This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
gv.c: Note that some GV_foo constants are documented
[perl5.git] / handy.h
CommitLineData
a0d0e21e 1/* handy.h
a687059c 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
da5d8dbb 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
a687059c 5 *
6e21c824
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8d063cd8 8 *
8d063cd8
LW
9 */
10
4650c663
KW
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
6a5bc5ac
KW
14#ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */
15#define PERL_HANDY_H_
9d745869 16
24792b8d
NC
17#ifndef PERL_CORE
18# define Null(type) ((type)NULL)
954c1994
GS
19
20/*
ccfc67b7 21=head1 Handy Values
954c1994 22
78342678 23=for apidoc AmnU||Nullch
72d33970
FC
24Null character pointer. (No longer available when C<PERL_CORE> is
25defined.)
2307c6d0 26
78342678 27=for apidoc AmnU||Nullsv
72d33970 28Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
29
30=cut
31*/
32
24792b8d
NC
33# define Nullch Null(char*)
34# define Nullfp Null(PerlIO*)
35# define Nullsv Null(SV*)
36#endif
8d063cd8 37
641d3f0b
PP
38#ifdef TRUE
39#undef TRUE
40#endif
41#ifdef FALSE
42#undef FALSE
43#endif
44#define TRUE (1)
45#define FALSE (0)
46
cf3f0ffb
DM
47/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
48 * (compiler permitting) that casting away const-ness will give a warning;
49 * e.g.:
50 *
51 * const SV *sv = ...;
52 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
53 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
54 */
55
b1bc3f34 56#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
f065df04 57# define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
b1bc3f34
NC
58#else
59# define MUTABLE_PTR(p) ((void *) (p))
60#endif
61
a062e10d 62#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 63#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 64#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 65#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 66#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 67#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 68
f789f6a4 69#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
70# include <stdbool.h>
71# ifndef HAS_BOOL
72# define HAS_BOOL 1
73# endif
74#endif
75
8e84507e 76/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
77 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
78 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 79 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
80 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
81 g++ can be identified by __GNUG__.
82 Andy Dougherty February 2000
5d94fbed 83*/
3609ea0d 84#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 85# ifndef PERL_BOOL_AS_CHAR
5d94fbed 86# ifndef HAS_BOOL
c1d22f6b 87# define HAS_BOOL 1
5d94fbed 88# endif
f789f6a4 89# endif
5d94fbed
AD
90#endif
91
92#ifndef HAS_BOOL
f789f6a4
FC
93# ifdef bool
94# undef bool
95# endif
70d5cb32 96# define bool char
c1d22f6b 97# define HAS_BOOL 1
a687059c 98#endif
0d3e774c 99
25ba28ce
KW
100/*
101=for apidoc Am|bool|cBOOL|bool expr
102
103Cast-to-bool. A simple S<C<(bool) I<expr>>> cast may not do the right thing:
104if C<bool> is defined as C<char>, for example, then the cast from C<int> is
105implementation-defined.
106
107C<(bool)!!(cbool)> in a ternary triggers a bug in xlc on AIX
108
109=cut
110*/
18f5643b 111#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 112
46c6c7e2 113/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
114 * XXX Should really be a Configure probe, with HAS__FUNCTION__
115 * and FUNCTION__ as results.
116 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
46c6c7e2
JH
117#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
118# define FUNCTION__ __func__
07798b17
AC
119#elif (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
120 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
121# define FUNCTION__ ""
46c6c7e2 122#else
07798b17 123# define FUNCTION__ __FUNCTION__ /* Common extension. */
46c6c7e2
JH
124#endif
125
27d4fb96
PP
126/* XXX A note on the perl source internal type system. The
127 original intent was that I32 be *exactly* 32 bits.
128
129 Currently, we only guarantee that I32 is *at least* 32 bits.
130 Specifically, if int is 64 bits, then so is I32. (This is the case
131 for the Cray.) This has the advantage of meshing nicely with
132 standard library calls (where we pass an I32 and the library is
133 expecting an int), but the disadvantage that an I32 is not 32 bits.
134 Andy Dougherty August 1996
24fef2a7 135
dc45a647
MB
136 There is no guarantee that there is *any* integral type with
137 exactly 32 bits. It is perfectly legal for a system to have
138 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 139
dc45a647
MB
140 Similarly, there is no guarantee that I16 and U16 have exactly 16
141 bits.
693762b4 142
8e84507e
NIS
143 For dealing with issues that may arise from various 32/64-bit
144 systems, we will ask Configure to check out
8175356b 145
3609ea0d
JH
146 SHORTSIZE == sizeof(short)
147 INTSIZE == sizeof(int)
148 LONGSIZE == sizeof(long)
dc45a647 149 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 150 PTRSIZE == sizeof(void *)
dc45a647
MB
151 DOUBLESIZE == sizeof(double)
152 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 153
27d4fb96
PP
154*/
155
69512466
JH
156#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
157# include <inttypes.h>
dd0eed91
JH
158# ifdef INT32_MIN_BROKEN
159# undef INT32_MIN
160# define INT32_MIN (-2147483647-1)
161# endif
162# ifdef INT64_MIN_BROKEN
163# undef INT64_MIN
164# define INT64_MIN (-9223372036854775807LL-1)
165# endif
69512466
JH
166#endif
167
8175356b
JH
168typedef I8TYPE I8;
169typedef U8TYPE U8;
170typedef I16TYPE I16;
171typedef U16TYPE U16;
172typedef I32TYPE I32;
173typedef U32TYPE U32;
16d89be8 174
74b807c7 175#ifdef QUADKIND
8175356b
JH
176typedef I64TYPE I64;
177typedef U64TYPE U64;
16d89be8 178#endif
8175356b 179
d8668976 180#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 181
5ff3f7a4
GS
182/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
183 Please search CHAR_MAX in perl.h for further details. */
184#define U8_MAX UINT8_MAX
185#define U8_MIN UINT8_MIN
186
5ff3f7a4
GS
187#define I16_MAX INT16_MAX
188#define I16_MIN INT16_MIN
189#define U16_MAX UINT16_MAX
190#define U16_MIN UINT16_MIN
191
5ff3f7a4
GS
192#define I32_MAX INT32_MAX
193#define I32_MIN INT32_MIN
0e983133
GS
194#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
195# define U32_MAX UINT32_MAX
196#else
197# define U32_MAX 4294967295U
198#endif
5ff3f7a4
GS
199#define U32_MIN UINT32_MIN
200
201#else
202
5c9fa16e
KA
203/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
204 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
205#define U8_MAX PERL_UCHAR_MAX
206#define U8_MIN PERL_UCHAR_MIN
79072805 207
27d4fb96
PP
208#define I16_MAX PERL_SHORT_MAX
209#define I16_MIN PERL_SHORT_MIN
210#define U16_MAX PERL_USHORT_MAX
211#define U16_MIN PERL_USHORT_MIN
79072805 212
c4f23d77 213#if LONGSIZE > 4
27d4fb96
PP
214# define I32_MAX PERL_INT_MAX
215# define I32_MIN PERL_INT_MIN
216# define U32_MAX PERL_UINT_MAX
217# define U32_MIN PERL_UINT_MIN
79072805 218#else
27d4fb96
PP
219# define I32_MAX PERL_LONG_MAX
220# define I32_MIN PERL_LONG_MIN
221# define U32_MAX PERL_ULONG_MAX
222# define U32_MIN PERL_ULONG_MIN
79072805
LW
223#endif
224
5ff3f7a4
GS
225#endif
226
247cee9f
KW
227/* These C99 typedefs are useful sometimes for, say, loop variables whose
228 * maximum values are small, but for which speed trumps size. If we have a C99
229 * compiler, use that. Otherwise, a plain 'int' should be good enough.
230 *
231 * Restrict these to core for now until we are more certain this is a good
232 * idea. */
233#if defined(PERL_CORE) || defined(PERL_EXT)
234# ifdef I_STDINT
235 typedef int_fast8_t PERL_INT_FAST8_T;
236 typedef uint_fast8_t PERL_UINT_FAST8_T;
237 typedef int_fast16_t PERL_INT_FAST16_T;
238 typedef uint_fast16_t PERL_UINT_FAST16_T;
239# else
240 typedef int PERL_INT_FAST8_T;
241 typedef unsigned int PERL_UINT_FAST8_T;
242 typedef int PERL_INT_FAST16_T;
243 typedef unsigned int PERL_UINT_FAST16_T;
244# endif
245#endif
246
464decb6
KW
247/* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
248 * anyone is grepping for it */
249#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
fc36a67e
PP
250#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
251#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
252
88794300 253/* Unused by core; should be deprecated */
ff68c719 254#define Ctl(ch) ((ch) & 037)
8d063cd8 255
98fce2a4
KW
256#if defined(PERL_CORE) || defined(PERL_EXT)
257# ifndef MIN
258# define MIN(a,b) ((a) < (b) ? (a) : (b))
259# endif
260# ifndef MAX
261# define MAX(a,b) ((a) > (b) ? (a) : (b))
262# endif
263#endif
264
84ff4fa9
KW
265/* Returns a boolean as to whether the input unsigned number is a power of 2
266 * (2**0, 2**1, etc). In other words if it has just a single bit set.
267 * If not, subtracting 1 would leave the uppermost bit set, so the & would
268 * yield non-zero */
269#if defined(PERL_CORE) || defined(PERL_EXT)
011b1419 270# define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
84ff4fa9
KW
271#endif
272
8d9433eb
KW
273/*
274=for apidoc Am|void|__ASSERT_|bool expr
275
276This is a helper macro to avoid preprocessor issues, replaced by nothing
277unless under DEBUGGING, where it expands to an assert of its argument,
278followed by a comma (hence the comma operator). If we just used a straight
279assert(), we would get a comma with nothing before it when not DEBUGGING.
280
281=cut
282
283We also use empty definition under Coverity since the __ASSERT__
284checks often check for things that Really Cannot Happen, and Coverity
285detects that and gets all excited. */
3e94db23
JH
286
287#if defined(DEBUGGING) && !defined(__COVERITY__)
0f092d08
KW
288# define __ASSERT_(statement) assert(statement),
289#else
290# define __ASSERT_(statement)
291#endif
292
3fe05580 293/*
a4ee4fb5 294=head1 SV Manipulation Functions
3fe05580 295
1568d13a
Z
296=for apidoc Ama|SV*|newSVpvs|"literal string" s
297Like C<newSVpvn>, but takes a literal string instead of a
30a15352 298string/length pair.
3fe05580 299
1568d13a
Z
300=for apidoc Ama|SV*|newSVpvs_flags|"literal string" s|U32 flags
301Like C<newSVpvn_flags>, but takes a literal string instead of
30a15352 302a string/length pair.
84bafc02 303
1568d13a
Z
304=for apidoc Ama|SV*|newSVpvs_share|"literal string" s
305Like C<newSVpvn_share>, but takes a literal string instead of
30a15352 306a string/length pair and omits the hash parameter.
3fe05580 307
1568d13a
Z
308=for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string" s|I32 flags
309Like C<sv_catpvn_flags>, but takes a literal string instead
30a15352 310of a string/length pair.
9dcc53ea 311
1568d13a
Z
312=for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string" s
313Like C<sv_catpvn_nomg>, but takes a literal string instead of
0c395ea5 314a string/length pair.
9dcc53ea 315
1568d13a
Z
316=for apidoc Am|void|sv_catpvs|SV* sv|"literal string" s
317Like C<sv_catpvn>, but takes a literal string instead of a
0c395ea5 318string/length pair.
3fe05580 319
1568d13a
Z
320=for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string" s
321Like C<sv_catpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
322string/length pair.
323
1568d13a
Z
324=for apidoc Am|void|sv_setpvs|SV* sv|"literal string" s
325Like C<sv_setpvn>, but takes a literal string instead of a
0c395ea5 326string/length pair.
3fe05580 327
1568d13a
Z
328=for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string" s
329Like C<sv_setpvn_mg>, but takes a literal string instead of a
9dcc53ea
Z
330string/length pair.
331
78342678 332=for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string" s
1568d13a 333Like C<sv_setref_pvn>, but takes a literal string instead of
0c395ea5 334a string/length pair.
9dcc53ea 335
3fe05580
MHM
336=head1 Memory Management
337
1568d13a
Z
338=for apidoc Ama|char*|savepvs|"literal string" s
339Like C<savepvn>, but takes a literal string instead of a
30a15352 340string/length pair.
3fe05580 341
1568d13a 342=for apidoc Ama|char*|savesharedpvs|"literal string" s
9dcc53ea
Z
343A version of C<savepvs()> which allocates the duplicate string in memory
344which is shared between threads.
345
3fe05580
MHM
346=head1 GV Functions
347
1568d13a
Z
348=for apidoc Am|HV*|gv_stashpvs|"literal string" name|I32 create
349Like C<gv_stashpvn>, but takes a literal string instead of a
0c395ea5 350string/length pair.
3fe05580
MHM
351
352=head1 Hash Manipulation Functions
353
1568d13a
Z
354=for apidoc Am|SV**|hv_fetchs|HV* tb|"literal string" key|I32 lval
355Like C<hv_fetch>, but takes a literal string instead of a
0c395ea5 356string/length pair.
3fe05580 357
d106cb46 358=for apidoc Am|SV**|hv_stores|HV* tb|"literal string" key|SV* val
1568d13a 359Like C<hv_store>, but takes a literal string instead of a
0c395ea5 360string/length pair
3fe05580
MHM
361and omits the hash parameter.
