
1385 * ASCII in the latter case) */
1386
1709d539
KW
1387# define _CC_WORDCHAR 0 /* \w and [:word:] */
1388# define _CC_DIGIT 1 /* \d and [:digit:] */
1389# define _CC_ALPHA 2 /* [:alpha:] */
1390# define _CC_LOWER 3 /* [:lower:] */
1391# define _CC_UPPER 4 /* [:upper:] */
1392# define _CC_PUNCT 5 /* [:punct:] */
1393# define _CC_PRINT 6 /* [:print:] */
15861f94 1394# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1395# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1396# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
779cf272 1397# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
1398# define _CC_BLANK 11 /* [:blank:] */
1399# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1400# define _CC_CNTRL 13 /* [:cntrl:] */
1401# define _CC_ASCII 14 /* [:ascii:] */
1402# define _CC_VERTSPACE 15 /* \v */
1709d539 1403
a0947d7b
KW
1404# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1405
1709d539 1406/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1407 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1408# define _CC_IDFIRST 16
1409# define _CC_CHARNAME_CONT 17
1410# define _CC_NONLATIN1_FOLD 18
1411# define _CC_NONLATIN1_SIMPLE_FOLD 19
1412# define _CC_QUOTEMETA 20
1413# define _CC_NON_FINAL_FOLD 21
1414# define _CC_IS_IN_SOME_FOLD 22
2ae9030c
KW
1415# define _CC_BINDIGIT 23
1416# define _CC_OCTDIGIT 24
1417# define _CC_MNEMONIC_CNTRL 25
073c22b3
KW
1418
1419/* This next group is only used on EBCDIC platforms, so theoretically could be
1420 * shared with something entirely different that's only on ASCII platforms */
abb8abf6 1421# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
0654f0ab 1422/* Unused: 26-30
f4cdb42c
KW
1423 * If more bits are needed, one could add a second word for non-64bit
1424 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1425 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1426 * is used only for optimization (as of this writing), and differs in the
1427 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1428 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1429 * would just cause /i regexes which match them to run less efficiently.
1430 * Similarly the EBCDIC-only bits are used just for speed, and could be
1431 * replaced by other means */
96ac0975 1432
3a371f2f
KW
1433#if defined(PERL_CORE) || defined(PERL_EXT)
1434/* An enum version of the character class numbers, to help compilers
1435 * optimize */
1436typedef enum {
3a371f2f 1437 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1438 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1439 _CC_ENUM_ASCII = _CC_ASCII,
1440 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1441 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1442 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1443 _CC_ENUM_DIGIT = _CC_DIGIT,
1444 _CC_ENUM_GRAPH = _CC_GRAPH,
1445 _CC_ENUM_LOWER = _CC_LOWER,
1446 _CC_ENUM_PRINT = _CC_PRINT,
1447 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1448 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1449 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1450 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1451 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1452 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1453} _char_class_number;
1454#endif
1455
86f72d56 1456#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1457
6635f04f 1458START_EXTERN_C
96ac0975
NC
1459# ifdef DOINIT
1460EXTCONST U32 PL_charclass[] = {
1461# include "l1_char_class_tab.h"
1462};
1463
1464# else /* ! DOINIT */
1465EXTCONST U32 PL_charclass[];
1466# endif
6635f04f 1467END_EXTERN_C
96ac0975 1468
265c1f46 1469 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1470# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1471
1472 /* For internal core Perl use only: the base macro for defining macros like
1473 * isALPHA */
ff7ecfc3 1474# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1475 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1476
f4cdb42c
KW
1477 /* The mask for the _A versions of the macros; it just adds in the bit for
1478 * ASCII. */
1479# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1480
4650c663
KW
1481 /* For internal core Perl use only: the base macro for defining macros like
1482 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1483 * the ASCII bit are present */
b7d90381
KW
1484# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1485 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1486 == _CC_mask_A(classnum)))
f4cdb42c 1487
26c1d9d8
KW
1488/* On ASCII platforms certain classes form a single range. It's faster to
1489 * special case these. isDIGIT is a single range on all platforms */
b877c1ff
KW
1490# ifdef EBCDIC
1491# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1492# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1493# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1494# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1495# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1496# else
26c1d9d8 1497 /* By folding the upper and lowercase, we can use a single range */
b877c1ff 1498# define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
26c1d9d8 1499# define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
b877c1ff
KW
1500# define isLOWER_A(c) inRANGE(c, 'a', 'z')
1501# define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1502# define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1503# endif
15861f94 1504# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1505# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1506# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
b877c1ff 1507# define isDIGIT_A(c) inRANGE(c, '0', '9')
f4cdb42c
KW
1508# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1509# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
f4cdb42c 1510# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1511# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1512 */
d95f8b6a 1513# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1514# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1515# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1516# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1517
1518 /* continuation character for legal NAME in \N{NAME} */
1519# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1520
1521# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1522# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1523# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1524# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1525# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1526# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1527# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1528# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1529# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1530# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1531
0852beac
KW
1532# ifdef EBCDIC
1533# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1534# endif
1535
f12c0118
KW
1536 /* Participates in a single-character fold with a character above 255 */
1537# 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)))
1538
1539 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1540# 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 1541
4eeeb416 1542# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1543# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1544 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1545# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1546 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
5e6ebb12
KW
1547
1548/* is c a control character for which we have a mnemonic? */
1549# if defined(PERL_CORE) || defined(PERL_EXT)
1550# define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1551# endif
687c8d01 1552#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1553
1554 /* If we don't have perl.h, we are compiling a utility program. Below we
1555 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1556 * to it. Most are coded based on first principles. These are written to
74665a89 1557 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
