This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
handy.h: Generate compile error if macros called wrong
[perl5.git] / handy.h
CommitLineData
a0d0e21e 1/* handy.h
a687059c 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
da5d8dbb 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
a687059c 5 *
6e21c824
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8d063cd8 8 *
8d063cd8
LW
9 */
10
4650c663
KW
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
9d745869
KW
14#ifndef HANDY_H /* Guard against nested #inclusion */
15#define HANDY_H
16
85e6fe83 17#if !defined(__STDC__)
378cc40b
LW
18#ifdef NULL
19#undef NULL
20#endif
a687059c 21# define NULL 0
85e6fe83
LW
22#endif
23
24792b8d
NC
24#ifndef PERL_CORE
25# define Null(type) ((type)NULL)
954c1994
GS
26
27/*
ccfc67b7 28=head1 Handy Values
954c1994 29
3609ea0d 30=for apidoc AmU||Nullch
72d33970
FC
31Null character pointer. (No longer available when C<PERL_CORE> is
32defined.)
2307c6d0 33
954c1994 34=for apidoc AmU||Nullsv
72d33970 35Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
36
37=cut
38*/
39
24792b8d
NC
40# define Nullch Null(char*)
41# define Nullfp Null(PerlIO*)
42# define Nullsv Null(SV*)
43#endif
8d063cd8 44
641d3f0b
PP
45#ifdef TRUE
46#undef TRUE
47#endif
48#ifdef FALSE
49#undef FALSE
50#endif
51#define TRUE (1)
52#define FALSE (0)
53
cf3f0ffb
DM
54/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
55 * (compiler permitting) that casting away const-ness will give a warning;
56 * e.g.:
57 *
58 * const SV *sv = ...;
59 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
60 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
61 */
62
b1bc3f34 63#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
f065df04 64# define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
b1bc3f34
NC
65#else
66# define MUTABLE_PTR(p) ((void *) (p))
67#endif
68
a062e10d 69#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 70#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 71#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 72#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 73#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 74#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 75
f789f6a4 76#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
77# include <stdbool.h>
78# ifndef HAS_BOOL
79# define HAS_BOOL 1
80# endif
81#endif
82
8e84507e 83/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
84 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
85 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 86 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
87 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
88 g++ can be identified by __GNUG__.
89 Andy Dougherty February 2000
5d94fbed 90*/
3609ea0d 91#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 92# ifndef PERL_BOOL_AS_CHAR
5d94fbed 93# ifndef HAS_BOOL
c1d22f6b 94# define HAS_BOOL 1
5d94fbed 95# endif
f789f6a4 96# endif
5d94fbed
AD
97#endif
98
99#ifndef HAS_BOOL
f789f6a4
FC
100# ifdef bool
101# undef bool
102# endif
70d5cb32 103# define bool char
c1d22f6b 104# define HAS_BOOL 1
a687059c 105#endif
0d3e774c 106
236415e2
KW
107/* cast-to-bool. A simple (bool) cast may not do the right thing: if bool is
108 * defined as char for example, then the cast from int is
109 * implementation-defined (bool)!!(cbool) in a ternary triggers a bug in xlc on
110 * AIX */
18f5643b 111#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 112
46c6c7e2 113/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
114 * XXX Should really be a Configure probe, with HAS__FUNCTION__
115 * and FUNCTION__ as results.
116 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
46c6c7e2
JH
117#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
118# define FUNCTION__ __func__
119#else
9ea40801 120# if (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
be73ebb0 121 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
46c6c7e2
JH
122# define FUNCTION__ ""
123# else
124# define FUNCTION__ __FUNCTION__ /* Common extension. */
125# endif
126#endif
127
27d4fb96
PP
128/* XXX A note on the perl source internal type system. The
129 original intent was that I32 be *exactly* 32 bits.
130
131 Currently, we only guarantee that I32 is *at least* 32 bits.
132 Specifically, if int is 64 bits, then so is I32. (This is the case
133 for the Cray.) This has the advantage of meshing nicely with
134 standard library calls (where we pass an I32 and the library is
135 expecting an int), but the disadvantage that an I32 is not 32 bits.