362
510966aa
Z
363=head1 Lexer interface
364
1568d13a 365=for apidoc Amx|void|lex_stuff_pvs|"literal string" pv|U32 flags
510966aa 366
1568d13a 367Like L</lex_stuff_pvn>, but takes a literal string instead of
0c395ea5 368a string/length pair.
510966aa 369
3fe05580
MHM
370=cut
371*/
372
926bb54c
KW
373/* concatenating with "" ensures that only literal strings are accepted as
374 * argument */
26bfd924 375#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
2efa8cc7 376
926bb54c
KW
377/* note that STR_WITH_LEN() can't be used as argument to macros or functions
378 * that under some configurations might be macros, which means that it requires
379 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
ba3a79e7
GA
380 */
381
382/* STR_WITH_LEN() shortcuts */
383#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
384#define newSVpvs_flags(str,flags) \
385 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 386#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
387#define sv_catpvs_flags(sv, str, flags) \
388 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
389#define sv_catpvs_nomg(sv, str) \
390 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
391#define sv_catpvs(sv, str) \
392 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
393#define sv_catpvs_mg(sv, str) \
394 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 395#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
396#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
397#define sv_setref_pvs(rv, classname, str) \
398 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 399#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
400#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
401#define gv_stashpvs(str, create) \
402 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
403#define gv_fetchpvs(namebeg, add, sv_type) \
404 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
405#define gv_fetchpvn(namebeg, len, add, sv_type) \
406 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
407#define sv_catxmlpvs(dsv, str, utf8) \
408 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 409
ba3a79e7 410
510966aa
Z
411#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
412
b96d8cd9
NC
413#define get_cvs(str, flags) \
414 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 415
954c1994 416/*
ccfc67b7
JH
417=head1 Miscellaneous Functions
418
954c1994 419=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
420Test two C<NUL>-terminated strings to see if they are different. Returns true
421or false.
954c1994
GS
422
423=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
424Test two C<NUL>-terminated strings to see if they are equal. Returns true or
425false.
954c1994
GS
426
427=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
428Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
429second, C<s2>. Returns true or false.
954c1994
GS
430
431=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
432Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
433equal to the second, C<s2>. Returns true or false.
954c1994
GS
434
435=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
436Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
437the second, C<s2>. Returns true or false.
954c1994
GS
438
439=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
440Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
441or equal to the second, C<s2>. Returns true or false.
954c1994
GS
442
443=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
444Test two C<NUL>-terminated strings to see if they are different. The C<len>
445parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
446wrapper for C<strncmp>).
447
448=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
449Test two C<NUL>-terminated strings to see if they are equal. The C<len>
450parameter indicates the number of bytes to compare. Returns true or false. (A
451wrapper for C<strncmp>).
954c1994 452
bd18bd40
KW
453=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
454Test two buffers (which may contain embedded C<NUL> characters, to see if they
455are equal. The C<len> parameter indicates the number of bytes to compare.
456Returns zero if equal, or non-zero if non-equal.
457
2d8eeddb
KW
458=for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"literal string" s2
459Like L</memEQ>, but the second string is a literal enclosed in double quotes,
460C<l1> gives the number of bytes in C<s1>.
461Returns zero if equal, or non-zero if non-equal.
462
bd18bd40
KW
463=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
464Test two buffers (which may contain embedded C<NUL> characters, to see if they
465are not equal. The C<len> parameter indicates the number of bytes to compare.
466Returns zero if non-equal, or non-zero if equal.
467
2d8eeddb
KW
468=for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"literal string" s2
469Like L</memNE>, but the second string is a literal enclosed in double quotes,
470C<l1> gives the number of bytes in C<s1>.
471Returns zero if non-equal, or zero if non-equal.
472
954c1994 473=cut
fc169e00
KW
474
475New macros should use the following conventions for their names (which are
476based on the underlying C library functions):
477
478 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
479
480 Each has two main parameters, string-like operands that are compared
481 against each other, as specified by the macro name. Some macros may
482 additionally have one or potentially even two length parameters. If a length
483 parameter applies to both string parameters, it will be positioned third;
484 otherwise any length parameter immediately follows the string parameter it
485 applies to.
486
487 If the prefix to the name is 'str', the string parameter is a pointer to a C
488 language string. Such a string does not contain embedded NUL bytes; its
489 length may be unknown, but can be calculated by C<strlen()>, since it is
490 terminated by a NUL, which isn't included in its length.
491
492 The optional 'n' following 'str' means that that there is a third parameter,
493 giving the maximum number of bytes to look at in each string. Even if both
494 strings are longer than the length parameter, those extra bytes will be
495 unexamined.
496
497 The 's' suffix means that the 2nd byte string parameter is a literal C
498 double-quoted string. Its length will automatically be calculated by the
499 macro, so no length parameter will ever be needed for it.
500
501 If the prefix is 'mem', the string parameters don't have to be C strings;
502 they may contain embedded NUL bytes, do not necessarily have a terminating
503 NUL, and their lengths can be known only through other means, which in
504 practice are additional parameter(s) passed to the function. All 'mem'
505 functions have at least one length parameter. Barring any 'l' or 's' suffix,
506 there is a single length parameter, in position 3, which applies to both
507 string parameters. The 's' suffix means, as described above, that the 2nd
508 string is a literal double-quoted C string (hence its length is calculated by
509 the macro, and the length parameter to the function applies just to the first
510 string parameter, and hence is positioned just after it). An 'l' suffix
511 means that the 2nd string parameter has its own length parameter, and the
512 signature will look like memFOOl(s1, l1, s2, l2).
513
514 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
515 substring of the 1st string. 'P' if present indicates that the substring
516 must be a "proper" one in tha mathematical sense that the first one must be
517 strictly larger than the 2nd.
518
954c1994
GS
519*/
520
62946e08 521
75400963
KW
522#define strNE(s1,s2) (strcmp(s1,s2) != 0)
523#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
524#define strLT(s1,s2) (strcmp(s1,s2) < 0)
525#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
526#define strGT(s1,s2) (strcmp(s1,s2) > 0)
527#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 528
75400963
KW
529#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
530#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 531
9d3980bc
KW
532#define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
533#define memNE(s1,s2,l) (! memEQ(s1,s2,l))
36477c24 534
085b7534 535/* memEQ and memNE where second comparand is a string constant */
568a785a 536#define memEQs(s1, l, s2) \
777fa2cb 537 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 538#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 539
fdbb9a7c
KW
540/* Keep these private until we decide it was a good idea */
541#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
542
543#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
544
bdb7e3f0 545#define memBEGINs(s1, l, s2) \
30a6480c 546 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 547 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158 548#define memBEGINPs(s1, l, s2) \
30a6480c 549 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
de627158 550 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0 551#define memENDs(s1, l, s2) \
30a6480c 552 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
bdb7e3f0 553 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424 554#define memENDPs(s1, l, s2) \
30a6480c 555 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
b80f8424 556 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 557#endif /* End of making macros private */
bdb7e3f0 558
062b6850
KW
559#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
560#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
561#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
562#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
563
bbce6d69
PP
564/*
565 * Character classes.
566 *
567 * Unfortunately, the introduction of locales means that we
568 * can't trust isupper(), etc. to tell the truth. And when
569 * it comes to /\w+/ with tainting enabled, we *must* be able
570 * to trust our character classes.
571 *
572 * Therefore, the default tests in the text of Perl will be
573 * independent of locale. Any code that wants to depend on
574 * the current locale will use the tests that begin with "lc".
575 */
576
2304df62
AD
577#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
578# ifndef CTYPE256
579# define CTYPE256
580# endif
581#endif
582
954c1994 583/*
ccfc67b7 584
dcccc8ff 585=head1 Character classification
243effed
KW
586This section is about functions (really macros) that classify characters
587into types, such as punctuation versus alphabetic, etc. Most of these are
588analogous to regular expression character classes. (See
589L<perlrecharclass/POSIX Character Classes>.) There are several variants for
590each class. (Not all macros have all variants; each item below lists the
591ones valid for it.) None are affected by C<use bytes>, and only the ones
592with C<LC> in the name are affected by the current locale.
593
d713f9d9
KW
594The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
595treating it as a code point, and returns a boolean as to whether or not the
596character represented by it is (or on non-ASCII platforms, corresponds to) an
6aff1f14
KW
597ASCII character in the named class based on platform, Unicode, and Perl rules.
598If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 599
c98722a4 600Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
601with no suffix C<"_A">. This variant is used to emphasize by its name that
602only ASCII-range characters can return TRUE.
4b9734bf 603
d60679e1 604Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
605onto the platform. That is, the code points that are ASCII are unaffected,
606since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
607as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
608true when called with the code point 0xDF, which is a word character in both
4650c663 609ASCII and EBCDIC (though it represents different characters in each).
d713f9d9
KW
610If the input is a number that doesn't fit in an octet, FALSE is returned.
611(Perl's documentation uses a colloquial definition of Latin-1, to include all
612code points below 256.)
243effed 613
d713f9d9
KW
614Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
615inputs below 256, but if the code point is larger than 255, Unicode rules are
616used to determine if it is in the character class. For example,
d0da05db 617C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 618WITH MACRON in Unicode, and is a word character.
243effed 619
c98722a4 620Variant C<isI<FOO>_utf8_safe> is like C<isI<FOO>_uvchr>, but is used for UTF-8
d713f9d9
KW
621encoded strings. Each call classifies the first character of the string. This
622variant takes two parameters. The first, C<p>, is a
da8c1a98
KW
623pointer to the first byte of the character to be classified. (Recall that it
624may take more than one byte to represent a character in UTF-8 strings.) The
625second parameter, C<e>, points to anywhere in the string beyond the first
626character, up to one byte past the end of the entire string. The suffix
627C<_safe> in the function's name indicates that it will not attempt to read
628beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
629is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
630character is malformed in some way, the program may croak, or the function may
631return FALSE, at the discretion of the implementation, and subject to change in
632future releases.
633
c98722a4 634Variant C<isI<FOO>_utf8> is like C<isI<FOO>_utf8_safe>, but takes just a single
da8c1a98 635parameter, C<p>, which has the same meaning as the corresponding parameter does
c98722a4 636in C<isI<FOO>_utf8_safe>. The function therefore can't check if it is reading
8b4b30c5 637beyond the end of the string. Starting in Perl v5.32, it will take a second
c98722a4 638parameter, becoming a synonym for C<isI<FOO>_utf8_safe>. At that time every
34aeb2e9 639program that uses it will have to be changed to successfully compile. In the
c98722a4 640meantime, the first runtime call to C<isI<FOO>_utf8> from each call point in the
34aeb2e9 641program will raise a deprecation warning, enabled by default. You can convert
c98722a4 642your program now to use C<isI<FOO>_utf8_safe>, and avoid the warnings, and get an
8b4b30c5 643extra measure of protection, or you can wait until v5.32, when you'll be forced
34aeb2e9 644to add the C<e> parameter.
243effed 645
d713f9d9
KW
646Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
647but the result is based on the current locale, which is what C<LC> in the name
648stands for. If Perl can determine that the current locale is a UTF-8 locale,
649it uses the published Unicode rules; otherwise, it uses the C library function
650that gives the named classification. For example, C<isDIGIT_LC()> when not in
651a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
652returned if the input won't fit into an octet. On some platforms where the C
653library function is known to be defective, Perl changes its result to follow
654the POSIX standard's rules.
243effed 655
d713f9d9
KW
656Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
657than 256, but for larger ones it returns the Unicode classification of the code
658point.
243effed 659
c98722a4 660Variant C<isI<FOO>_LC_utf8_safe> is like C<isI<FOO>_LC_uvchr>, but is used for UTF-8
d713f9d9
KW
661encoded strings. Each call classifies the first character of the string. This
662variant takes two parameters. The first, C<p>, is a pointer to the first byte
663of the character to be classified. (Recall that it may take more than one byte
664to represent a character in UTF-8 strings.) The second parameter, C<e>,
665points to anywhere in the string beyond the first character, up to one byte
666past the end of the entire string. The suffix C<_safe> in the function's name
667indicates that it will not attempt to read beyond S<C<e - 1>>, provided that
668the constraint S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING>
669builds). If the UTF-8 for the input character is malformed in some way, the
670program may croak, or the function may return FALSE, at the discretion of the
671implementation, and subject to change in future releases.
da8c1a98 672
c98722a4 673Variant C<isI<FOO>_LC_utf8> is like C<isI<FOO>_LC_utf8_safe>, but takes just a single
da8c1a98 674parameter, C<p>, which has the same meaning as the corresponding parameter does
c98722a4 675in C<isI<FOO>_LC_utf8_safe>. The function therefore can't check if it is reading
8b4b30c5 676beyond the end of the string. Starting in Perl v5.32, it will take a second
c98722a4 677parameter, becoming a synonym for C<isI<FOO>_LC_utf8_safe>. At that time every
34aeb2e9 678program that uses it will have to be changed to successfully compile. In the
c98722a4 679meantime, the first runtime call to C<isI<FOO>_LC_utf8> from each call point in
34aeb2e9 680the program will raise a deprecation warning, enabled by default. You can
c98722a4 681convert your program now to use C<isI<FOO>_LC_utf8_safe>, and avoid the warnings,
8b4b30c5 682and get an extra measure of protection, or you can wait until v5.32, when
34aeb2e9 683you'll be forced to add the C<e> parameter.
ccfc67b7 684
d713f9d9
KW
685=for apidoc Am|bool|isALPHA|int ch
686Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
687analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
688See the L<top of this section|/Character classification> for an explanation of
689variants
da8c1a98
KW
690C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8_safe>,
691C<isALPHA_LC>, C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8_safe>.