182c4ace 1558# define isDIGIT_A(c) inRANGE(c, '0', '9')
0852beac 1559# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1560# define isSPACE_A(c) (isBLANK_A(c) \
1561 || (c) == '\n' \
1562 || (c) == '\r' \
1563 || (c) == '\v' \
0852beac 1564 || (c) == '\f')
74665a89
KW
1565 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1566 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1567 * performance (if optimization isn't on), and allow the same code to be
1568 * used for both platform types */
182c4ace
KW
1569# define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1570 || inRANGE((c), 'j', 'r') \
1571 || inRANGE((c), 's', 'z')
1572# define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1573 || inRANGE((c), 'J', 'R') \
1574 || inRANGE((c), 'S', 'Z')
a4d7a999
KW
1575# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1576# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1577# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1578# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
182c4ace
KW
1579# define isXDIGIT_A(c) ( isDIGIT_A(c) \
1580 || inRANGE((c), 'a', 'f') \
1581 || inRANGE((c), 'A', 'F')
74665a89
KW
1582# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1583 || (c) == '#' || (c) == '$' || (c) == '%' \
1584 || (c) == '&' || (c) == '\'' || (c) == '(' \
1585 || (c) == ')' || (c) == '*' || (c) == '+' \
1586 || (c) == ',' || (c) == '.' || (c) == '/' \
1587 || (c) == ':' || (c) == ';' || (c) == '<' \
1588 || (c) == '=' || (c) == '>' || (c) == '?' \
1589 || (c) == '@' || (c) == '[' || (c) == '\\' \
1590 || (c) == ']' || (c) == '^' || (c) == '_' \
1591 || (c) == '`' || (c) == '{' || (c) == '|' \
1592 || (c) == '}' || (c) == '~')
1593# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1594# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1595
0852beac 1596# ifdef EBCDIC
74665a89
KW
1597 /* The below is accurate for the 3 EBCDIC code pages traditionally
1598 * supported by perl. The only difference between them in the controls
1599 * is the position of \n, and that is represented symbolically below */
1600# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1601 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1602 || (c) == '\t' || (c) == '\v' \
182c4ace 1603 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
8ec0a736 1604 || (c) == 7F /* U+7F DEL */ \
182c4ace
KW
1605 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1606 DC[1-3] */ \
74665a89
KW
1607 || (c) == 0x18 /* U+18 CAN */ \
1608 || (c) == 0x19 /* U+19 EOM */ \
182c4ace 1609 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
74665a89
KW
1610 || (c) == 0x26 /* U+17 ETB */ \
1611 || (c) == 0x27 /* U+1B ESC */ \
1612 || (c) == 0x2D /* U+05 ENQ */ \
1613 || (c) == 0x2E /* U+06 ACK */ \
1614 || (c) == 0x32 /* U+16 SYN */ \
1615 || (c) == 0x37 /* U+04 EOT */ \
1616 || (c) == 0x3C /* U+14 DC4 */ \
1617 || (c) == 0x3D /* U+15 NAK */ \
1618 || (c) == 0x3F)/* U+1A SUB */
0852beac 1619# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1620# else /* isASCII is already defined for ASCII platforms, so can use that to
1621 define isCNTRL */
1622# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1623# endif
1624
3ffc8c70 1625 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1626 * during debugging, and it seems best to keep them. We may be called
1627 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1628 * do anything anyway, so make it not a problem */
1629# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1630# define NATIVE_TO_LATIN1(ch) (ch)
1631# endif
3ded5eb0
KW
1632# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1633# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1634# define isBLANK_L1(c) (isBLANK_A(c) \
1635 || (FITS_IN_8_BITS(c) \
1636 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1637# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1638# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1639# define isLOWER_L1(c) (isLOWER_A(c) \
1640 || (FITS_IN_8_BITS(c) \
ae683a5f 1641 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1642 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1643 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1644 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1645 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1646# define isPRINT_L1(c) (isPRINT_A(c) \
1647 || (FITS_IN_8_BITS(c) \
1648 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1649# define isPUNCT_L1(c) (isPUNCT_A(c) \
1650 || (FITS_IN_8_BITS(c) \
ae683a5f 1651 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1652 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1653 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1654 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1655 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1656 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1657 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1658# define isSPACE_L1(c) (isSPACE_A(c) \
1659 || (FITS_IN_8_BITS(c) \
ae683a5f 1660 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1661 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1662# define isUPPER_L1(c) (isUPPER_A(c) \
1663 || (FITS_IN_8_BITS(c) \
182c4ace
KW
1664 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1665 0xC0, 0xDE) \
3ded5eb0
KW
1666 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1667# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1668# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1669# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1670 || isBLANK_L1(c) \
1671 || (c) == '-' \
1672 || (c) == '(' \
1673 || (c) == ')')
1674 /* The following are not fully accurate in the above-ASCII range. I (khw)
1675 * don't think it's necessary to be so for the purposes where this gets
1676 * compiled */
1677# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1678# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1679
1680 /* And these aren't accurate at all. They are useful only for above
1681 * Latin1, which utilities and bootstrapping don't deal with */
1682# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1683# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1684# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1685
1686 /* Many of the macros later in this file are defined in terms of these. By
1687 * implementing them with a function, which converts the class number into
1688 * a call to the desired macro, all of the later ones work. However, that
1689 * function won't be actually defined when building a utility program (no
1690 * perl.h), and so a compiler error will be generated if one is attempted
1691 * to be used. And the above-Latin1 code points require Unicode tables to
1692 * be present, something unlikely to be the case when bootstrapping */
1693# define _generic_isCC(c, classnum) \
1694 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1695# define _generic_isCC_A(c, classnum) \
1696 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1697#endif /* End of no perl.h H_PERL */
8a58bdcf 1698
e66b99e9
KW
1699#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1700#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1701#define isASCII_A(c) isASCII(c)
1702#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1703#define isBLANK(c) isBLANK_A(c)
1704#define isCNTRL(c) isCNTRL_A(c)
1705#define isDIGIT(c) isDIGIT_A(c)
1706#define isGRAPH(c) isGRAPH_A(c)
1707#define isIDFIRST(c) isIDFIRST_A(c)
1708#define isLOWER(c) isLOWER_A(c)
1709#define isPRINT(c) isPRINT_A(c)
779cf272 1710#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1711#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1712#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1713#define isPUNCT(c) isPUNCT_A(c)
1714#define isSPACE(c) isSPACE_A(c)
1715#define isUPPER(c) isUPPER_A(c)
1716#define isWORDCHAR(c) isWORDCHAR_A(c)
1717#define isXDIGIT(c) isXDIGIT_A(c)
1718
1719/* ASCII casing. These could also be written as
1720 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1721 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1722 which uses table lookup and mask instead of subtraction. (This would
c5e9991e
KW
1723 work because the _MOD does not apply in the ASCII range).