136 Andy Dougherty August 1996
24fef2a7 137
dc45a647
MB
138 There is no guarantee that there is *any* integral type with
139 exactly 32 bits. It is perfectly legal for a system to have
140 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 141
dc45a647
MB
142 Similarly, there is no guarantee that I16 and U16 have exactly 16
143 bits.
693762b4 144
8e84507e
NIS
145 For dealing with issues that may arise from various 32/64-bit
146 systems, we will ask Configure to check out
8175356b 147
3609ea0d
JH
148 SHORTSIZE == sizeof(short)
149 INTSIZE == sizeof(int)
150 LONGSIZE == sizeof(long)
dc45a647 151 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 152 PTRSIZE == sizeof(void *)
dc45a647
MB
153 DOUBLESIZE == sizeof(double)
154 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 155
27d4fb96
PP
156*/
157
69512466
JH
158#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
159# include <inttypes.h>
dd0eed91
JH
160# ifdef INT32_MIN_BROKEN
161# undef INT32_MIN
162# define INT32_MIN (-2147483647-1)
163# endif
164# ifdef INT64_MIN_BROKEN
165# undef INT64_MIN
166# define INT64_MIN (-9223372036854775807LL-1)
167# endif
69512466
JH
168#endif
169
8175356b
JH
170typedef I8TYPE I8;
171typedef U8TYPE U8;
172typedef I16TYPE I16;
173typedef U16TYPE U16;
174typedef I32TYPE I32;
175typedef U32TYPE U32;
16d89be8
JH
176
177#ifdef HAS_QUAD
8175356b
JH
178typedef I64TYPE I64;
179typedef U64TYPE U64;
16d89be8 180#endif
8175356b 181
9c39b21c
JH
182/* INT64_C/UINT64_C are C99 from <stdint.h> (so they will not be
183 * available in strict C89 mode), but they are nice, so let's define
184 * them if necessary. */
0d3bd0a9 185#if defined(HAS_QUAD)
9c39b21c
JH
186# undef PeRl_INT64_C
187# undef PeRl_UINT64_C
188/* Prefer the native integer types (int and long) over long long
189 * (which is not C89) and Win32-specific __int64. */
190# if QUADKIND == QUAD_IS_INT && INTSIZE == 8
191# define PeRl_INT64_C(c) (c)
192# define PeRl_UINT64_C(c) CAT2(c,U)
193# endif
194# if QUADKIND == QUAD_IS_LONG && LONGSIZE == 8
195# define PeRl_INT64_C(c) CAT2(c,L)
196# define PeRl_UINT64_C(c) CAT2(c,UL)
197# endif
198# if QUADKIND == QUAD_IS_LONG_LONG && defined(HAS_LONG_LONG)
199# define PeRl_INT64_C(c) CAT2(c,LL)
200# define PeRl_UINT64_C(c) CAT2(c,ULL)
201# endif
202# if QUADKIND == QUAD_IS___INT64
203# define PeRl_INT64_C(c) CAT2(c,I64)
204# define PeRl_UINT64_C(c) CAT2(c,UI64)
205# endif
206# ifndef PeRl_INT64_C
702c92eb
JH
207# define PeRl_INT64_C(c) ((I64)(c)) /* last resort */
208# define PeRl_UINT64_C(c) ((U64)(c))
9c39b21c
JH
209# endif
210/* In OS X the INT64_C/UINT64_C are defined with LL/ULL, which will
211 * not fly with C89-pedantic gcc, so let's undefine them first so that
212 * we can redefine them with our native integer preferring versions. */
8eafc6eb 213# if defined(PERL_DARWIN) && defined(PERL_GCC_PEDANTIC)
9c39b21c
JH
214# undef INT64_C
215# undef UINT64_C
216# endif
217# ifndef INT64_C
218# define INT64_C(c) PeRl_INT64_C(c)
219# endif
220# ifndef UINT64_C
221# define UINT64_C(c) PeRl_UINT64_C(c)
222# endif
e8c95190 223#endif
dc750f2e 224
d8668976 225#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 226
5ff3f7a4
GS
227/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
228 Please search CHAR_MAX in perl.h for further details. */
229#define U8_MAX UINT8_MAX
230#define U8_MIN UINT8_MIN
231
5ff3f7a4
GS