8a58bdcf 692
f16858ed
KW
693=cut
694
695Here and below, we add the protoypes of these macros for downstream programs
696that would be interested in them, such as Devel::PPPort
697
698=for apidoc Amh|bool|isALPHA_A|int ch
699=for apidoc Amh|bool|isALPHA_L1|int ch
700=for apidoc Amh|bool|isALPHA_uvchr|int ch
701=for apidoc Amh|bool|isALPHA_utf8_safe|U8 * s|U8 * end
702=for apidoc Amh|bool|isALPHA_utf8|U8 * s
703=for apidoc Amh|bool|isALPHA_LC|int ch
704=for apidoc Amh|bool|isALPHA_LC_uvchr|int ch
705=for apidoc Amh|bool|isALPHA_LC_utf8_safe|U8 * s| U8 *end
706
d713f9d9
KW
707=for apidoc Am|bool|isALPHANUMERIC|int ch
708Returns a boolean indicating whether the specified character is one of
709C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
710See the L<top of this section|/Character classification> for an explanation of
711variants
d0da05db 712C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
da8c1a98
KW
713C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>,
714and C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 715
255b632a
KW
716A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
717this corresponds to the C language alphanumeric definition). Also
718there are the variants
719C<isALNUMC_A>, C<isALNUMC_L1>
720C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
721
f16858ed
KW
722=for apidoc Amh|bool|isALPHANUMERIC_A|int ch
723=for apidoc Amh|bool|isALPHANUMERIC_L1|int ch
724=for apidoc Amh|bool|isALPHANUMERIC_uvchr|int ch
725=for apidoc Amh|bool|isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
726=for apidoc Amh|bool|isALPHANUMERIC_utf8|U8 * s
727=for apidoc Amh|bool|isALPHANUMERIC_LC|int ch
728=for apidoc Amh|bool|isALPHANUMERIC_LC_uvchr|int ch
729=for apidoc Amh|bool|isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
730=for apidoc Amh|bool|isALNUMC|int ch
731=for apidoc Amh|bool|isALNUMC_A|int ch
732=for apidoc Amh|bool|isALNUMC_L1|int ch
733=for apidoc Amh|bool|isALNUMC_LC|int ch
734=for apidoc Amh|bool|isALNUMC_LC_uvchr|int ch
735
d713f9d9 736=for apidoc Am|bool|isASCII|int ch
8a58bdcf 737Returns a boolean indicating whether the specified character is one of the 128
243effed 738characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 739On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
740character corresponds to an ASCII character. Variants C<isASCII_A()> and
741C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
742See the L<top of this section|/Character classification> for an explanation of
743variants
da8c1a98
KW
744C<isASCII_uvchr>, C<isASCII_utf8_safe>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
745C<isASCII_LC_utf8_safe>. Note, however, that some platforms do not have the C
243effed
KW
746library routine C<isascii()>. In these cases, the variants whose names contain
747C<LC> are the same as the corresponding ones without.
748
f16858ed
KW
749=for apidoc Amh|bool|isASCII_A|int ch
750=for apidoc Amh|bool|isASCII_L1|int ch
751=for apidoc Amh|bool|isASCII_uvchr|int ch
752=for apidoc Amh|bool|isASCII_utf8_safe|U8 * s|U8 * end
753=for apidoc Amh|bool|isASCII_utf8|U8 * s
754=for apidoc Amh|bool|isASCII_LC|int ch
755=for apidoc Amh|bool|isASCII_LC_uvchr|int ch
756=for apidoc Amh|bool|isASCII_LC_utf8_safe|U8 * s| U8 *end
757
d98532ea
KW
758Also note, that because all ASCII characters are UTF-8 invariant (meaning they
759have the exact same representation (always a single byte) whether encoded in
760UTF-8 or not), C<isASCII> will give the correct results when called with any
da8c1a98
KW
761byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8_safe>
762will work properly on any string encoded or not in UTF-8.
d98532ea 763
243effed
KW
764=for apidoc Am|bool|isBLANK|char ch
765Returns a boolean indicating whether the specified character is a
6aff1f14 766character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
767See the L<top of this section|/Character classification> for an explanation of
768variants
da8c1a98
KW
769C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8_safe>,
770C<isBLANK_LC>, C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8_safe>. Note,
771however, that some platforms do not have the C library routine
772C<isblank()>. In these cases, the variants whose names contain C<LC> are
773the same as the corresponding ones without.
243effed 774
f16858ed
KW
775=for apidoc Amh|bool|isBLANK_A|int ch
776=for apidoc Amh|bool|isBLANK_L1|int ch
777=for apidoc Amh|bool|isBLANK_uvchr|int ch
778=for apidoc Amh|bool|isBLANK_utf8_safe|U8 * s|U8 * end
779=for apidoc Amh|bool|isBLANK_utf8|U8 * s
780=for apidoc Amh|bool|isBLANK_LC|int ch
781=for apidoc Amh|bool|isBLANK_LC_uvchr|int ch
782=for apidoc Amh|bool|isBLANK_LC_utf8_safe|U8 * s| U8 *end
783
243effed
KW
784=for apidoc Am|bool|isCNTRL|char ch
785Returns a boolean indicating whether the specified character is a
6aff1f14 786control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
787See the L<top of this section|/Character classification> for an explanation of
788variants
da8c1a98
KW
789C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8_safe>,
790C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8_safe> On EBCDIC
791platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 792
f16858ed
KW
793=for apidoc Amh|bool|isCNTRL_A|int ch
794=for apidoc Amh|bool|isCNTRL_L1|int ch
795=for apidoc Amh|bool|isCNTRL_uvchr|int ch
796=for apidoc Amh|bool|isCNTRL_utf8_safe|U8 * s|U8 * end
797=for apidoc Amh|bool|isCNTRL_utf8|U8 * s
798=for apidoc Amh|bool|isCNTRL_LC|int ch
799=for apidoc Amh|bool|isCNTRL_LC_uvchr|int ch
800=for apidoc Amh|bool|isCNTRL_LC_utf8_safe|U8 * s| U8 *end
801
954c1994 802=for apidoc Am|bool|isDIGIT|char ch
2787a470 803Returns a boolean indicating whether the specified character is a
6aff1f14 804digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 805Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
806See the L<top of this section|/Character classification> for an explanation of
807variants
da8c1a98
KW
808C<isDIGIT_uvchr>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
809C<isDIGIT_LC_utf8_safe>.
243effed 810
f16858ed
KW
811=for apidoc Amh|bool|isDIGIT_A|int ch
812=for apidoc Amh|bool|isDIGIT_L1|int ch
813=for apidoc Amh|bool|isDIGIT_uvchr|int ch
814=for apidoc Amh|bool|isDIGIT_utf8_safe|U8 * s|U8 * end
815=for apidoc Amh|bool|isDIGIT_utf8|U8 * s
816=for apidoc Amh|bool|isDIGIT_LC|int ch
817=for apidoc Amh|bool|isDIGIT_LC_uvchr|int ch
818=for apidoc Amh|bool|isDIGIT_LC_utf8_safe|U8 * s| U8 *end
819
243effed
KW
820=for apidoc Am|bool|isGRAPH|char ch
821Returns a boolean indicating whether the specified character is a
6aff1f14 822graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 823See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
824variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8_safe>,
825C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8_safe>.
954c1994 826
f16858ed
KW
827=for apidoc Amh|bool|isGRAPH_A|int ch
828=for apidoc Amh|bool|isGRAPH_L1|int ch
829=for apidoc Amh|bool|isGRAPH_uvchr|int ch
830=for apidoc Amh|bool|isGRAPH_utf8_safe|U8 * s|U8 * end
831=for apidoc Amh|bool|isGRAPH_utf8|U8 * s
832=for apidoc Amh|bool|isGRAPH_LC|int ch
833=for apidoc Amh|bool|isGRAPH_LC_uvchr|int ch
834=for apidoc Amh|bool|isGRAPH_LC_utf8_safe|U8 * s| U8 *end
835
0c82b6df 836=for apidoc Am|bool|isLOWER|char ch
2787a470 837Returns a boolean indicating whether the specified character is a
6aff1f14 838lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
839See the L<top of this section|/Character classification> for an explanation of
840variants
da8c1a98
KW
841C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8_safe>,
842C<isLOWER_LC>, C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8_safe>.
0c82b6df 843
f16858ed
KW
844=for apidoc Amh|bool|isLOWER_A|int ch
845=for apidoc Amh|bool|isLOWER_L1|int ch
846=for apidoc Amh|bool|isLOWER_uvchr|int ch
847=for apidoc Amh|bool|isLOWER_utf8_safe|U8 * s|U8 * end
848=for apidoc Amh|bool|isLOWER_utf8|U8 * s
849=for apidoc Amh|bool|isLOWER_LC|int ch
850=for apidoc Amh|bool|isLOWER_LC_uvchr|int ch
851=for apidoc Amh|bool|isLOWER_LC_utf8_safe|U8 * s| U8 *end
852
c99e91e9 853=for apidoc Am|bool|isOCTAL|char ch
2787a470 854Returns a boolean indicating whether the specified character is an
6aff1f14 855octal digit, [0-7].
243effed
KW
856The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
857C<isOCTAL>.
858
f16858ed
KW
859=for apidoc Amh|bool|isOCTAL_A|int ch
860=for apidoc Amh|bool|isOCTAL_L1|int ch
861
243effed
KW
862=for apidoc Am|bool|isPUNCT|char ch
863Returns a boolean indicating whether the specified character is a
6aff1f14
KW
864punctuation character, analogous to C<m/[[:punct:]]/>.
865Note that the definition of what is punctuation isn't as
243effed
KW
866straightforward as one might desire. See L<perlrecharclass/POSIX Character
867Classes> for details.
dcccc8ff 868See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
869variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8_safe>,
870C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8_safe>.
c99e91e9 871
f16858ed
KW
872=for apidoc Amh|bool|isPUNCT_A|int ch
873=for apidoc Amh|bool|isPUNCT_L1|int ch
874=for apidoc Amh|bool|isPUNCT_uvchr|int ch
875=for apidoc Amh|bool|isPUNCT_utf8_safe|U8 * s|U8 * end
876=for apidoc Amh|bool|isPUNCT_utf8|U8 * s
877=for apidoc Amh|bool|isPUNCT_LC|int ch
878=for apidoc Amh|bool|isPUNCT_LC_uvchr|int ch
879=for apidoc Amh|bool|isPUNCT_LC_utf8_safe|U8 * s| U8 *end
880
0c82b6df 881=for apidoc Am|bool|isSPACE|char ch
2787a470 882Returns a boolean indicating whether the specified character is a
6aff1f14 883whitespace character. This is analogous
398d098a 884to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 885this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
886locale forms of this macro (the ones with C<LC> in their names) matched
887precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
888in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
889(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
890See the L<top of this section|/Character classification> for an explanation of
891variants
da8c1a98
KW
892C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8_safe>,
893C<isSPACE_LC>, C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8_safe>.
0c82b6df 894
f16858ed
KW
895=for apidoc Amh|bool|isSPACE_A|int ch
896=for apidoc Amh|bool|isSPACE_L1|int ch
897=for apidoc Amh|bool|isSPACE_uvchr|int ch
898=for apidoc Amh|bool|isSPACE_utf8_safe|U8 * s|U8 * end
899=for apidoc Amh|bool|isSPACE_utf8|U8 * s
900=for apidoc Amh|bool|isSPACE_LC|int ch
901=for apidoc Amh|bool|isSPACE_LC_uvchr|int ch
902=for apidoc Amh|bool|isSPACE_LC_utf8_safe|U8 * s| U8 *end
903
398d098a
KW
904=for apidoc Am|bool|isPSXSPC|char ch
905(short for Posix Space)
779cf272
KW
906Starting in 5.18, this is identical in all its forms to the
907corresponding C<isSPACE()> macros.
398d098a
KW
908The locale forms of this macro are identical to their corresponding
909C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
910non-locale forms differ from their C<isSPACE()> forms only in that the
911C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
912Otherwise they are identical. Thus this macro is analogous to what
913C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 914See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
915variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8_safe>,
916C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 917
f16858ed
KW
918=for apidoc Amh|bool|isPSXSPC_A|int ch
919=for apidoc Amh|bool|isPSXSPC_L1|int ch
920=for apidoc Amh|bool|isPSXSPC_uvchr|int ch
921=for apidoc Amh|bool|isPSXSPC_utf8_safe|U8 * s|U8 * end
922=for apidoc Amh|bool|isPSXSPC_utf8|U8 * s
923=for apidoc Amh|bool|isPSXSPC_LC|int ch
924=for apidoc Amh|bool|isPSXSPC_LC_uvchr|int ch
925=for apidoc Amh|bool|isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
926
954c1994 927=for apidoc Am|bool|isUPPER|char ch
2787a470 928Returns a boolean indicating whether the specified character is an
6aff1f14 929uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 930See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
931variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8_safe>,
932C<isUPPER_LC>, C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8_safe>.