1724
1725 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1726 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1727 there non-ASCII invariants, and all of them are controls.) */
68067e4e
DM
1728#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1729#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1730
25200305
KW
1731/* In the ASCII range, these are equivalent to what they're here defined to be.
1732 * But by creating these definitions, other code doesn't have to be aware of
c5e9991e
KW
1733 * this detail. Actually this works for all UTF-8 invariants, not just the
1734 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
25200305 1735#define toFOLD(c) toLOWER(c)
25200305
KW
1736#define toTITLE(c) toUPPER(c)
1737
c753c8d3
KW
1738#define toLOWER_A(c) toLOWER(c)
1739#define toUPPER_A(c) toUPPER(c)
25200305
KW
1740#define toFOLD_A(c) toFOLD(c)
1741#define toTITLE_A(c) toTITLE(c)
1a0901db 1742
4650c663 1743/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1744#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1745 ? (c) \
f4cd282c 1746 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1747#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1748
1a0901db 1749/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1750 * all mapped to one of them, and these need special handling; returns the
1751 * input itself if is out-of-range */
b2bf251f 1752#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1753 ? (c) \
f4cd282c 1754 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1755#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1756
beab9ebe
KW
1757/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1758
1759/* For internal core Perl use only: the base macro for defining macros like
1760 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1761 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1762 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1763 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1764 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1765 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1766 * code points under locale rules, use the macros based on _generic_LC_uvchr
1767 * instead of this. */
beab9ebe
KW
1768#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1769 (! FITS_IN_8_BITS(c) \
1770 ? 0 \
31f05a37
KW
1771 : IN_UTF8_CTYPE_LOCALE \
1772 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1773 : cBOOL(non_utf8))
1774
1775/* For internal core Perl use only: a helper macro for defining macros like
1776 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1777 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1778 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1779 * _generic_LC_base for more info */
1780#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1781 _generic_LC_base(c,utf8_locale_classnum, \
1782 non_utf8_func( (_LC_CAST) (c)))
1783
1784/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1785 * 'c' is the platform's native underscore character */
1786#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1787 _generic_LC_base(c, utf8_locale_classnum, \
1788 (non_utf8_func( (_LC_CAST) (c)) \
1789 || (char)(c) == '_'))
1790
1791/* These next three are also for internal core Perl use only: case-change
247985d4
KW
1792 * helper macros. The reason for using the PL_latin arrays is in case the
1793 * system function is defective; it ensures uniform results that conform to the
b257a28c 1794 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
beab9ebe
KW
1795#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1796 ? (c) \
31f05a37
KW
1797 : (IN_UTF8_CTYPE_LOCALE) \
1798 ? PL_latin1_lc[ (U8) (c) ] \
5a10328c 1799 : (cast)function((cast)(c)))
beab9ebe 1800
31f05a37
KW
1801/* Note that the result can be larger than a byte in a UTF-8 locale. It
1802 * returns a single value, so can't adequately return the upper case of LATIN
1803 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1804 * values "SS"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1805 * otherwise returns its input. It does not handle the anomalies in UTF-8
1806 * Turkic locales. */
beab9ebe
KW
1807#define _generic_toUPPER_LC(c, function, cast) \
1808 (! FITS_IN_8_BITS(c) \
1809 ? (c) \
31f05a37 1810 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1811 ? (cast)function((cast)(c)) \
31f05a37
KW
1812 : ((((U8)(c)) == MICRO_SIGN) \
1813 ? GREEK_CAPITAL_LETTER_MU \
1814 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1815 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1816 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1817 ? (__ASSERT_(0) (c)) \
1818 : PL_mod_latin1_uc[ (U8) (c) ])))))
1819
1820/* Note that the result can be larger than a byte in a UTF-8 locale. It
1821 * returns a single value, so can't adequately return the fold case of LATIN
1822 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1823 * values "ss"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1824 * otherwise returns its input. It does not handle the anomalies in UTF-8
1825 * Turkic locales */
beab9ebe 1826#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1827 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1828 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1829 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1830 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1831 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1832
84061b6a 1833/* Use the libc versions for these if available. */
f05550c0 1834#if defined(HAS_ISASCII)
84061b6a
KW
1835# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1836#else
1837# define isASCII_LC(c) isASCII(c)
1838#endif
1839
f05550c0 1840#if defined(HAS_ISBLANK)
84061b6a 1841# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1842#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1843# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1844#endif
1845
f05550c0 1846#define _LC_CAST U8
bbce6d69 1847
f05550c0 1848#ifdef WIN32
375f5f06
KW
1849 /* The Windows functions don't bother to follow the POSIX standard, which
1850 * for example says that something can't both be a printable and a control.