232#define I16_MAX INT16_MAX
233#define I16_MIN INT16_MIN
234#define U16_MAX UINT16_MAX
235#define U16_MIN UINT16_MIN
236
5ff3f7a4
GS
237#define I32_MAX INT32_MAX
238#define I32_MIN INT32_MIN
0e983133
GS
239#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
240# define U32_MAX UINT32_MAX
241#else
242# define U32_MAX 4294967295U
243#endif
5ff3f7a4
GS
244#define U32_MIN UINT32_MIN
245
246#else
247
5c9fa16e
KA
248/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
249 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
250#define U8_MAX PERL_UCHAR_MAX
251#define U8_MIN PERL_UCHAR_MIN
79072805 252
27d4fb96
PP
253#define I16_MAX PERL_SHORT_MAX
254#define I16_MIN PERL_SHORT_MIN
255#define U16_MAX PERL_USHORT_MAX
256#define U16_MIN PERL_USHORT_MIN
79072805 257
c4f23d77 258#if LONGSIZE > 4
27d4fb96
PP
259# define I32_MAX PERL_INT_MAX
260# define I32_MIN PERL_INT_MIN
261# define U32_MAX PERL_UINT_MAX
262# define U32_MIN PERL_UINT_MIN
79072805 263#else
27d4fb96
PP
264# define I32_MAX PERL_LONG_MAX
265# define I32_MIN PERL_LONG_MIN
266# define U32_MAX PERL_ULONG_MAX
267# define U32_MIN PERL_ULONG_MIN
79072805
LW
268#endif
269
5ff3f7a4
GS
270#endif
271
58a9a5d5 272/* log(2) is pretty close to 0.30103, just in case anyone is grepping for it */
fc36a67e
PP
273#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
274#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
275#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
276
88794300 277/* Unused by core; should be deprecated */
ff68c719 278#define Ctl(ch) ((ch) & 037)
8d063cd8 279
4650c663
KW
280/* This is a helper macro to avoid preprocessor issues, replaced by nothing
281 * unless under DEBUGGING, where it expands to an assert of its argument,
282 * followed by a comma (hence the comma operator). If we just used a straight
3e94db23
JH
283 * assert(), we would get a comma with nothing before it when not DEBUGGING.
284 *
285 * We also use empty definition under Coverity since the __ASSERT__
286 * checks often check for things that Really Cannot Happen, and Coverity
287 * detects that and gets all excited. */
288
289#if defined(DEBUGGING) && !defined(__COVERITY__)
0f092d08
KW
290# define __ASSERT_(statement) assert(statement),
291#else
292# define __ASSERT_(statement)
293#endif
294
3fe05580
MHM
295/*
296=head1 SV-Body Allocation
297
298=for apidoc Ama|SV*|newSVpvs|const char* s
0c395ea5 299Like C<newSVpvn>, but takes a C<NUL>-terminated literal string instead of a
30a15352 300string/length pair.
3fe05580 301
84bafc02 302=for apidoc Ama|SV*|newSVpvs_flags|const char* s|U32 flags
0c395ea5 303Like C<newSVpvn_flags>, but takes a C<NUL>-terminated literal string instead of
30a15352 304a string/length pair.
84bafc02 305
3fe05580 306=for apidoc Ama|SV*|newSVpvs_share|const char* s
0c395ea5 307Like C<newSVpvn_share>, but takes a C<NUL>-terminated literal string instead of
30a15352 308a string/length pair and omits the hash parameter.
3fe05580 309
9dcc53ea 310=for apidoc Am|void|sv_catpvs_flags|SV* sv|const char* s|I32 flags
0c395ea5 311Like C<sv_catpvn_flags>, but takes a C<NUL>-terminated literal string instead
30a15352 312of a string/length pair.
9dcc53ea
Z
313
314=for apidoc Am|void|sv_catpvs_nomg|SV* sv|const char* s
0c395ea5
KW
315Like C<sv_catpvn_nomg>, but takes a C<NUL>-terminated literal string instead of
316a string/length pair.