954c1994 933
f16858ed
KW
934=for apidoc Amh|bool|isUPPER_A|int ch
935=for apidoc Amh|bool|isUPPER_L1|int ch
936=for apidoc Amh|bool|isUPPER_uvchr|int ch
937=for apidoc Amh|bool|isUPPER_utf8_safe|U8 * s|U8 * end
938=for apidoc Amh|bool|isUPPER_utf8|U8 * s
939=for apidoc Amh|bool|isUPPER_LC|int ch
940=for apidoc Amh|bool|isUPPER_LC_uvchr|int ch
941=for apidoc Amh|bool|isUPPER_LC_utf8_safe|U8 * s| U8 *end
942
243effed 943=for apidoc Am|bool|isPRINT|char ch
8eea39dd 944Returns a boolean indicating whether the specified character is a
6aff1f14 945printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
946See the L<top of this section|/Character classification> for an explanation of
947variants
da8c1a98
KW
948C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8_safe>,
949C<isPRINT_LC>, C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8_safe>.
243effed 950
f16858ed
KW
951=for apidoc Amh|bool|isPRINT_A|int ch
952=for apidoc Amh|bool|isPRINT_L1|int ch
953=for apidoc Amh|bool|isPRINT_uvchr|int ch
954=for apidoc Amh|bool|isPRINT_utf8_safe|U8 * s|U8 * end
955=for apidoc Amh|bool|isPRINT_utf8|U8 * s
956=for apidoc Amh|bool|isPRINT_LC|int ch
957=for apidoc Amh|bool|isPRINT_LC_uvchr|int ch
958=for apidoc Amh|bool|isPRINT_LC_utf8_safe|U8 * s| U8 *end
959
243effed
KW
960=for apidoc Am|bool|isWORDCHAR|char ch
961Returns a boolean indicating whether the specified character is a character
962that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
963in a regular expression. A word character is an alphabetic character, a
964decimal digit, a connecting punctuation character (such as an underscore), or
965a "mark" character that attaches to one of those (like some sort of accent).
966C<isALNUM()> is a synonym provided for backward compatibility, even though a
967word character includes more than the standard C language meaning of
968alphanumeric.
dcccc8ff 969See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
970variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and
971C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and
972C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
973include the platform's native underscore.
8a58bdcf 974
f16858ed
KW
975=for apidoc Amh|bool|isWORDCHAR_A|int ch
976=for apidoc Amh|bool|isWORDCHAR_L1|int ch
977=for apidoc Amh|bool|isWORDCHAR_uvchr|int ch
978=for apidoc Amh|bool|isWORDCHAR_utf8_safe|U8 * s|U8 * end
979=for apidoc Amh|bool|isWORDCHAR_utf8|U8 * s
980=for apidoc Amh|bool|isWORDCHAR_LC|int ch
981=for apidoc Amh|bool|isWORDCHAR_LC_uvchr|int ch
982=for apidoc Amh|bool|isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
983=for apidoc Amh|bool|isALNUM|int ch
984=for apidoc Amh|bool|isALNUM_A|int ch
985=for apidoc Amh|bool|isALNUM_LC|int ch
986=for apidoc Amh|bool|isALNUM_LC_uvchr|int ch
987
8a58bdcf
KW
988=for apidoc Am|bool|isXDIGIT|char ch
989Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
990digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
991and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
992See the L<top of this section|/Character classification> for an explanation of
993variants
da8c1a98
KW
994C<isXDIGIT_uvchr>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>,
995and C<isXDIGIT_LC_utf8_safe>.
243effed 996
f16858ed
KW
997=for apidoc Amh|bool|isXDIGIT_A|int ch
998=for apidoc Amh|bool|isXDIGIT_L1|int ch
999=for apidoc Amh|bool|isXDIGIT_uvchr|int ch
1000=for apidoc Amh|bool|isXDIGIT_utf8_safe|U8 * s|U8 * end
1001=for apidoc Amh|bool|isXDIGIT_utf8|U8 * s
1002=for apidoc Amh|bool|isXDIGIT_LC|int ch
1003=for apidoc Amh|bool|isXDIGIT_LC_uvchr|int ch
1004=for apidoc Amh|bool|isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1005
3c3ecf18
KW
1006=for apidoc Am|bool|isIDFIRST|char ch
1007Returns a boolean indicating whether the specified character can be the first
1008character of an identifier. This is very close to, but not quite the same as
1009the official Unicode property C<XID_Start>. The difference is that this
1010returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
1011See the L<top of this section|/Character classification> for an explanation of
1012variants
da8c1a98
KW
1013C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8_safe>,
1014C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18 1015
f16858ed
KW
1016=for apidoc Amh|bool|isIDFIRST_A|int ch
1017=for apidoc Amh|bool|isIDFIRST_L1|int ch
1018=for apidoc Amh|bool|isIDFIRST_uvchr|int ch
1019=for apidoc Amh|bool|isIDFIRST_utf8_safe|U8 * s|U8 * end
1020=for apidoc Amh|bool|isIDFIRST_utf8|U8 * s
1021=for apidoc Amh|bool|isIDFIRST_LC|int ch
1022=for apidoc Amh|bool|isIDFIRST_LC_uvchr|int ch
1023=for apidoc Amh|bool|isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1024
3c3ecf18
KW
1025=for apidoc Am|bool|isIDCONT|char ch
1026Returns a boolean indicating whether the specified character can be the
1027second or succeeding character of an identifier. This is very close to, but
1028not quite the same as the official Unicode property C<XID_Continue>. The
1029difference is that this returns true only if the input character also matches
dcccc8ff
KW
1030L</isWORDCHAR>. See the L<top of this section|/Character classification> for
1031an
d0da05db 1032explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
da8c1a98
KW
1033C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
1034C<isIDCONT_LC_utf8_safe>.
3c3ecf18 1035
f16858ed
KW
1036=for apidoc Amh|bool|isIDCONT_A|int ch
1037=for apidoc Amh|bool|isIDCONT_L1|int ch
1038=for apidoc Amh|bool|isIDCONT_uvchr|int ch
1039=for apidoc Amh|bool|isIDCONT_utf8_safe|U8 * s|U8 * end
1040=for apidoc Amh|bool|isIDCONT_utf8|U8 * s
1041=for apidoc Amh|bool|isIDCONT_LC|int ch
1042=for apidoc Amh|bool|isIDCONT_LC_uvchr|int ch
1043=for apidoc Amh|bool|isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1044
243effed 1045=head1 Miscellaneous Functions
8eea39dd 1046
95a59cab 1047=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 1048Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
1049Behaviour is only well defined when isXDIGIT(*str) is true.
1050
e7c1e6c1 1051=head1 Character case changing
21da7284
KW
1052Perl uses "full" Unicode case mappings. This means that converting a single
1053character to another case may result in a sequence of more than one character.
1054For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1055character sequence C<SS>. This presents some complications The lowercase of
1056all characters in the range 0..255 is a single character, and thus
1057C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1058return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1059an API that does allow every possible legal result to be returned.) Likewise
1060no other function that is crippled by not being able to give the correct
1061results for the full range of possible inputs has been implemented here.
e7c1e6c1 1062
d713f9d9 1063=for apidoc Am|U8|toUPPER|int ch
1f607577
KW
1064Converts the specified character to uppercase. If the input is anything but an
1065ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 1066C<toUPPER_A> is equivalent.
954c1994 1067
d0da05db
KW
1068=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1069Converts the code point C<cp> to its uppercase version, and
1070stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1071point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1072that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1073bytes since the uppercase version may be longer than the original character.
1074
1075The first code point of the uppercased version is returned
21da7284
KW
1076(but note, as explained at L<the top of this section|/Character case
1077changing>, that there may be more.)
1f607577 1078
a239b1e2
KW
1079=for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1080Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1081extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
1082stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1083that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1084bytes since the uppercase version may be longer than the original character.
1085
1086The first code point of the uppercased version is returned
21da7284
KW
1087(but note, as explained at L<the top of this section|/Character case
1088changing>, that there may be more).
1f607577 1089
a239b1e2
KW
1090The suffix C<_safe> in the function's name indicates that it will not attempt
1091to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1092true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1093input character is malformed in some way, the program may croak, or the
1094function may return the REPLACEMENT CHARACTER, at the discretion of the
1095implementation, and subject to change in future releases.
1096
1097=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
1098This is like C<L</toUPPER_utf8_safe>>, but doesn't have the C<e>
1099parameter The function therefore can't check if it is reading
8b4b30c5 1100beyond the end of the string. Starting in Perl v5.32, it will take the C<e>
607313a1
KW
1101parameter, becoming a synonym for C<toUPPER_utf8_safe>. At that time every
1102program that uses it will have to be changed to successfully compile. In the
1103meantime, the first runtime call to C<toUPPER_utf8> from each call point in the
1104program will raise a deprecation warning, enabled by default. You can convert
1105your program now to use C<toUPPER_utf8_safe>, and avoid the warnings, and get an
8b4b30c5 1106extra measure of protection, or you can wait until v5.32, when you'll be forced
607313a1 1107to add the C<e> parameter.
1f607577 1108
25200305
KW
1109=for apidoc Am|U8|toFOLD|U8 ch
1110Converts the specified character to foldcase. If the input is anything but an
1111ASCII uppercase character, that input character itself is returned. Variant
1112C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 1113Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 1114
d0da05db
KW
1115=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1116Converts the code point C<cp> to its foldcase version, and
1117stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1118point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1119that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1120bytes since the foldcase version may be longer than the original character.
1121
1122The first code point of the foldcased version is returned
21da7284
KW
1123(but note, as explained at L<the top of this section|/Character case
1124changing>, that there may be more).
1f607577 1125
a239b1e2
KW
1126=for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1127Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1128extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
1129stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1130that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1131bytes since the foldcase version may be longer than the original character.
1132
1133The first code point of the foldcased version is returned
21da7284
KW
1134(but note, as explained at L<the top of this section|/Character case
1135changing>, that there may be more).
1f607577 1136
a239b1e2
KW
1137The suffix C<_safe> in the function's name indicates that it will not attempt
1138to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1139true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1140input character is malformed in some way, the program may croak, or the
1141function may return the REPLACEMENT CHARACTER, at the discretion of the
1142implementation, and subject to change in future releases.
1143
1144=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
1145This is like C<L</toFOLD_utf8_safe>>, but doesn't have the C<e>
1146parameter The function therefore can't check if it is reading
8b4b30c5 1147beyond the end of the string. Starting in Perl v5.32, it will take the C<e>
607313a1
KW
1148parameter, becoming a synonym for C<toFOLD_utf8_safe>. At that time every
1149program that uses it will have to be changed to successfully compile. In the
1150meantime, the first runtime call to C<toFOLD_utf8> from each call point in the
1151program will raise a deprecation warning, enabled by default. You can convert
1152your program now to use C<toFOLD_utf8_safe>, and avoid the warnings, and get an
8b4b30c5 1153extra measure of protection, or you can wait until v5.32, when you'll be forced
607313a1 1154to add the C<e> parameter.
1f607577
KW
1155
1156=for apidoc Am|U8|toLOWER|U8 ch
1157Converts the specified character to lowercase. If the input is anything but an
1158ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 1159C<toLOWER_A> is equivalent.
954c1994 1160
1f607577 1161=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
1162Converts the specified Latin1 character to lowercase. The results are
1163undefined if the input doesn't fit in a byte.
1f607577
KW
1164
1165=for apidoc Am|U8|toLOWER_LC|U8 ch
1166Converts the specified character to lowercase using the current locale's rules,
1167if possible; otherwise returns the input character itself.
1168
d0da05db
KW
1169=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1170Converts the code point C<cp> to its lowercase version, and
1171stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1172point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1173that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1174bytes since the lowercase version may be longer than the original character.
1175
1176The first code point of the lowercased version is returned
21da7284
KW
1177(but note, as explained at L<the top of this section|/Character case
1178changing>, that there may be more).
1f607577 1179
a239b1e2
KW
1180
1181=for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1182Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1183extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
1184stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1185that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1186bytes since the lowercase version may be longer than the original character.
1187
1188The first code point of the lowercased version is returned
21da7284
KW
1189(but note, as explained at L<the top of this section|/Character case
1190changing>, that there may be more).
1f607577 1191
a239b1e2
KW
1192The suffix C<_safe> in the function's name indicates that it will not attempt
1193to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1194true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1195input character is malformed in some way, the program may croak, or the
1196function may return the REPLACEMENT CHARACTER, at the discretion of the
1197implementation, and subject to change in future releases.
1198
1199=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
1200This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1201parameter The function therefore can't check if it is reading
8b4b30c5 1202beyond the end of the string. Starting in Perl v5.32, it will take the C<e>
607313a1
KW
1203parameter, becoming a synonym for C<toLOWER_utf8_safe>. At that time every
1204program that uses it will have to be changed to successfully compile. In the
1205meantime, the first runtime call to C<toLOWER_utf8> from each call point in the
1206program will raise a deprecation warning, enabled by default. You can convert
1207your program now to use C<toLOWER_utf8_safe>, and avoid the warnings, and get an
8b4b30c5 1208extra measure of protection, or you can wait until v5.32, when you'll be forced
607313a1 1209to add the C<e> parameter.
1f607577 1210
25200305
KW
1211=for apidoc Am|U8|toTITLE|U8 ch
1212Converts the specified character to titlecase. If the input is anything but an
1213ASCII lowercase character, that input character itself is returned. Variant
b7d90381 1214C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 1215range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 1216not a concept used in locale handling, so there is no functionality for that.)
25200305 1217
d0da05db
KW
1218=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1219Converts the code point C<cp> to its titlecase version, and
1220stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1221point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
1222that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1223bytes since the titlecase version may be longer than the original character.