1851 * But Windows treats the \t control as a printable, and does such things
1852 * as making superscripts into both digits and punctuation. This tames
1853 * these flaws by assuming that the definitions of both controls and space
1854 * are correct, and then making sure that other definitions don't have
1855 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1856 * Not all possible weirdnesses are checked for, just the ones that were
1857 * detected on actual Microsoft code pages */
1858
b7d90381
KW
1859# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1860# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1861
1862# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1863 && isALPHANUMERIC_LC(c))
1864# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1865 ! isPUNCT_LC(c))
1866# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1867 isALPHANUMERIC_LC(c))
1868# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1869# define isIDFIRST_LC(c) (((c) == '_') \
1870 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1871# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1872# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1873# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1874# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1875# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1876# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1877 && isALPHANUMERIC_LC(c))
f05550c0
BF
1878
1879# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1880# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1881# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1882
1883#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1884 /* For most other platforms */
beab9ebe 1885
f05550c0
BF
1886# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1887# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1888# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1889# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1890# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1891# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1892# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1893# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1894# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1895# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1896# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1897# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1898# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1899
1900
1901# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1902# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1903# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1904
1905#else /* The final fallback position */
1906
b7d90381
KW
1907# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1908# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1909# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1910# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1911# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1912# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1913# define isLOWER_LC(c) (isascii(c) && islower(c))
1914# define isPRINT_LC(c) (isascii(c) && isprint(c))
1915# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1916# define isSPACE_LC(c) (isascii(c) && isspace(c))
1917# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1918# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1919# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1920
1921# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1922# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1923# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1924
f05550c0 1925#endif
55204971 1926
eba68aa0
KW
1927#define isIDCONT(c) isWORDCHAR(c)
1928#define isIDCONT_A(c) isWORDCHAR_A(c)
1929#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1930#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1931#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1932
4650c663 1933/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1934 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1935 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1936 * classes where there is a macro or function 'above_latin1' that takes the
1937 * single argument 'c' and returns the desired value. These exist for those
2366ba44
KW
1938 * classes which have simple definitions, avoiding the overhead of an inversion
1939 * list binary search. _generic_invlist_uvchr() can be used
4650c663 1940 * for classes where that overhead is faster than a direct lookup.
d0da05db 1941 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1942 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1943 * have duplicate checks here, so could create versions of the macros that
1944 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1945
1946/* Note that all ignore 'use bytes' */
1e222e4f
KW
1947#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1948 ? _generic_isCC(c, classnum) \
cd500f2f 1949 : above_latin1(c))
2366ba44 1950#define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
1e222e4f 1951 ? _generic_isCC(c, classnum) \
922e8cb4 1952 : _is_uni_FOO(classnum, c))
2366ba44
KW
1953#define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
1954#define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
d0da05db
KW
1955#define isASCII_uvchr(c) isASCII(c)
1956#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1957#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2366ba44
KW
1958#define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
1959#define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1960#define isIDCONT_uvchr(c) \
1961 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1962#define isIDFIRST_uvchr(c) \
1963 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2366ba44
KW
1964#define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
1965#define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
d0da05db 1966
2366ba44 1967#define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
d0da05db
KW
1968#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1969#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1970
2366ba44 1971#define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
d0da05db 1972#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2366ba44 1973#define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
d0da05db
KW
1974#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1975
1976#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1977#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1978#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1979#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1980
1981/* For backwards compatibility, even though '_uni' should mean official Unicode
1982 * code points, in Perl it means native for those below 256 */
1983#define isALPHA_uni(c) isALPHA_uvchr(c)
1984#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1985#define isASCII_uni(c) isASCII_uvchr(c)
1986#define isBLANK_uni(c) isBLANK_uvchr(c)
1987#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1988#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1989#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1990#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1991#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1992#define isLOWER_uni(c) isLOWER_uvchr(c)
1993#define isPRINT_uni(c) isPRINT_uvchr(c)
1994#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1995#define isSPACE_uni(c) isSPACE_uvchr(c)
1996#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1997#define isUPPER_uni(c) isUPPER_uvchr(c)
1998#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1999#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2000#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2001#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2002#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2003#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2004#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 2005
4650c663
KW
2006/* For internal core Perl use only: the base macros for defining macros like
2007 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 2008 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 2009 * simple class definitions; the other for more complex. These are like
d0da05db 2010 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
2011#define _generic_LC_uvchr(latin1, above_latin1, c) \
2012 (c < 256 ? latin1(c) : above_latin1(c))
2366ba44 2013#define _generic_LC_invlist_uvchr(latin1, classnum, c) \
cd500f2f
KW
2014 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2015
2366ba44
KW
2016#define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2017#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
cd500f2f 2018 _CC_ALPHANUMERIC, c)
b7d90381
KW
2019#define isASCII_LC_uvchr(c) isASCII_LC(c)
2020#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2021 is_HORIZWS_cp_high, c)
feeab5a9 2022#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2366ba44
KW
2023#define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2024#define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 2025#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 2026 _is_uni_perl_idcont, c)
b7d90381 2027#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f 2028 _is_uni_perl_idstart, c)
2366ba44
KW
2029#define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2030#define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 2031#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2366ba44 2032#define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 2033#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 2034 is_XPERLSPACE_cp_high, c)
2366ba44
KW
2035#define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2036#define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
cd500f2f 2037 _CC_WORDCHAR, c)
b7d90381
KW
2038#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2039 is_XDIGIT_cp_high, c)
e712593e 2040
b7d90381 2041#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 2042
da8c1a98
KW
2043/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2044 * they don't otherwise go out of their way to look for malformed UTF-8. If
2045 * they can return accurate results without knowing if the input is otherwise
2046 * malformed, they do so. For example isASCII is accurate in spite of any
2047 * non-length malformations because it looks only at a single byte. Likewise
2048 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2049 * variant ones return FALSE. But, if the input has to be well-formed in order
2050 * for the results to be accurate, the macros will test and if malformed will
2051 * call a routine to die
2052 *
2053 * Except for toke.c, the macros do assume that e > p, asserting that on
2054 * DEBUGGING builds. Much code that calls these depends on this being true,
2055 * for other reasons. toke.c is treated specially as using the regular
2056 * assertion breaks it in many ways. All strings that these operate on there
2057 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2058 * bunch of code in toke.c assumes that this is true, so the assertion allows
2059 * for that */
2060#ifdef PERL_IN_TOKE_C
2061# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2062#else
2063# define _utf8_safe_assert(p,e) ((e) > (p))
2064#endif
2065
2066#define _generic_utf8_safe(classnum, p, e, above_latin1) \
c81b3562
KW
2067 ((! _utf8_safe_assert(p, e)) \
2068 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2069 : (UTF8_IS_INVARIANT(*(p))) \
da8c1a98
KW
2070 ? _generic_isCC(*(p), classnum) \
2071 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2072 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2073 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2074 classnum) \
2075 : (_force_out_malformed_utf8_message( \
2076 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2077 : above_latin1))
b7d90381
KW
2078/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2079 * 'above_latin1' can be a macro */
da8c1a98
KW
2080#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2081 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2366ba44 2082#define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2083 _generic_utf8_safe(classnum, p, e, \
2084 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2085 ? (_force_out_malformed_utf8_message( \
2086 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2087 : above_latin1(p)))
2366ba44
KW
2088/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2089 * 'above_latin1' parameter */
2090#define _generic_invlist_utf8_safe(classnum, p, e) \
2091 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
922e8cb4 2092
cc8ab7c0 2093/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
2094 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2095 * class is TRUE for. Hence it can skip the tests for this range.