9dcc53ea 317
7b4e7469 318=for apidoc Am|void|sv_catpvs|SV* sv|const char* s
0c395ea5
KW
319Like C<sv_catpvn>, but takes a C<NUL>-terminated literal string instead of a
320string/length pair.
3fe05580 321
9dcc53ea 322=for apidoc Am|void|sv_catpvs_mg|SV* sv|const char* s
0c395ea5 323Like C<sv_catpvn_mg>, but takes a C<NUL>-terminated literal string instead of a
9dcc53ea
Z
324string/length pair.
325
7b4e7469 326=for apidoc Am|void|sv_setpvs|SV* sv|const char* s
0c395ea5
KW
327Like C<sv_setpvn>, but takes a C<NUL>-terminated literal string instead of a
328string/length pair.
3fe05580 329
9dcc53ea 330=for apidoc Am|void|sv_setpvs_mg|SV* sv|const char* s
0c395ea5 331Like C<sv_setpvn_mg>, but takes a C<NUL>-terminated literal string instead of a
9dcc53ea
Z
332string/length pair.
333
334=for apidoc Am|SV *|sv_setref_pvs|const char* s
0c395ea5
KW
335Like C<sv_setref_pvn>, but takes a C<NUL>-terminated literal string instead of
336a string/length pair.
9dcc53ea 337
3fe05580
MHM
338=head1 Memory Management
339
340=for apidoc Ama|char*|savepvs|const char* s
0c395ea5 341Like C<savepvn>, but takes a C<NUL>-terminated literal string instead of a
30a15352 342string/length pair.
3fe05580 343
9dcc53ea
Z
344=for apidoc Ama|char*|savesharedpvs|const char* s
345A version of C<savepvs()> which allocates the duplicate string in memory
346which is shared between threads.
347
3fe05580
MHM
348=head1 GV Functions
349
350=for apidoc Am|HV*|gv_stashpvs|const char* name|I32 create
0c395ea5
KW
351Like C<gv_stashpvn>, but takes a C<NUL>-terminated literal string instead of a
352string/length pair.
3fe05580
MHM
353
354=head1 Hash Manipulation Functions
355
356=for apidoc Am|SV**|hv_fetchs|HV* tb|const char* key|I32 lval
0c395ea5
KW
357Like C<hv_fetch>, but takes a C<NUL>-terminated literal string instead of a
358string/length pair.
3fe05580
MHM
359
360=for apidoc Am|SV**|hv_stores|HV* tb|const char* key|NULLOK SV* val
0c395ea5
KW
361Like C<hv_store>, but takes a C<NUL>-terminated literal string instead of a
362string/length pair
3fe05580
MHM
363and omits the hash parameter.
364
510966aa
Z
365=head1 Lexer interface
366
367=for apidoc Amx|void|lex_stuff_pvs|const char *pv|U32 flags
368
0c395ea5
KW
369Like L</lex_stuff_pvn>, but takes a C<NUL>-terminated literal string instead of
370a string/length pair.
510966aa 371
3fe05580
MHM
372=cut
373*/
374
926bb54c
KW
375/* concatenating with "" ensures that only literal strings are accepted as
376 * argument */
26bfd924 377#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
2efa8cc7 378
926bb54c
KW
379/* note that STR_WITH_LEN() can't be used as argument to macros or functions
380 * that under some configurations might be macros, which means that it requires
381 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
ba3a79e7
GA
382 */
383
384/* STR_WITH_LEN() shortcuts */
385#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
386#define newSVpvs_flags(str,flags) \
387 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 388#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
389#define sv_catpvs_flags(sv, str, flags) \
390 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
391#define sv_catpvs_nomg(sv, str) \
392 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
393#define sv_catpvs(sv, str) \
394 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
395#define sv_catpvs_mg(sv, str) \
396 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 397#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
398#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
399#define sv_setref_pvs(rv, classname, str) \
400 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 401#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
402#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
403#define gv_stashpvs(str, create) \
404 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
405#define gv_fetchpvs(namebeg, add, sv_type) \
406 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
407#define gv_fetchpvn(namebeg, len, add, sv_type) \
408 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
409#define sv_catxmlpvs(dsv, str, utf8) \
410 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
a038e571
NC
411#define hv_fetchs(hv,key,lval) \
412 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
413 (lval) ? (HV_FETCH_JUST_SV | HV_FETCH_LVALUE) \
414 : HV_FETCH_JUST_SV, NULL, 0))
415
416#define hv_stores(hv,key,val) \
417 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
418 (HV_FETCH_ISSTORE|HV_FETCH_JUST_SV), (val), 0))
ba3a79e7 419
510966aa
Z
420#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
421
b96d8cd9
NC
422#define get_cvs(str, flags) \
423 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 424
954c1994 425/*
ccfc67b7
JH
426=head1 Miscellaneous Functions
427
954c1994 428=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
429Test two C<NUL>-terminated strings to see if they are different. Returns true
430or false.