1224
1225The first code point of the titlecased version is returned
21da7284
KW
1226(but note, as explained at L<the top of this section|/Character case
1227changing>, that there may be more).
1f607577 1228
a239b1e2
KW
1229=for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1230Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1231extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1232stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1233that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1234bytes since the titlecase version may be longer than the original character.
1235
1236The first code point of the titlecased version is returned
21da7284
KW
1237(but note, as explained at L<the top of this section|/Character case
1238changing>, that there may be more).
1f607577 1239
a239b1e2
KW
1240The suffix C<_safe> in the function's name indicates that it will not attempt
1241to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1242true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1243input character is malformed in some way, the program may croak, or the
1244function may return the REPLACEMENT CHARACTER, at the discretion of the
1245implementation, and subject to change in future releases.
1246
1247=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
1248This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1249parameter The function therefore can't check if it is reading
8b4b30c5 1250beyond the end of the string. Starting in Perl v5.32, it will take the C<e>
607313a1
KW
1251parameter, becoming a synonym for C<toTITLE_utf8_safe>. At that time every
1252program that uses it will have to be changed to successfully compile. In the
1253meantime, the first runtime call to C<toTITLE_utf8> from each call point in the
1254program will raise a deprecation warning, enabled by default. You can convert
1255your program now to use C<toTITLE_utf8_safe>, and avoid the warnings, and get an
8b4b30c5 1256extra measure of protection, or you can wait until v5.32, when you'll be forced
607313a1 1257to add the C<e> parameter.
1f607577 1258
954c1994 1259=cut
353c9b6f 1260
d0da05db 1261XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1262really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1263and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1264
8a58bdcf 1265Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1266patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1267
954c1994
GS
1268*/
1269
de9e2639
KW
1270/* Specify the widest unsigned type on the platform. */
1271#ifdef QUADKIND
1272# define WIDEST_UTYPE U64
7c062697
KW
1273#else
1274# define WIDEST_UTYPE U32
1275#endif
1276
3912bc88
KW
1277/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1278 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1279 * bits of information are lost even on a 64-bit machine, but to get the
1280 * compiler to optimize it out if possible. This is because Configure makes
1281 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1282 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1283 *
1284 * For Coverity, be always true, because otherwise Coverity thinks
1285 * it finds several expressions that are always true, independent
1286 * of operands. Well, they are, but that is kind of the point.
220c71bf 1287 */
7e75d1a1 1288#ifndef __COVERITY__
6c5b02ac
KW
1289 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1290 * pointer) */
1291#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1292 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1293#else
1294#define FITS_IN_8_BITS(c) (1)
1295#endif
cf301eb7 1296
94250c4f
KW
1297/* Returns true if c is in the range l..u, where 'l' is non-negative
1298 * Written this way so that after optimization, only one conditional test is
1299 * needed.
1300 *
1301 * This isn't fully general, except for the special cased 'signed char' (which
1302 * should be resolved at compile time): It won't work if 'c' is negative, and
1303 * 'l' is larger than the max for that signed type. Thus if 'c' is a negative
1304 * int, and 'l' is larger than INT_MAX, it will fail. To protect agains this
1305 * happening, there is an assert that will generate a warning if c is larger
1306 * than e.g. INT_MAX if it is an 'unsigned int'. This could be a false
1307 * positive, but khw couldn't figure out a way to make it better. It's good
1308 * enough so far */
1309#define inRANGE(c, l, u) (__ASSERT_((l) >= 0) __ASSERT_((u) >= (l)) \
1310 ((sizeof(c) == 1) \
1311 ? (((WIDEST_UTYPE) ((((U8) (c))|0) - (l))) <= ((WIDEST_UTYPE) ((u) - (l)))) \
1312 : (__ASSERT_( (((WIDEST_UTYPE) 1) << (CHARBITS * sizeof(c) - 1) & (c)) \
1313 /* sign bit of c is 0 */ == 0 \
1314 || (((~ ((WIDEST_UTYPE) 1) << ((CHARBITS * sizeof(c) - 1) - 1))\
1315 /* l not larger than largest value in c's signed type */ \
1316 & ~ ((WIDEST_UTYPE) 0)) & (l)) == 0) \
1317 ((WIDEST_UTYPE) (((c) - (l)) | 0) <= ((WIDEST_UTYPE) ((u) - (l)))))))
305fe86e 1318
41f43cc2 1319#ifdef EBCDIC
b6340bd0 1320# ifndef _ALL_SOURCE
0852beac
KW
1321 /* The native libc isascii() et.al. functions return the wrong results
1322 * on at least z/OS unless this is defined. */
b6340bd0
KW
1323# error _ALL_SOURCE should probably be defined
1324# endif
41f43cc2 1325#else
0852beac
KW
1326 /* There is a simple definition of ASCII for ASCII platforms. But the
1327 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1328 * other macros below.
3f3c579d
KW
1329 *
1330 * The cast here is used instead of '(c) >= 0', because some compilers emit
1331 * a warning that that test is always true when the parameter is an
1332 * unsigned type. khw supposes that it could be written as
1333 * && ((c) == '\0' || (c) > 0)
1334 * to avoid the message, but the cast will likely avoid extra branches even
1335 * with stupid compilers.
1336 *
1337 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1338 * a pointer) */
9c903d59 1339# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1340#endif
1341
38694112
KW
1342/* Take the eight possible bit patterns of the lower 3 bits and you get the
1343 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1344 * can be ignored. If the rest match '0', we have an octal */
1345#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1346
9fb1bf9d 1347#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1348
a500dc72
KW
1349/* Character class numbers. For internal core Perl use only. The ones less
1350 * than 32 are used in PL_charclass[] and the ones up through the one that
1351 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1352 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1353 * their actual definitions are here. If that file has a name not used here,
1354 * it won't compile.
1709d539
KW
1355 *
1356 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561 1357 * frequency of their use. This gives a slight edge to exiting a loop earlier
58a3ba2c
KW
1358 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1359 * real Posix character class, and some (small) inefficiencies in regular
1360 * expression handling would be introduced by putting it in the middle of those
1361 * that are. Also, cntrl and ascii come after the others as it may be useful
1362 * to group these which have no members that match above Latin1, (or above
1363 * ASCII in the latter case) */
1364
1709d539
KW
1365# define _CC_WORDCHAR 0 /* \w and [:word:] */
1366# define _CC_DIGIT 1 /* \d and [:digit:] */
1367# define _CC_ALPHA 2 /* [:alpha:] */
1368# define _CC_LOWER 3 /* [:lower:] */
1369# define _CC_UPPER 4 /* [:upper:] */
1370# define _CC_PUNCT 5 /* [:punct:] */
1371# define _CC_PRINT 6 /* [:print:] */
15861f94 1372# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1373# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1374# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
779cf272 1375# define _CC_SPACE 10 /* \s, [:space:] */
34aeb2e9
KW
1376# define _CC_PSXSPC _CC_SPACE /* XXX Temporary, can be removed
1377 when the deprecated isFOO_utf8()
1378 functions are removed */
b0d691b2
KW
1379# define _CC_BLANK 11 /* [:blank:] */
1380# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1381# define _CC_CNTRL 13 /* [:cntrl:] */
1382# define _CC_ASCII 14 /* [:ascii:] */
1383# define _CC_VERTSPACE 15 /* \v */
1709d539 1384
a0947d7b
KW
1385# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1386
1709d539 1387/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1388 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1389# define _CC_IDFIRST 16
1390# define _CC_CHARNAME_CONT 17
1391# define _CC_NONLATIN1_FOLD 18
1392# define _CC_NONLATIN1_SIMPLE_FOLD 19
1393# define _CC_QUOTEMETA 20
1394# define _CC_NON_FINAL_FOLD 21
1395# define _CC_IS_IN_SOME_FOLD 22
1396# define _CC_MNEMONIC_CNTRL 23
073c22b3 1397
34aeb2e9
KW
1398# define _CC_IDCONT 24 /* XXX Temporary, can be removed when the deprecated
1399 isFOO_utf8() functions are removed */
1400
073c22b3
KW
1401/* This next group is only used on EBCDIC platforms, so theoretically could be
1402 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
1403# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1404# define _CC_UTF8_IS_START 29
1405# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1406# define _CC_UTF8_IS_CONTINUATION 31
1407/* Unused: 24-27
f4cdb42c
KW
1408 * If more bits are needed, one could add a second word for non-64bit
1409 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1410 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1411 * is used only for optimization (as of this writing), and differs in the
1412 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1413 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1414 * would just cause /i regexes which match them to run less efficiently.
1415 * Similarly the EBCDIC-only bits are used just for speed, and could be
1416 * replaced by other means */
96ac0975 1417
3a371f2f
KW
1418#if defined(PERL_CORE) || defined(PERL_EXT)
1419/* An enum version of the character class numbers, to help compilers
1420 * optimize */
1421typedef enum {
3a371f2f 1422 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1423 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1424 _CC_ENUM_ASCII = _CC_ASCII,
1425 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1426 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1427 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1428 _CC_ENUM_DIGIT = _CC_DIGIT,
1429 _CC_ENUM_GRAPH = _CC_GRAPH,
1430 _CC_ENUM_LOWER = _CC_LOWER,
1431 _CC_ENUM_PRINT = _CC_PRINT,
1432 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1433 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1434 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1435 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1436 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1437 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1438} _char_class_number;
1439#endif
1440
86f72d56 1441#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1442
6635f04f 1443START_EXTERN_C
96ac0975
NC
1444# ifdef DOINIT
1445EXTCONST U32 PL_charclass[] = {
1446# include "l1_char_class_tab.h"
1447};
1448
1449# else /* ! DOINIT */
1450EXTCONST U32 PL_charclass[];
1451# endif
6635f04f 1452END_EXTERN_C
96ac0975 1453
265c1f46 1454 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1455# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1456
1457 /* For internal core Perl use only: the base macro for defining macros like
1458 * isALPHA */
ff7ecfc3 1459# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1460 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1461
f4cdb42c
KW
1462 /* The mask for the _A versions of the macros; it just adds in the bit for
1463 * ASCII. */
1464# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1465
4650c663
KW
1466 /* For internal core Perl use only: the base macro for defining macros like
1467 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1468 * the ASCII bit are present */
b7d90381
KW
1469# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1470 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1471 == _CC_mask_A(classnum)))
f4cdb42c 1472
26c1d9d8
KW
1473/* On ASCII platforms certain classes form a single range. It's faster to
1474 * special case these. isDIGIT is a single range on all platforms */
b877c1ff
KW
1475# ifdef EBCDIC
1476# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1477# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1478# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1479# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1480# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1481# else
26c1d9d8 1482 /* By folding the upper and lowercase, we can use a single range */
b877c1ff 1483# define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
26c1d9d8 1484# define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
b877c1ff
KW
1485# define isLOWER_A(c) inRANGE(c, 'a', 'z')
1486# define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1487# define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1488# endif
15861f94 1489# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1490# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1491# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
b877c1ff 1492# define isDIGIT_A(c) inRANGE(c, '0', '9')
f4cdb42c
KW
1493# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1494# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
f4cdb42c 1495# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1496# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1497 */
d95f8b6a 1498# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1499# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1500# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1501# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1502
1503 /* continuation character for legal NAME in \N{NAME} */
1504# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1505
1506# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1507# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1508# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1509# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1510# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1511# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1512# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1513# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1514# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1515# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1516
0852beac
KW
1517# ifdef EBCDIC
1518# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1519# endif
1520
f12c0118
KW
1521 /* Participates in a single-character fold with a character above 255 */
1522# 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)))
1523
1524 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1525# 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 1526
4eeeb416 1527# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1528# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1529 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1530# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1531 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1532# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1533 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1534#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1535
1536 /* If we don't have perl.h, we are compiling a utility program. Below we
1537 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1538 * to it. Most are coded based on first principles. These are written to
74665a89 1539 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
182c4ace 1540# define isDIGIT_A(c) inRANGE(c, '0', '9')
0852beac 1541# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1542# define isSPACE_A(c) (isBLANK_A(c) \
1543 || (c) == '\n' \
1544 || (c) == '\r' \
1545 || (c) == '\v' \
0852beac 1546 || (c) == '\f')
74665a89
KW
1547 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1548 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1549 * performance (if optimization isn't on), and allow the same code to be
1550 * used for both platform types */
182c4ace
KW
1551# define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1552 || inRANGE((c), 'j', 'r') \
1553 || inRANGE((c), 's', 'z')
1554# define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1555 || inRANGE((c), 'J', 'R') \
1556 || inRANGE((c), 'S', 'Z')
a4d7a999
KW
1557# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1558# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1559# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1560# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
182c4ace
KW
1561# define isXDIGIT_A(c) ( isDIGIT_A(c) \
1562 || inRANGE((c), 'a', 'f') \
1563 || inRANGE((c), 'A', 'F')
74665a89
KW
1564# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1565 || (c) == '#' || (c) == '$' || (c) == '%' \
1566 || (c) == '&' || (c) == '\'' || (c) == '(' \
1567 || (c) == ')' || (c) == '*' || (c) == '+' \
1568 || (c) == ',' || (c) == '.' || (c) == '/' \
1569 || (c) == ':' || (c) == ';' || (c) == '<' \
1570 || (c) == '=' || (c) == '>' || (c) == '?' \
1571 || (c) == '@' || (c) == '[' || (c) == '\\' \
1572 || (c) == ']' || (c) == '^' || (c) == '_' \