2096 * 'above_latin1' should include its arguments */
da8c1a98
KW
2097#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2098 (__ASSERT_(_utf8_safe_assert(p, e)) \
2099 (UTF8_IS_INVARIANT(*(p))) \
2100 ? _generic_isCC(*(p), classnum) \
2101 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2102 ? 0 /* Note that doesn't check validity for latin1 */ \
2103 : above_latin1)
2104
84238efa 2105
059703b0
KW
2106#define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2107#define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2108#define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2109#define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2110#define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2111#define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2112#define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2113#define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2114#define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2115#define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2116#define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2117#define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2118#define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2119#define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2120#define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2121#define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2122#define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2123#define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
e8fa43e2 2124
2366ba44 2125#define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
da8c1a98 2126#define isALPHANUMERIC_utf8_safe(p, e) \
2366ba44 2127 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
da8c1a98
KW
2128#define isASCII_utf8_safe(p, e) \
2129 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2130 * works */ \
2131 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2132#define isBLANK_utf8_safe(p, e) \
2366ba44 2133 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
da8c1a98 2134
e8fa43e2
KW
2135#ifdef EBCDIC
2136 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2137 * more efficient macro instead of the more general one */
da8c1a98 2138# define isCNTRL_utf8_safe(p, e) \
56d02b8c 2139 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 2140#else
da8c1a98 2141# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
2142#endif
2143
da8c1a98
KW
2144#define isDIGIT_utf8_safe(p, e) \
2145 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2366ba44
KW
2146 _is_utf8_FOO(_CC_DIGIT, p, e))
2147#define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
da8c1a98 2148#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
dd1a3ba7 2149 _is_utf8_perl_idcont, p, e)
e5dcd934 2150
c11ff943
KW
2151/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2152 * IDFIRST is an alnum. See
8034715d 2153 * https://github.com/Perl/perl5/issues/10275 for more detail than you
f91dcd13
KW
2154 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2155 * This used to be not the XID version, but we decided to go with the more
2156 * modern Unicode definition */
da8c1a98
KW
2157#define isIDFIRST_utf8_safe(p, e) \
2158 _generic_func_utf8_safe(_CC_IDFIRST, \
dd1a3ba7 2159 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
da8c1a98 2160
2366ba44
KW
2161#define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2162#define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
da8c1a98 2163#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2366ba44 2164#define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
da8c1a98 2165#define isSPACE_utf8_safe(p, e) \
2366ba44
KW
2166 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2167#define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
da8c1a98 2168#define isVERTWS_utf8_safe(p, e) \
2366ba44 2169 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
da8c1a98 2170#define isWORDCHAR_utf8_safe(p, e) \
2366ba44 2171 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
da8c1a98
KW
2172#define isXDIGIT_utf8_safe(p, e) \
2173 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2174 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2175 ? (_force_out_malformed_utf8_message( \
2176 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2177 : is_XDIGIT_high(p)))
a0ed51b3 2178
059703b0
KW
2179#define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2180#define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2181#define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2182#define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2e8adce6 2183
567b353c 2184/* For internal core use only, subject to change */
059703b0
KW
2185#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2186#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2187#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2188#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
a1a5ec35
KW
2189
2190#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2191#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2192#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2193#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 2194
059703b0
KW
2195#define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2196#define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2197#define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2198#define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2199#define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2200#define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2201#define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2202#define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2203#define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2204#define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2205#define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2206#define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2207#define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2208#define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2209#define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2210#define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2211#define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
34aeb2e9 2212
da8c1a98
KW
2213/* For internal core Perl use only: the base macros for defining macros like
2214 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2215 * point in 'p' is within the 0-255 range, it uses locale rules from the
2216 * passed-in 'macro' parameter */
2217#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2218 (__ASSERT_(_utf8_safe_assert(p, e)) \
2219 (UTF8_IS_INVARIANT(*(p))) \
2220 ? macro(*(p)) \
2221 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2222 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2223 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2224 : (_force_out_malformed_utf8_message( \
2225 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2226 : above_latin1))
2227
2366ba44
KW
2228#define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2229 _generic_LC_utf8_safe(macro, p, e, \
2230 _is_utf8_FOO(classnum, p, e))
da8c1a98
KW
2231
2232#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2233 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2234
2366ba44 2235#define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
da8c1a98
KW
2236 _generic_LC_utf8_safe(classnum, p, e, \
2237 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2238 ? (_force_out_malformed_utf8_message( \
2239 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2240 : above_latin1(p)))
2241
2242#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2366ba44 2243 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
da8c1a98
KW
2244 _CC_ALPHANUMERIC, p, e)
2245#define isALPHA_LC_utf8_safe(p, e) \