954c1994
GS
431
432=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
433Test two C<NUL>-terminated strings to see if they are equal. Returns true or
434false.
954c1994
GS
435
436=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
437Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
438second, C<s2>. Returns true or false.
954c1994
GS
439
440=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
441Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
442equal to the second, C<s2>. Returns true or false.
954c1994
GS
443
444=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
445Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
446the second, C<s2>. Returns true or false.
954c1994
GS
447
448=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
449Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
450or equal to the second, C<s2>. Returns true or false.
954c1994
GS
451
452=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
453Test two C<NUL>-terminated strings to see if they are different. The C<len>
454parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
455wrapper for C<strncmp>).
456
457=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
458Test two C<NUL>-terminated strings to see if they are equal. The C<len>
459parameter indicates the number of bytes to compare. Returns true or false. (A
460wrapper for C<strncmp>).
954c1994 461
bd18bd40
KW
462=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
463Test two buffers (which may contain embedded C<NUL> characters, to see if they
464are equal. The C<len> parameter indicates the number of bytes to compare.
465Returns zero if equal, or non-zero if non-equal.
466
467=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
468Test two buffers (which may contain embedded C<NUL> characters, to see if they
469are not equal. The C<len> parameter indicates the number of bytes to compare.
470Returns zero if non-equal, or non-zero if equal.
471
954c1994
GS
472=cut
473*/
474
8d063cd8
LW
475#define strNE(s1,s2) (strcmp(s1,s2))
476#define strEQ(s1,s2) (!strcmp(s1,s2))
477#define strLT(s1,s2) (strcmp(s1,s2) < 0)
478#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
479#define strGT(s1,s2) (strcmp(s1,s2) > 0)
480#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
481#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
482#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
378cc40b 483
36477c24
PP
484#ifdef HAS_MEMCMP
485# define memNE(s1,s2,l) (memcmp(s1,s2,l))
486# define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
487#else
488# define memNE(s1,s2,l) (bcmp(s1,s2,l))
489# define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
490#endif
491
568a785a 492#define memEQs(s1, l, s2) \
8e6497b6 493 (sizeof(s2)-1 == l && memEQ(s1, ("" s2 ""), (sizeof(s2)-1)))
568a785a
NC
494#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
495
bbce6d69
PP
496/*
497 * Character classes.
498 *
499 * Unfortunately, the introduction of locales means that we
500 * can't trust isupper(), etc. to tell the truth. And when
501 * it comes to /\w+/ with tainting enabled, we *must* be able
502 * to trust our character classes.
503 *
504 * Therefore, the default tests in the text of Perl will be
505 * independent of locale. Any code that wants to depend on
506 * the current locale will use the tests that begin with "lc".
507 */
508
2304df62
AD
509#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
510# ifndef CTYPE256
511# define CTYPE256
512# endif
513#endif
514
954c1994 515/*
ccfc67b7 516
dcccc8ff 517=head1 Character classification
243effed
KW
518This section is about functions (really macros) that classify characters
519into types, such as punctuation versus alphabetic, etc. Most of these are
520analogous to regular expression character classes. (See
521L<perlrecharclass/POSIX Character Classes>.) There are several variants for
522each class. (Not all macros have all variants; each item below lists the
523ones valid for it.) None are affected by C<use bytes>, and only the ones
524with C<LC> in the name are affected by the current locale.
525
526The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
527C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
528represented by that octet is (or on non-ASCII platforms, corresponds to) an
529ASCII character in the named class based on platform, Unicode, and Perl rules.