1573 || (c) == '`' || (c) == '{' || (c) == '|' \
1574 || (c) == '}' || (c) == '~')
1575# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1576# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1577
0852beac 1578# ifdef EBCDIC
74665a89
KW
1579 /* The below is accurate for the 3 EBCDIC code pages traditionally
1580 * supported by perl. The only difference between them in the controls
1581 * is the position of \n, and that is represented symbolically below */
1582# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1583 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1584 || (c) == '\t' || (c) == '\v' \
182c4ace 1585 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
74665a89 1586 || (c) == 7 /* U+7F DEL */ \
182c4ace
KW
1587 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1588 DC[1-3] */ \
74665a89
KW
1589 || (c) == 0x18 /* U+18 CAN */ \
1590 || (c) == 0x19 /* U+19 EOM */ \
182c4ace 1591 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
74665a89
KW
1592 || (c) == 0x26 /* U+17 ETB */ \
1593 || (c) == 0x27 /* U+1B ESC */ \
1594 || (c) == 0x2D /* U+05 ENQ */ \
1595 || (c) == 0x2E /* U+06 ACK */ \
1596 || (c) == 0x32 /* U+16 SYN */ \
1597 || (c) == 0x37 /* U+04 EOT */ \
1598 || (c) == 0x3C /* U+14 DC4 */ \
1599 || (c) == 0x3D /* U+15 NAK */ \
1600 || (c) == 0x3F)/* U+1A SUB */
0852beac 1601# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1602# else /* isASCII is already defined for ASCII platforms, so can use that to
1603 define isCNTRL */
1604# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1605# endif
1606
3ffc8c70 1607 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1608 * during debugging, and it seems best to keep them. We may be called
1609 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1610 * do anything anyway, so make it not a problem */
1611# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1612# define NATIVE_TO_LATIN1(ch) (ch)
1613# endif
3ded5eb0
KW
1614# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1615# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1616# define isBLANK_L1(c) (isBLANK_A(c) \
1617 || (FITS_IN_8_BITS(c) \
1618 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1619# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1620# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1621# define isLOWER_L1(c) (isLOWER_A(c) \
1622 || (FITS_IN_8_BITS(c) \
ae683a5f 1623 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1624 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1625 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1626 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1627 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1628# define isPRINT_L1(c) (isPRINT_A(c) \
1629 || (FITS_IN_8_BITS(c) \
1630 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1631# define isPUNCT_L1(c) (isPUNCT_A(c) \
1632 || (FITS_IN_8_BITS(c) \
ae683a5f 1633 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1634 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1635 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1636 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1637 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1638 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1639 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1640# define isSPACE_L1(c) (isSPACE_A(c) \
1641 || (FITS_IN_8_BITS(c) \
ae683a5f 1642 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1643 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1644# define isUPPER_L1(c) (isUPPER_A(c) \
1645 || (FITS_IN_8_BITS(c) \
182c4ace
KW
1646 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1647 0xC0, 0xDE) \
3ded5eb0
KW
1648 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1649# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1650# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1651# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1652 || isBLANK_L1(c) \
1653 || (c) == '-' \
1654 || (c) == '(' \
1655 || (c) == ')')
1656 /* The following are not fully accurate in the above-ASCII range. I (khw)
1657 * don't think it's necessary to be so for the purposes where this gets
1658 * compiled */
1659# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1660# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1661
1662 /* And these aren't accurate at all. They are useful only for above
1663 * Latin1, which utilities and bootstrapping don't deal with */
1664# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1665# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1666# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1667
1668 /* Many of the macros later in this file are defined in terms of these. By
1669 * implementing them with a function, which converts the class number into
1670 * a call to the desired macro, all of the later ones work. However, that
1671 * function won't be actually defined when building a utility program (no
1672 * perl.h), and so a compiler error will be generated if one is attempted
1673 * to be used. And the above-Latin1 code points require Unicode tables to
1674 * be present, something unlikely to be the case when bootstrapping */
1675# define _generic_isCC(c, classnum) \
1676 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1677# define _generic_isCC_A(c, classnum) \
1678 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1679#endif /* End of no perl.h H_PERL */
8a58bdcf 1680
e66b99e9
KW
1681#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1682#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1683#define isASCII_A(c) isASCII(c)
1684#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1685#define isBLANK(c) isBLANK_A(c)
1686#define isCNTRL(c) isCNTRL_A(c)
1687#define isDIGIT(c) isDIGIT_A(c)
1688#define isGRAPH(c) isGRAPH_A(c)
1689#define isIDFIRST(c) isIDFIRST_A(c)
1690#define isLOWER(c) isLOWER_A(c)
1691#define isPRINT(c) isPRINT_A(c)
779cf272 1692#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1693#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1694#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1695#define isPUNCT(c) isPUNCT_A(c)
1696#define isSPACE(c) isSPACE_A(c)
1697#define isUPPER(c) isUPPER_A(c)
1698#define isWORDCHAR(c) isWORDCHAR_A(c)
1699#define isXDIGIT(c) isXDIGIT_A(c)
1700
1701/* ASCII casing. These could also be written as
1702 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1703 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1704 which uses table lookup and mask instead of subtraction. (This would
c5e9991e
KW
1705 work because the _MOD does not apply in the ASCII range).
1706
1707 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1708 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1709 there non-ASCII invariants, and all of them are controls.) */
68067e4e
DM
1710#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1711#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1712
25200305
KW
1713/* In the ASCII range, these are equivalent to what they're here defined to be.
1714 * But by creating these definitions, other code doesn't have to be aware of
c5e9991e
KW
1715 * this detail. Actually this works for all UTF-8 invariants, not just the
1716 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
25200305 1717#define toFOLD(c) toLOWER(c)
25200305
KW
1718#define toTITLE(c) toUPPER(c)
1719
c753c8d3
KW
1720#define toLOWER_A(c) toLOWER(c)
1721#define toUPPER_A(c) toUPPER(c)
25200305
KW
1722#define toFOLD_A(c) toFOLD(c)
1723#define toTITLE_A(c) toTITLE(c)
1a0901db 1724
4650c663 1725/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1726#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1727 ? (c) \
f4cd282c 1728 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1729#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1730
1a0901db 1731/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1732 * all mapped to one of them, and these need special handling; returns the
1733 * input itself if is out-of-range */
b2bf251f 1734#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1735 ? (c) \
f4cd282c 1736 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1737#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1738
beab9ebe
KW
1739/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1740
1741/* For internal core Perl use only: the base macro for defining macros like
1742 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1743 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1744 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1745 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1746 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1747 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1748 * code points under locale rules, use the macros based on _generic_LC_uvchr
1749 * instead of this. */
beab9ebe
KW
1750#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1751 (! FITS_IN_8_BITS(c) \
1752 ? 0 \
31f05a37
KW
1753 : IN_UTF8_CTYPE_LOCALE \
1754 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1755 : cBOOL(non_utf8))
1756
1757/* For internal core Perl use only: a helper macro for defining macros like
1758 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1759 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1760 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1761 * _generic_LC_base for more info */
1762#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1763 _generic_LC_base(c,utf8_locale_classnum, \
1764 non_utf8_func( (_LC_CAST) (c)))
1765
1766/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1767 * 'c' is the platform's native underscore character */
1768#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1769 _generic_LC_base(c, utf8_locale_classnum, \
1770 (non_utf8_func( (_LC_CAST) (c)) \
1771 || (char)(c) == '_'))
1772
1773/* These next three are also for internal core Perl use only: case-change
247985d4
KW
1774 * helper macros. The reason for using the PL_latin arrays is in case the
1775 * system function is defective; it ensures uniform results that conform to the
b257a28c 1776 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
beab9ebe
KW
1777#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1778 ? (c) \
31f05a37
KW
1779 : (IN_UTF8_CTYPE_LOCALE) \
1780 ? PL_latin1_lc[ (U8) (c) ] \
5a10328c 1781 : (cast)function((cast)(c)))
beab9ebe 1782
31f05a37
KW
1783/* Note that the result can be larger than a byte in a UTF-8 locale. It
1784 * returns a single value, so can't adequately return the upper case of LATIN
1785 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1786 * values "SS"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1787 * otherwise returns its input. It does not handle the anomalies in UTF-8
1788 * Turkic locales. */
beab9ebe
KW
1789#define _generic_toUPPER_LC(c, function, cast) \
1790 (! FITS_IN_8_BITS(c) \
1791 ? (c) \
31f05a37 1792 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1793 ? (cast)function((cast)(c)) \
31f05a37
KW
1794 : ((((U8)(c)) == MICRO_SIGN) \
1795 ? GREEK_CAPITAL_LETTER_MU \
1796 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1797 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1798 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1799 ? (__ASSERT_(0) (c)) \
1800 : PL_mod_latin1_uc[ (U8) (c) ])))))
1801
1802/* Note that the result can be larger than a byte in a UTF-8 locale. It
1803 * returns a single value, so can't adequately return the fold case of LATIN
1804 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1805 * values "ss"); instead it asserts against that under DEBUGGING, and
b257a28c
KW
1806 * otherwise returns its input. It does not handle the anomalies in UTF-8
1807 * Turkic locales */
beab9ebe 1808#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1809 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1810 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1811 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1812 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1813 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1814
84061b6a 1815/* Use the libc versions for these if available. */
f05550c0 1816#if defined(HAS_ISASCII)
84061b6a
KW
1817# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1818#else
1819# define isASCII_LC(c) isASCII(c)
1820#endif
1821
f05550c0 1822#if defined(HAS_ISBLANK)
84061b6a 1823# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1824#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1825# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1826#endif
1827
f05550c0 1828#define _LC_CAST U8
bbce6d69 1829
f05550c0 1830#ifdef WIN32
375f5f06
KW
1831 /* The Windows functions don't bother to follow the POSIX standard, which
1832 * for example says that something can't both be a printable and a control.
1833 * But Windows treats the \t control as a printable, and does such things
1834 * as making superscripts into both digits and punctuation. This tames
1835 * these flaws by assuming that the definitions of both controls and space
1836 * are correct, and then making sure that other definitions don't have
1837 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1838 * Not all possible weirdnesses are checked for, just the ones that were
1839 * detected on actual Microsoft code pages */
1840
b7d90381
KW
1841# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1842# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1843
1844# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1845 && isALPHANUMERIC_LC(c))
1846# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1847 ! isPUNCT_LC(c))
1848# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1849 isALPHANUMERIC_LC(c))
1850# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1851# define isIDFIRST_LC(c) (((c) == '_') \
1852 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1853# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1854# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1855# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1856# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1857# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1858# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1859 && isALPHANUMERIC_LC(c))
f05550c0
BF
1860
1861# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1862# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1863# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1864
1865#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1866 /* For most other platforms */
beab9ebe 1867
f05550c0
BF
1868# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1869# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1870# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1871# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1872# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1873# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1874# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1875# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1876# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1877# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1878# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1879# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1880# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1881
1882
1883# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1884# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1885# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1886
1887#else /* The final fallback position */
1888
b7d90381
KW
1889# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1890# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1891# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1892# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1893# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1894# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1895# define isLOWER_LC(c) (isascii(c) && islower(c))
1896# define isPRINT_LC(c) (isascii(c) && isprint(c))
1897# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1898# define isSPACE_LC(c) (isascii(c) && isspace(c))
1899# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1900# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1901# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1902
1903# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1904# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1905# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1906
f05550c0 1907#endif
55204971 1908
eba68aa0
KW
1909#define isIDCONT(c) isWORDCHAR(c)
1910#define isIDCONT_A(c) isWORDCHAR_A(c)
1911#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1912#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1913#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1914
4650c663 1915/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1916 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1917 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1918 * classes where there is a macro or function 'above_latin1' that takes the
1919 * single argument 'c' and returns the desired value. These exist for those
1920 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1921 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1922 * for classes where that overhead is faster than a direct lookup.