2366ba44 2246 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
da8c1a98
KW
2247#define isASCII_LC_utf8_safe(p, e) \
2248 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2249#define isBLANK_LC_utf8_safe(p, e) \
2366ba44 2250 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
da8c1a98
KW
2251#define isCNTRL_LC_utf8_safe(p, e) \
2252 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2253#define isDIGIT_LC_utf8_safe(p, e) \
2366ba44 2254 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
da8c1a98 2255#define isGRAPH_LC_utf8_safe(p, e) \
2366ba44 2256 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
da8c1a98
KW
2257#define isIDCONT_LC_utf8_safe(p, e) \
2258 _generic_LC_func_utf8_safe(isIDCONT_LC, \
dd1a3ba7 2259 _is_utf8_perl_idcont, p, e)
da8c1a98
KW
2260#define isIDFIRST_LC_utf8_safe(p, e) \
2261 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
dd1a3ba7 2262 _is_utf8_perl_idstart, p, e)
da8c1a98 2263#define isLOWER_LC_utf8_safe(p, e) \
2366ba44 2264 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
da8c1a98 2265#define isPRINT_LC_utf8_safe(p, e) \
2366ba44 2266 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
da8c1a98
KW
2267#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2268#define isPUNCT_LC_utf8_safe(p, e) \
2366ba44 2269 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
da8c1a98 2270#define isSPACE_LC_utf8_safe(p, e) \
2366ba44 2271 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
da8c1a98 2272#define isUPPER_LC_utf8_safe(p, e) \
2366ba44 2273 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
da8c1a98 2274#define isWORDCHAR_LC_utf8_safe(p, e) \
2366ba44 2275 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
da8c1a98 2276#define isXDIGIT_LC_utf8_safe(p, e) \
2366ba44 2277 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2278
fbc19f27
KW
2279/* Macros for backwards compatibility and for completeness when the ASCII and
2280 * Latin1 values are identical */
b7d90381
KW
2281#define isALPHAU(c) isALPHA_L1(c)
2282#define isDIGIT_L1(c) isDIGIT_A(c)
2283#define isOCTAL(c) isOCTAL_A(c)
2284#define isOCTAL_L1(c) isOCTAL_A(c)
2285#define isXDIGIT_L1(c) isXDIGIT_A(c)
2286#define isALNUM(c) isWORDCHAR(c)
a377c856 2287#define isALNUM_A(c) isALNUM(c)
b7d90381
KW
2288#define isALNUMU(c) isWORDCHAR_L1(c)
2289#define isALNUM_LC(c) isWORDCHAR_LC(c)
2290#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2291#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
059703b0 2292#define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
4c1d9526 2293#define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
059703b0 2294#define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
4c1d9526 2295#define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
b7d90381
KW
2296#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2297#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2298#define isALNUMC(c) isALPHANUMERIC(c)
2299#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2300#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2301#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
059703b0 2302#define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
4c1d9526
KW
2303#define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2304#define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
fbc19f27 2305
2bd1cbf6
KW
2306/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2307 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2308 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2309 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2310 * the outlier from the block that contains the other controls, just like
2311 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2312 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2313 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2314 * etc. */
2bd1cbf6 2315#ifndef EBCDIC
75763b3a 2316# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2317#else
75763b3a
KW
2318# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2319 ((isPRINT_A(c)) \
2320 ? (UNLIKELY((c) == '?') \
2321 ? QUESTION_MARK_CTRL \
2322 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2323 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2324 ? '?' \
2325 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2326#endif
bbce6d69 2327
dea28490
JJ
2328/* Line numbers are unsigned, 32 bits. */
2329typedef U32 line_t;
e5dcd934 2330#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2331
91152fc1
DG
2332/* Helpful alias for version prescan */
2333#define is_LAX_VERSION(a,b) \
2334 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2335
2336#define is_STRICT_VERSION(a,b) \
2337 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2338
2339#define BADVERSION(a,b,c) \
2340 if (b) { \
2341 *b = c; \
2342 } \
2343 return a;
8c52afec 2344
1ce77b7d
KW
2345/* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2346 * a-f) to its numeric value without using any branches. The input is
2347 * validated only by an assert() in DEBUGGING builds.
2348 *
2349 * It works by right shifting and isolating the bit that is 0 for the digits,
2350 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2351 * position, and then to the eights position. Both are added together to form
2352 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2353 * final four bits of the input to form the correct value. */
2354#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2355 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2356 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2357 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2358
2359/* The argument is a string pointer, which is advanced. */
2360#define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
95a59cab 2361
cb27eebd
KW
2362/* Converts a character known to represent an octal digit (0-7) to its numeric
2363 * value. The input is validated only by an assert() in DEBUGGING builds. In
2364 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2365#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2366
305b8651
KW
2367/* Efficiently returns a boolean as to if two native characters are equivalent
2368 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2369 * the ALPHA in the name is to remind you of that. This is asserted() in
2370 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2371 * works (on valid input) for both non- and UTF-8-encoded bytes.
2372 *
2373 * When one of the inputs is a compile-time constant and gets folded by the
2374 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2375 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2376 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2377 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2378 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2379 * just a single 0, in the bit position where the upper- and lowercase differ.
2380 * */
305b8651
KW
2381#define isALPHA_FOLD_EQ(c1, c2) \
2382 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2383 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2384#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2385
8e84507e 2386/*
ccfc67b7
JH
2387=head1 Memory Management
2388
a02a5408 2389=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2390The XSUB-writer's interface to the C C<malloc> function.
2391
596f7718 2392Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2393
c5008215
JC
2394In 5.9.3, Newx() and friends replace the older New() API, and drops
2395the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2396themselves. This aid has been superseded by a new build option,
d10b4965 2397PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2398there for use in XS modules supporting older perls.
2399
a02a5408 2400=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2401The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2402cast. See also C<L</Newx>>.