d0da05db 1923 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1924 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1925 * have duplicate checks here, so could create versions of the macros that
1926 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1927
1928/* Note that all ignore 'use bytes' */
1e222e4f
KW
1929#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1930 ? _generic_isCC(c, classnum) \
cd500f2f 1931 : above_latin1(c))
1e222e4f
KW
1932#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
1933 ? _generic_isCC(c, classnum) \
922e8cb4 1934 : _is_uni_FOO(classnum, c))
d0da05db
KW
1935#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1936#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1937#define isASCII_uvchr(c) isASCII(c)
1938#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1939#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1940#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1941#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1942#define isIDCONT_uvchr(c) \
1943 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1944#define isIDFIRST_uvchr(c) \
1945 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
d0da05db
KW
1946#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1947#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1948
1949#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1950#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1951#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1952
1953#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1954#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1955#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1956#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1957
1958#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1959#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1960#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1961#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1962
1963/* For backwards compatibility, even though '_uni' should mean official Unicode
1964 * code points, in Perl it means native for those below 256 */
1965#define isALPHA_uni(c) isALPHA_uvchr(c)
1966#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1967#define isASCII_uni(c) isASCII_uvchr(c)
1968#define isBLANK_uni(c) isBLANK_uvchr(c)
1969#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1970#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1971#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1972#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1973#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1974#define isLOWER_uni(c) isLOWER_uvchr(c)
1975#define isPRINT_uni(c) isPRINT_uvchr(c)
1976#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1977#define isSPACE_uni(c) isSPACE_uvchr(c)
1978#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1979#define isUPPER_uni(c) isUPPER_uvchr(c)
1980#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1981#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1982#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1983#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1984#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1985#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1986#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1987
4650c663
KW
1988/* For internal core Perl use only: the base macros for defining macros like
1989 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1990 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1991 * simple class definitions; the other for more complex. These are like
d0da05db 1992 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1993#define _generic_LC_uvchr(latin1, above_latin1, c) \
1994 (c < 256 ? latin1(c) : above_latin1(c))
1995#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1996 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1997
1998#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1999#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
2000 _CC_ALPHANUMERIC, c)
b7d90381
KW
2001#define isASCII_LC_uvchr(c) isASCII_LC(c)
2002#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2003 is_HORIZWS_cp_high, c)
feeab5a9 2004#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
2005#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2006#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 2007#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 2008 _is_uni_perl_idcont, c)
b7d90381 2009#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
2010 _is_uni_perl_idstart, c)
2011#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
2012#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 2013#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 2014#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 2015#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 2016 is_XPERLSPACE_cp_high, c)
cd500f2f 2017#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 2018#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 2019 _CC_WORDCHAR, c)
b7d90381
KW
2020#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2021 is_XDIGIT_cp_high, c)
e712593e 2022
b7d90381 2023#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 2024
4650c663
KW
2025/* For internal core Perl use only: the base macros for defining macros like
2026 * isALPHA_utf8. These are like the earlier defined macros, but take an input
2027 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
2028 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
2029 * 'utf8' parameter. This relies on the fact that ASCII characters have the
2030 * same representation whether utf8 or not. Note that it assumes that the utf8
2031 * has been validated, and ignores 'use bytes' */
34aeb2e9
KW
2032#define _base_generic_utf8(enum_name, name, p, use_locale ) \
2033 _is_utf8_FOO(CAT2(_CC_, enum_name), \
2034 (const U8 *) p, \
2035 "is" STRINGIFY(name) "_utf8", \
2036 "is" STRINGIFY(name) "_utf8_safe", \
2037 1, use_locale, __FILE__,__LINE__)
2038
2039#define _generic_utf8(name, p) _base_generic_utf8(name, name, p, 0)
da8c1a98
KW
2040
2041/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2042 * they don't otherwise go out of their way to look for malformed UTF-8. If
2043 * they can return accurate results without knowing if the input is otherwise
2044 * malformed, they do so. For example isASCII is accurate in spite of any
2045 * non-length malformations because it looks only at a single byte. Likewise
2046 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2047 * variant ones return FALSE. But, if the input has to be well-formed in order
2048 * for the results to be accurate, the macros will test and if malformed will
2049 * call a routine to die
2050 *
2051 * Except for toke.c, the macros do assume that e > p, asserting that on
2052 * DEBUGGING builds. Much code that calls these depends on this being true,
2053 * for other reasons. toke.c is treated specially as using the regular
2054 * assertion breaks it in many ways. All strings that these operate on there
2055 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2056 * bunch of code in toke.c assumes that this is true, so the assertion allows
2057 * for that */
2058#ifdef PERL_IN_TOKE_C
2059# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2060#else
2061# define _utf8_safe_assert(p,e) ((e) > (p))
2062#endif
2063
2064#define _generic_utf8_safe(classnum, p, e, above_latin1) \
2065 (__ASSERT_(_utf8_safe_assert(p, e)) \
2066 (UTF8_IS_INVARIANT(*(p))) \
2067 ? _generic_isCC(*(p), classnum) \
2068 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2069 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2070 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2071 classnum) \
2072 : (_force_out_malformed_utf8_message( \
2073 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2074 : above_latin1))
b7d90381
KW
2075/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2076 * 'above_latin1' can be a macro */
da8c1a98
KW
2077#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2078 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2079#define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \
2080 _generic_utf8_safe(classnum, p, e, \
2081 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2082 ? (_force_out_malformed_utf8_message( \
2083 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2084 : above_latin1(p)))
4650c663 2085/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 2086 * 'above_latin1' parameter */
da8c1a98
KW
2087#define _generic_swash_utf8_safe(classnum, p, e) \
2088_generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e))
922e8cb4 2089
cc8ab7c0 2090/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
2091 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2092 * class is TRUE for. Hence it can skip the tests for this range.
2093 * 'above_latin1' should include its arguments */
da8c1a98
KW
2094#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2095 (__ASSERT_(_utf8_safe_assert(p, e)) \
2096 (UTF8_IS_INVARIANT(*(p))) \
2097 ? _generic_isCC(*(p), classnum) \
2098 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2099 ? 0 /* Note that doesn't check validity for latin1 */ \
2100 : above_latin1)
2101
84238efa 2102
34aeb2e9
KW
2103#define isALPHA_utf8(p) _generic_utf8(ALPHA, p)
2104#define isALPHANUMERIC_utf8(p) _generic_utf8(ALPHANUMERIC, p)
2105#define isASCII_utf8(p) _generic_utf8(ASCII, p)
2106#define isBLANK_utf8(p) _generic_utf8(BLANK, p)
2107#define isCNTRL_utf8(p) _generic_utf8(CNTRL, p)
2108#define isDIGIT_utf8(p) _generic_utf8(DIGIT, p)
2109#define isGRAPH_utf8(p) _generic_utf8(GRAPH, p)
2110#define isIDCONT_utf8(p) _generic_utf8(IDCONT, p)
2111#define isIDFIRST_utf8(p) _generic_utf8(IDFIRST, p)
2112#define isLOWER_utf8(p) _generic_utf8(LOWER, p)
2113#define isPRINT_utf8(p) _generic_utf8(PRINT, p)
2114#define isPSXSPC_utf8(p) _generic_utf8(PSXSPC, p)
2115#define isPUNCT_utf8(p) _generic_utf8(PUNCT, p)
2116#define isSPACE_utf8(p) _generic_utf8(SPACE, p)
2117#define isUPPER_utf8(p) _generic_utf8(UPPER, p)
2118#define isVERTWS_utf8(p) _generic_utf8(VERTSPACE, p)
2119#define isWORDCHAR_utf8(p) _generic_utf8(WORDCHAR, p)
2120#define isXDIGIT_utf8(p) _generic_utf8(XDIGIT, p)
e8fa43e2 2121
da8c1a98
KW
2122#define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e)
2123#define isALPHANUMERIC_utf8_safe(p, e) \
2124 _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e)
2125#define isASCII_utf8_safe(p, e) \
2126 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2127 * works */ \
2128 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2129#define isBLANK_utf8_safe(p, e) \
2130 _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
2131
e8fa43e2
KW
2132#ifdef EBCDIC
2133 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2134 * more efficient macro instead of the more general one */
da8c1a98 2135# define isCNTRL_utf8_safe(p, e) \
56d02b8c 2136 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 2137#else
da8c1a98 2138# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
2139#endif
2140
da8c1a98
KW
2141#define isDIGIT_utf8_safe(p, e) \
2142 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2143 _is_utf8_FOO_with_len(_CC_DIGIT, p, e))
2144#define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e)
2145#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
2146 _is_utf8_perl_idcont_with_len, p, e)
e5dcd934 2147
c11ff943
KW
2148/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2149 * IDFIRST is an alnum. See
c8362b00 2150 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
2151 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2152 * This used to be not the XID version, but we decided to go with the more
2153 * modern Unicode definition */
da8c1a98
KW
2154#define isIDFIRST_utf8_safe(p, e) \
2155 _generic_func_utf8_safe(_CC_IDFIRST, \
2156 _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e))
2157
2158#define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e)
2159#define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e)
2160#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2161#define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e)
2162#define isSPACE_utf8_safe(p, e) \
2163 _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2164#define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e)
2165#define isVERTWS_utf8_safe(p, e) \
2166 _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
2167#define isWORDCHAR_utf8_safe(p, e) \
2168 _generic_swash_utf8_safe(_CC_WORDCHAR, p, e)
2169#define isXDIGIT_utf8_safe(p, e) \
2170 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2171 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2172 ? (_force_out_malformed_utf8_message( \
2173 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2174 : is_XDIGIT_high(p)))
a0ed51b3 2175
7d2b61dd 2176#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
2177#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
2178#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
2179#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 2180
567b353c 2181/* For internal core use only, subject to change */
607313a1
KW
2182#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f, "", 0)
2183#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f, "", 0)
2184#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f, "", 0)
2185#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f, "", 0)
a1a5ec35
KW
2186
2187#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2188#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2189#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2190#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 2191
4650c663
KW
2192/* For internal core Perl use only: the base macros for defining macros like
2193 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
2194 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
2195 * 'macro' parameter */
34aeb2e9
KW
2196#define _generic_LC_utf8(name, p) _base_generic_utf8(name, name, p, 1)
2197
2198#define isALPHA_LC_utf8(p) _generic_LC_utf8(ALPHA, p)
2199#define isALPHANUMERIC_LC_utf8(p) _generic_LC_utf8(ALPHANUMERIC, p)
2200#define isASCII_LC_utf8(p) _generic_LC_utf8(ASCII, p)
2201#define isBLANK_LC_utf8(p) _generic_LC_utf8(BLANK, p)
2202#define isCNTRL_LC_utf8(p) _generic_LC_utf8(CNTRL, p)
2203#define isDIGIT_LC_utf8(p) _generic_LC_utf8(DIGIT, p)
2204#define isGRAPH_LC_utf8(p) _generic_LC_utf8(GRAPH, p)
2205#define isIDCONT_LC_utf8(p) _generic_LC_utf8(IDCONT, p)
2206#define isIDFIRST_LC_utf8(p) _generic_LC_utf8(IDFIRST, p)
2207#define isLOWER_LC_utf8(p) _generic_LC_utf8(LOWER, p)
2208#define isPRINT_LC_utf8(p) _generic_LC_utf8(PRINT, p)
2209#define isPSXSPC_LC_utf8(p) _generic_LC_utf8(PSXSPC, p)
2210#define isPUNCT_LC_utf8(p) _generic_LC_utf8(PUNCT, p)
2211#define isSPACE_LC_utf8(p) _generic_LC_utf8(SPACE, p)
2212#define isUPPER_LC_utf8(p) _generic_LC_utf8(UPPER, p)
2213#define isWORDCHAR_LC_utf8(p) _generic_LC_utf8(WORDCHAR, p)
2214#define isXDIGIT_LC_utf8(p) _generic_LC_utf8(XDIGIT, p)
2215
da8c1a98
KW
2216/* For internal core Perl use only: the base macros for defining macros like
2217 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2218 * point in 'p' is within the 0-255 range, it uses locale rules from the
2219 * passed-in 'macro' parameter */
2220#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2221 (__ASSERT_(_utf8_safe_assert(p, e)) \
2222 (UTF8_IS_INVARIANT(*(p))) \
2223 ? macro(*(p)) \
2224 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2225 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2226 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2227 : (_force_out_malformed_utf8_message( \
2228 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2229 : above_latin1))
2230
2231#define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \
2232 _generic_LC_utf8_safe(macro, p, e, \
2233 _is_utf8_FOO_with_len(classnum, p, e))
2234
2235#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2236 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2237
2238#define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \
2239 _generic_LC_utf8_safe(classnum, p, e, \
2240 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2241 ? (_force_out_malformed_utf8_message( \
2242 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2243 : above_latin1(p)))
2244
2245#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2246 _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \
2247 _CC_ALPHANUMERIC, p, e)
2248#define isALPHA_LC_utf8_safe(p, e) \
2249 _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2250#define isASCII_LC_utf8_safe(p, e) \
2251 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2252#define isBLANK_LC_utf8_safe(p, e) \
2253 _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2254#define isCNTRL_LC_utf8_safe(p, e) \
2255 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2256#define isDIGIT_LC_utf8_safe(p, e) \
2257 _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2258#define isGRAPH_LC_utf8_safe(p, e) \
2259 _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2260#define isIDCONT_LC_utf8_safe(p, e) \
2261 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2262 _is_utf8_perl_idcont_with_len, p, e)
2263#define isIDFIRST_LC_utf8_safe(p, e) \
2264 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2265 _is_utf8_perl_idstart_with_len, p, e)
2266#define isLOWER_LC_utf8_safe(p, e) \
2267 _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2268#define isPRINT_LC_utf8_safe(p, e) \
2269 _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2270#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2271#define isPUNCT_LC_utf8_safe(p, e) \
2272 _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2273#define isSPACE_LC_utf8_safe(p, e) \
2274 _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2275#define isUPPER_LC_utf8_safe(p, e) \
2276 _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2277#define isWORDCHAR_LC_utf8_safe(p, e) \
2278 _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2279#define isXDIGIT_LC_utf8_safe(p, e) \
2280 _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2281
fbc19f27
KW
2282/* Macros for backwards compatibility and for completeness when the ASCII and
2283 * Latin1 values are identical */
b7d90381
KW
2284#define isALPHAU(c) isALPHA_L1(c)
2285#define isDIGIT_L1(c) isDIGIT_A(c)
2286#define isOCTAL(c) isOCTAL_A(c)
2287#define isOCTAL_L1(c) isOCTAL_A(c)
2288#define isXDIGIT_L1(c) isXDIGIT_A(c)
2289#define isALNUM(c) isWORDCHAR(c)
a377c856 2290#define isALNUM_A(c) isALNUM(c)
b7d90381
KW
2291#define isALNUMU(c) isWORDCHAR_L1(c)
2292#define isALNUM_LC(c) isWORDCHAR_LC(c)
2293#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2294#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
2295#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
2296#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
2297#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2298#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2299#define isALNUMC(c) isALPHANUMERIC(c)
2300#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2301#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2302#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 2303#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 2304#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
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
cb27eebd
KW
2345/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2346 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2347 * which is advanced. The input is validated only by an assert() in DEBUGGING
2348 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2349 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2350#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2351 ? (c) \
2352 : ((c) + 9))))
902008b8
KW
2353#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2354 ? (*(s)++) \
2355 : (*(s)++ + 9))))
95a59cab 2356
cb27eebd
KW
2357/* Converts a character known to represent an octal digit (0-7) to its numeric
2358 * value. The input is validated only by an assert() in DEBUGGING builds. In
2359 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2360#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2361
305b8651
KW
2362/* Efficiently returns a boolean as to if two native characters are equivalent
2363 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2364 * the ALPHA in the name is to remind you of that. This is asserted() in
2365 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2366 * works (on valid input) for both non- and UTF-8-encoded bytes.