954c1994 2403
596f7718 2404Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2405
a02a5408 2406=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2407The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2408memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2409
596f7718 2410Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2411
954c1994
GS
2412=for apidoc Am|void|Renew|void* ptr|int nitems|type
2413The XSUB-writer's interface to the C C<realloc> function.
2414
596f7718 2415Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2416
954c1994
GS
2417=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2418The XSUB-writer's interface to the C C<realloc> function, with
2419cast.
2420
596f7718 2421Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2422
49b8b560 2423=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2424The XSUB-writer's interface to the C C<free> function.
2425
596f7718 2426This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2427
954c1994
GS
2428=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2429The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2430source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2431C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2432
e90e2364 2433=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2434Like C<Move> but returns C<dest>. Useful
72d33970 2435for encouraging compilers to tail-call
e90e2364
NC
2436optimise.
2437
954c1994
GS
2438=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2439The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2440source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2441C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2442
e90e2364
NC
2443=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2444
796b6530 2445Like C<Copy> but returns C<dest>. Useful
72d33970 2446for encouraging compilers to tail-call
e90e2364
NC
2447optimise.
2448
954c1994
GS
2449=for apidoc Am|void|Zero|void* dest|int nitems|type
2450
2451The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2452destination, C<nitems> is the number of items, and C<type> is the type.
2453
e90e2364
NC
2454=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2455
72d33970
FC
2456Like C<Zero> but returns dest. Useful
2457for encouraging compilers to tail-call
e90e2364
NC
2458optimise.
2459
da5d8dbb 2460=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2461This is an architecture-independent macro to copy one structure to another.
954c1994 2462
7e337ee0
JH
2463=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2464
2465Fill up memory with a byte pattern (a byte repeated over and over
2466again) that hopefully catches attempts to access uninitialized memory.
2467
2468=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2469
2470PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2471
1c12ffb4 2472=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2473
2474PoisonWith(0xEF) for catching access to freed memory.
2475
9965345d
JH
2476=for apidoc Am|void|Poison|void* dest|int nitems|type
2477
7e337ee0 2478PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2479
2480=cut */
954c1994 2481
561b68a9
SH
2482/* Maintained for backwards-compatibility only. Use newSV() instead. */
2483#ifndef PERL_CORE
ff06c60c 2484#define NEWSV(x,len) newSV(len)
561b68a9 2485#endif
ff06c60c 2486
b7112dce 2487#define MEM_SIZE_MAX ((MEM_SIZE)-1)
19a94d75 2488
a500027b 2489#define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
e6bdf523 2490
27d5b266 2491#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2492
2493/* This expression will be constant-folded at compile time. It checks
2494 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2495 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2496 * As well as avoiding the need for a run-time check in some cases, it's
2497 * designed to avoid compiler warnings like:
2498 * comparison is always false due to limited range of data type
73e8ff00
DM
2499 * It's mathematically equivalent to
2500 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2501 */
2502
2503# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
445198b9
LM
2504 ( sizeof(MEM_SIZE) < sizeof(n) \
2505 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 2506
88f9f128 2507/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2508 * compiler warnings. We *want* to write the expression as
2509 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2510 * (for some compile-time constant C), but even when the LHS
2511 * constant-folds to false at compile-time, g++ insists on emitting
2512 * warnings about the RHS (e.g. "comparison is always false"), so instead
2513 * we write it as
e6bdf523 2514 *
d98e5cde 2515 * (cond ? n : X) > C
88f9f128 2516 *
d98e5cde
DM
2517 * where X is a constant with X > C always false. Choosing a value for X
2518 * is tricky. If 0, some compilers will complain about 0 > C always being
2519 * false; if 1, Coverity complains when n happens to be the constant value
2520 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2521 * for X and hope that nothing else whines.
e6bdf523
DM
2522 */
2523
2524# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2525 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2526 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2527
2528# define MEM_WRAP_CHECK(n,t) \
2529 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2530 && (croak_memory_wrap(),0))
2531
2532# define MEM_WRAP_CHECK_1(n,t,a) \
2533 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2534 && (Perl_croak_nocontext("%s",(a)),0))
2535
814eedc8
DD
2536/* "a" arg must be a string literal */
2537# define MEM_WRAP_CHECK_s(n,t,a) \
2538 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2539 && (Perl_croak_nocontext("" a ""),0))
2540
8b44ba4c 2541#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2542
a500027b 2543#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
2544#else
2545
410319be
NC
2546#define MEM_WRAP_CHECK(n,t)
2547#define MEM_WRAP_CHECK_1(n,t,a)
814eedc8 2548#define MEM_WRAP_CHECK_s(n,t,a)
8b44ba4c
NC
2549#define MEM_WRAP_CHECK_(n,t)
2550
a500027b 2551#define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
27d5b266 2552
1936d2a7 2553#endif
8b44ba4c 2554
fe4f188c 2555#ifdef PERL_MEM_LOG
46c6c7e2 2556/*
9f653bb5 2557 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2558 * go through functions, which are handy for debugging breakpoints, but
2559 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2560 * line number, and C function name if available) passed in. This info can
2561 * then be used for logging the calls, for which one gets a sample
73d1d973 2562 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2563 *
46c6c7e2 2564 * Known problems:
94e892a6 2565 * - not all memory allocs get logged, only those
46c6c7e2 2566 * that go through Newx() and derivatives (while all
94e892a6 2567 * Safefrees do get logged)
46c6c7e2
JH
2568 * - __FILE__ and __LINE__ do not work everywhere
2569 * - __func__ or __FUNCTION__ even less so
2570 * - I think more goes on after the perlio frees but
2571 * the thing is that STDERR gets closed (as do all
2572 * the file descriptors)
2573 * - no deeper calling stack than the caller of the Newx()
2574 * or the kind, but do I look like a C reflection/introspection
2575 * utility to you?