2367 *
2368 * When one of the inputs is a compile-time constant and gets folded by the
2369 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2370 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2371 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2372 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2373 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2374 * just a single 0, in the bit position where the upper- and lowercase differ.
2375 * */
305b8651
KW
2376#define isALPHA_FOLD_EQ(c1, c2) \
2377 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2378 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2379#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2380
8e84507e 2381/*
ccfc67b7
JH
2382=head1 Memory Management
2383
a02a5408 2384=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2385The XSUB-writer's interface to the C C<malloc> function.
2386
596f7718 2387Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2388
c5008215
JC
2389In 5.9.3, Newx() and friends replace the older New() API, and drops
2390the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2391themselves. This aid has been superseded by a new build option,
d10b4965 2392PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2393there for use in XS modules supporting older perls.
2394
a02a5408 2395=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2396The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2397cast. See also C<L</Newx>>.
954c1994 2398
596f7718 2399Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2400
a02a5408 2401=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2402The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2403memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2404
596f7718 2405Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2406
954c1994
GS
2407=for apidoc Am|void|Renew|void* ptr|int nitems|type
2408The XSUB-writer's interface to the C C<realloc> function.
2409
596f7718 2410Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2411
954c1994
GS
2412=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2413The XSUB-writer's interface to the C C<realloc> function, with
2414cast.
2415
596f7718 2416Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2417
49b8b560 2418=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2419The XSUB-writer's interface to the C C<free> function.
2420
596f7718 2421This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2422
954c1994
GS
2423=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2424The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2425source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2426C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2427
e90e2364 2428=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2429Like C<Move> but returns C<dest>. Useful
72d33970 2430for encouraging compilers to tail-call
e90e2364
NC
2431optimise.
2432
954c1994
GS
2433=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2434The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2435source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2436C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2437
e90e2364
NC
2438=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2439
796b6530 2440Like C<Copy> but returns C<dest>. Useful
72d33970 2441for encouraging compilers to tail-call
e90e2364
NC
2442optimise.
2443
954c1994
GS
2444=for apidoc Am|void|Zero|void* dest|int nitems|type
2445
2446The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2447destination, C<nitems> is the number of items, and C<type> is the type.
2448
e90e2364
NC
2449=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2450
72d33970
FC
2451Like C<Zero> but returns dest. Useful
2452for encouraging compilers to tail-call
e90e2364
NC
2453optimise.
2454
da5d8dbb 2455=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2456This is an architecture-independent macro to copy one structure to another.
954c1994 2457
7e337ee0
JH
2458=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2459
2460Fill up memory with a byte pattern (a byte repeated over and over
2461again) that hopefully catches attempts to access uninitialized memory.
2462
2463=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2464
2465PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2466
1c12ffb4 2467=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2468
2469PoisonWith(0xEF) for catching access to freed memory.
2470
9965345d
JH
2471=for apidoc Am|void|Poison|void* dest|int nitems|type
2472
7e337ee0 2473PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2474
2475=cut */
954c1994 2476
561b68a9
SH
2477/* Maintained for backwards-compatibility only. Use newSV() instead. */
2478#ifndef PERL_CORE
ff06c60c 2479#define NEWSV(x,len) newSV(len)
561b68a9 2480#endif
ff06c60c 2481
b7112dce 2482#define MEM_SIZE_MAX ((MEM_SIZE)-1)
19a94d75 2483
a500027b 2484#define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
e6bdf523 2485
27d5b266 2486#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2487
2488/* This expression will be constant-folded at compile time. It checks
2489 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2490 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2491 * As well as avoiding the need for a run-time check in some cases, it's
2492 * designed to avoid compiler warnings like:
2493 * comparison is always false due to limited range of data type
73e8ff00
DM
2494 * It's mathematically equivalent to
2495 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2496 */
2497
2498# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
445198b9
LM
2499 ( sizeof(MEM_SIZE) < sizeof(n) \
2500 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 2501
88f9f128 2502/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2503 * compiler warnings. We *want* to write the expression as
2504 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2505 * (for some compile-time constant C), but even when the LHS
2506 * constant-folds to false at compile-time, g++ insists on emitting
2507 * warnings about the RHS (e.g. "comparison is always false"), so instead
2508 * we write it as
e6bdf523 2509 *
d98e5cde 2510 * (cond ? n : X) > C
88f9f128 2511 *
d98e5cde
DM
2512 * where X is a constant with X > C always false. Choosing a value for X
2513 * is tricky. If 0, some compilers will complain about 0 > C always being
2514 * false; if 1, Coverity complains when n happens to be the constant value
2515 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2516 * for X and hope that nothing else whines.
e6bdf523
DM
2517 */
2518
2519# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2520 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2521 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2522
2523# define MEM_WRAP_CHECK(n,t) \
2524 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2525 && (croak_memory_wrap(),0))
2526
2527# define MEM_WRAP_CHECK_1(n,t,a) \
2528 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2529 && (Perl_croak_nocontext("%s",(a)),0))
2530
814eedc8
DD
2531/* "a" arg must be a string literal */
2532# define MEM_WRAP_CHECK_s(n,t,a) \
2533 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2534 && (Perl_croak_nocontext("" a ""),0))
2535
8b44ba4c 2536#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2537
a500027b 2538#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
2539#else
2540
410319be
NC
2541#define MEM_WRAP_CHECK(n,t)
2542#define MEM_WRAP_CHECK_1(n,t,a)
814eedc8 2543#define MEM_WRAP_CHECK_s(n,t,a)
8b44ba4c
NC
2544#define MEM_WRAP_CHECK_(n,t)
2545
a500027b 2546#define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
27d5b266 2547
1936d2a7 2548#endif
8b44ba4c 2549
fe4f188c 2550#ifdef PERL_MEM_LOG
46c6c7e2 2551/*
9f653bb5 2552 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2553 * go through functions, which are handy for debugging breakpoints, but
2554 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2555 * line number, and C function name if available) passed in. This info can
2556 * then be used for logging the calls, for which one gets a sample
73d1d973 2557 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2558 *
46c6c7e2 2559 * Known problems:
94e892a6 2560 * - not all memory allocs get logged, only those
46c6c7e2 2561 * that go through Newx() and derivatives (while all
94e892a6 2562 * Safefrees do get logged)
46c6c7e2
JH
2563 * - __FILE__ and __LINE__ do not work everywhere
2564 * - __func__ or __FUNCTION__ even less so
2565 * - I think more goes on after the perlio frees but
2566 * the thing is that STDERR gets closed (as do all
2567 * the file descriptors)
2568 * - no deeper calling stack than the caller of the Newx()
2569 * or the kind, but do I look like a C reflection/introspection
2570 * utility to you?
2571 * - the function prototypes for the logging functions
2572 * probably should maybe be somewhere else than handy.h
2573 * - one could consider inlining (macrofying) the logging
2574 * for speed, but I am too lazy
2575 * - one could imagine recording the allocations in a hash,
2576 * (keyed by the allocation address?), and maintain that
2577 * through reallocs and frees, but how to do that without
2578 * any News() happening...?
73d1d973 2579 * - lots of -Ddefines to get useful/controllable output
b953482e 2580 * - lots of ENV reads
46c6c7e2
JH
2581 */
2582
0b0ab801 2583# ifdef PERL_CORE
73d1d973 2584# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2585enum mem_log_type {
2586 MLT_ALLOC,
2587 MLT_REALLOC,
d7a2c63c
MHM
2588 MLT_FREE,
2589 MLT_NEW_SV,
2590 MLT_DEL_SV
0b0ab801
MHM
2591};
2592# endif
12754f92 2593# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2594void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2595void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2596# endif
0b0ab801
MHM
2597# endif
2598
fe4f188c
JH
2599#endif
2600
2601#ifdef PERL_MEM_LOG
d1401ee9
MHM
2602#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2603#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2604#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2605#endif
2606
2607#ifndef MEM_LOG_ALLOC
2608#define MEM_LOG_ALLOC(n,t,a) (a)
2609#endif
2610#ifndef MEM_LOG_REALLOC
2611#define MEM_LOG_REALLOC(n,t,v,a) (a)
2612#endif
2613#ifndef MEM_LOG_FREE
2614#define MEM_LOG_FREE(a) (a)
2615#endif
2616
d1401ee9
MHM
2617#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2618#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2619#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2620
2621#ifndef PERL_CORE
a02a5408
JC
2622/* pre 5.9.x compatibility */
2623#define New(x,v,n,t) Newx(v,n,t)
2624#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2625#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2626#endif
a02a5408 2627
ff68c719 2628#define Renew(v,n,t) \
d1401ee9 2629 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2630#define Renewc(v,n,t,c) \
d1401ee9 2631 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2632
2633#ifdef PERL_POISON
2634#define Safefree(d) \
06c0cc96 2635 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2636#else
fe4f188c 2637#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2638#endif
55497cff 2639
dbb57106
YO
2640/* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2641 * as it handles cases like constant string arguments without throwing warnings *
2642 * the cast is required, as is the inequality check, to avoid warnings */
45908e4d 2643#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2644
45908e4d
YO
2645
2646#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)))
2647#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)))
2648#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2649
bdd1531d 2650/* Like above, but returns a pointer to 'd' */
45908e4d
YO
2651#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)))
2652#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 2653#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2654
7e337ee0
JH
2655#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2656#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2657#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2658#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2659
caa674f3
DD
2660#ifdef PERL_POISON
2661# define PERL_POISON_EXPR(x) x
2662#else
2663# define PERL_POISON_EXPR(x)
2664#endif
2665
ff68c719 2666#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2667
1b7e2294
KW
2668/*
2669=head1 Handy Values
2670
2671=for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2672
2673Returns the number of elements in the input C array (so you want your
2674zero-based indices to be less than but not equal to).
2675
2676=for apidoc Am|void *|C_ARRAY_END|void *a
2677
2678Returns a pointer to one element past the final element of the input C array.
2679
2680=cut
2681
2682C_ARRAY_END is one past the last: half-open/half-closed range, not
2683last-inclusive range.
2684*/
622913ab 2685#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2686#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2687
2cc61e15
DD
2688#ifdef NEED_VA_COPY
2689# ifdef va_copy
2690# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2691# elif defined(__va_copy)
2692# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2693# else
07798b17 2694# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2695# endif
2696#endif
2697
472d47bc
SB
2698/* convenience debug macros */
2699#ifdef USE_ITHREADS
2700#define pTHX_FORMAT "Perl interpreter: 0x%p"
2701#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2702#define pTHX_VALUE_ (void *)my_perl,
2703#define pTHX_VALUE (void *)my_perl
2704#define pTHX__VALUE_ ,(void *)my_perl,
2705#define pTHX__VALUE ,(void *)my_perl
472d47bc 2706#else
3609ea0d 2707#define pTHX_FORMAT
472d47bc 2708#define pTHX__FORMAT
3609ea0d 2709#define pTHX_VALUE_
472d47bc 2710#define pTHX_VALUE
3609ea0d 2711#define pTHX__VALUE_
472d47bc
SB
2712#define pTHX__VALUE
2713#endif /* USE_ITHREADS */
3609ea0d 2714
2acdbac1
NC
2715/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2716 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2717 CPAN::Unpack show any users outside the core. */
2718#ifdef PERL_CORE
dc6e8de0
A
2719# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2720 "Use of " s " is deprecated")
c9680906
A
2721# define deprecate_disappears_in(when,message) \
2722 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2723 message ", and will disappear in Perl " when)
ac641426
A
2724# define deprecate_fatal_in(when,message) \
2725 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2726 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2727#endif
2728
dfff4baf
BF
2729/* Internal macros to deal with gids and uids */
2730#ifdef PERL_CORE
2731
2732# if Uid_t_size > IVSIZE
2733# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2734# define SvUID(sv) SvNV(sv)
07798b17
AC
2735# elif Uid_t_sign <= 0
2736# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2737# define SvUID(sv) SvIV(sv)
dfff4baf 2738# else
07798b17
AC
2739# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2740# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2741# endif /* Uid_t_size */
2742
2743# if Gid_t_size > IVSIZE
2744# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2745# define SvGID(sv) SvNV(sv)
07798b17
AC
2746# elif Gid_t_sign <= 0
2747# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2748# define SvGID(sv) SvIV(sv)
dfff4baf 2749# else
07798b17
AC
2750# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2751# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2752# endif /* Gid_t_size */
2753
2754#endif
2755
6a5bc5ac 2756#endif /* PERL_HANDY_H_ */
9d745869 2757
e9a8c099 2758/*
14d04a33 2759 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2760 */