2576 * - the function prototypes for the logging functions
2577 * probably should maybe be somewhere else than handy.h
2578 * - one could consider inlining (macrofying) the logging
2579 * for speed, but I am too lazy
2580 * - one could imagine recording the allocations in a hash,
2581 * (keyed by the allocation address?), and maintain that
2582 * through reallocs and frees, but how to do that without
2583 * any News() happening...?
73d1d973 2584 * - lots of -Ddefines to get useful/controllable output
b953482e 2585 * - lots of ENV reads
46c6c7e2
JH
2586 */
2587
0b0ab801 2588# ifdef PERL_CORE
73d1d973 2589# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2590enum mem_log_type {
2591 MLT_ALLOC,
2592 MLT_REALLOC,
d7a2c63c
MHM
2593 MLT_FREE,
2594 MLT_NEW_SV,
2595 MLT_DEL_SV
0b0ab801
MHM
2596};
2597# endif
12754f92 2598# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2599void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2600void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2601# endif
0b0ab801
MHM
2602# endif
2603
fe4f188c
JH
2604#endif
2605
2606#ifdef PERL_MEM_LOG
d1401ee9
MHM
2607#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2608#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2609#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2610#endif
2611
2612#ifndef MEM_LOG_ALLOC
2613#define MEM_LOG_ALLOC(n,t,a) (a)
2614#endif
2615#ifndef MEM_LOG_REALLOC
2616#define MEM_LOG_REALLOC(n,t,v,a) (a)
2617#endif
2618#ifndef MEM_LOG_FREE
2619#define MEM_LOG_FREE(a) (a)
2620#endif
2621
d1401ee9
MHM
2622#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2623#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2624#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2625
2626#ifndef PERL_CORE
a02a5408
JC
2627/* pre 5.9.x compatibility */
2628#define New(x,v,n,t) Newx(v,n,t)
2629#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2630#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2631#endif
a02a5408 2632
ff68c719 2633#define Renew(v,n,t) \
d1401ee9 2634 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2635#define Renewc(v,n,t,c) \
d1401ee9 2636 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2637
2638#ifdef PERL_POISON
2639#define Safefree(d) \
06c0cc96 2640 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2641#else
fe4f188c 2642#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2643#endif
55497cff 2644
dbb57106
YO
2645/* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2646 * as it handles cases like constant string arguments without throwing warnings *
2647 * the cast is required, as is the inequality check, to avoid warnings */
45908e4d 2648#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2649
45908e4d
YO
2650
2651#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)))
2652#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)))
2653#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2654
bdd1531d 2655/* Like above, but returns a pointer to 'd' */
45908e4d
YO
2656#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)))
2657#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 2658#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2659
7e337ee0
JH
2660#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2661#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2662#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2663#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2664
caa674f3
DD
2665#ifdef PERL_POISON
2666# define PERL_POISON_EXPR(x) x
2667#else
2668# define PERL_POISON_EXPR(x)
2669#endif
2670
ff68c719 2671#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2672
1b7e2294
KW
2673/*
2674=head1 Handy Values
2675
2676=for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2677
2678Returns the number of elements in the input C array (so you want your
2679zero-based indices to be less than but not equal to).
2680
2681=for apidoc Am|void *|C_ARRAY_END|void *a
2682
2683Returns a pointer to one element past the final element of the input C array.
2684
2685=cut
2686
2687C_ARRAY_END is one past the last: half-open/half-closed range, not
2688last-inclusive range.
2689*/
622913ab 2690#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2691#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2692
2cc61e15
DD
2693#ifdef NEED_VA_COPY
2694# ifdef va_copy
2695# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2696# elif defined(__va_copy)
2697# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2698# else
07798b17 2699# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2700# endif
2701#endif
2702
472d47bc
SB
2703/* convenience debug macros */
2704#ifdef USE_ITHREADS
2705#define pTHX_FORMAT "Perl interpreter: 0x%p"
2706#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2707#define pTHX_VALUE_ (void *)my_perl,
2708#define pTHX_VALUE (void *)my_perl
2709#define pTHX__VALUE_ ,(void *)my_perl,
2710#define pTHX__VALUE ,(void *)my_perl
472d47bc 2711#else
3609ea0d 2712#define pTHX_FORMAT
472d47bc 2713#define pTHX__FORMAT
3609ea0d 2714#define pTHX_VALUE_
472d47bc 2715#define pTHX_VALUE
3609ea0d 2716#define pTHX__VALUE_
472d47bc
SB
2717#define pTHX__VALUE
2718#endif /* USE_ITHREADS */
3609ea0d 2719
2acdbac1
NC
2720/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2721 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2722 CPAN::Unpack show any users outside the core. */
2723#ifdef PERL_CORE
dc6e8de0
A
2724# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2725 "Use of " s " is deprecated")
c9680906
A
2726# define deprecate_disappears_in(when,message) \
2727 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2728 message ", and will disappear in Perl " when)
ac641426
A
2729# define deprecate_fatal_in(when,message) \
2730 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2731 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2732#endif
2733
dfff4baf
BF
2734/* Internal macros to deal with gids and uids */
2735#ifdef PERL_CORE
2736
2737# if Uid_t_size > IVSIZE
2738# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2739# define SvUID(sv) SvNV(sv)
07798b17
AC
2740# elif Uid_t_sign <= 0
2741# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2742# define SvUID(sv) SvIV(sv)
dfff4baf 2743# else
07798b17
AC
2744# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2745# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2746# endif /* Uid_t_size */
2747
2748# if Gid_t_size > IVSIZE
2749# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2750# define SvGID(sv) SvNV(sv)
07798b17
AC
2751# elif Gid_t_sign <= 0
2752# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2753# define SvGID(sv) SvIV(sv)
dfff4baf 2754# else
07798b17
AC
2755# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2756# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2757# endif /* Gid_t_size */
2758
2759#endif
2760
6a5bc5ac 2761#endif /* PERL_HANDY_H_ */
9d745869 2762
e9a8c099 2763/*
14d04a33 2764 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2765 */