530If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 531
6aff1f14
KW
532Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
533with no suffix C<"_A">.
4b9734bf
KW
534
535Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
536onto the platform. That is, the code points that are ASCII are unaffected,
537since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
538as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
539true when called with the code point 0xDF, which is a word character in both
4650c663 540ASCII and EBCDIC (though it represents different characters in each).
243effed 541
d0da05db 542Variant C<isFOO_uvchr> is like the C<isFOO_L1> variant, but accepts any UV code
243effed
KW
543point as input. If the code point is larger than 255, Unicode rules are used
544to determine if it is in the character class. For example,
d0da05db 545C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 546WITH MACRON in Unicode, and is a word character.
243effed 547
d0da05db 548Variant C<isFOO_utf8> is like C<isFOO_uvchr>, but the input is a pointer to a
dc6b0978
KW
549(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
550possibly containing embedded C<NUL> characters). The classification of just
551the first (possibly multi-byte) character in the string is tested.
243effed 552
31f05a37
KW
553Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
554result is based on the current locale, which is what C<LC> in the name stands
555for. If Perl can determine that the current locale is a UTF-8 locale, it uses
556the published Unicode rules; otherwise, it uses the C library function that
557gives the named classification. For example, C<isDIGIT_LC()> when not in a
558UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
559returned if the input won't fit into an octet. On some platforms where the C
560library function is known to be defective, Perl changes its result to follow
561the POSIX standard's rules.
243effed
KW
562
563Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
564returns the same as C<isFOO_LC> for input code points less than 256, and
565returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
566
b7d90381 567Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer
dc6b0978
KW
568to a (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
569possibly containing embedded C<NUL> characters). The classification of just
570the first (possibly multi-byte) character in the string is tested.
ccfc67b7 571
954c1994 572=for apidoc Am|bool|isALPHA|char ch
2787a470 573Returns a boolean indicating whether the specified character is an
6aff1f14 574alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
575See the L<top of this section|/Character classification> for an explanation of
576variants
d0da05db 577C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>, C<isALPHA_LC>,
243effed 578C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
8a58bdcf 579
15861f94
KW
580=for apidoc Am|bool|isALPHANUMERIC|char ch
581Returns a boolean indicating whether the specified character is a either an
6aff1f14 582alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
583See the L<top of this section|/Character classification> for an explanation of
584variants
d0da05db 585C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
e5ad6aba 586C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
15861f94
KW
587C<isALPHANUMERIC_LC_utf8>.
588
8a58bdcf
KW
589=for apidoc Am|bool|isASCII|char ch
590Returns a boolean indicating whether the specified character is one of the 128
243effed 591characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 592On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
593character corresponds to an ASCII character. Variants C<isASCII_A()> and
594C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
595See the L<top of this section|/Character classification> for an explanation of
596variants
d0da05db 597C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
243effed
KW
598C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
599library routine C<isascii()>. In these cases, the variants whose names contain
600C<LC> are the same as the corresponding ones without.
601
d98532ea
KW
602Also note, that because all ASCII characters are UTF-8 invariant (meaning they
603have the exact same representation (always a single byte) whether encoded in
604UTF-8 or not), C<isASCII> will give the correct results when called with any
605byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
606work properly on any string encoded or not in UTF-8.
607
243effed
KW
608=for apidoc Am|bool|isBLANK|char ch
609Returns a boolean indicating whether the specified character is a
6aff1f14 610character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
611See the L<top of this section|/Character classification> for an explanation of
612variants
d0da05db 613C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>, C<isBLANK_LC>,
243effed
KW
614C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
615platforms do not have the C library routine C<isblank()>. In these cases, the
616variants whose names contain C<LC> are the same as the corresponding ones
617without.
618
619=for apidoc Am|bool|isCNTRL|char ch
620Returns a boolean indicating whether the specified character is a
6aff1f14 621control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
622See the L<top of this section|/Character classification> for an explanation of
623variants
d0da05db 624C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
6aff1f14
KW
625C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
626On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 627
954c1994 628=for apidoc Am|bool|isDIGIT|char ch
2787a470 629Returns a boolean indicating whether the specified character is a
6aff1f14 630digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 631Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
632See the L<top of this section|/Character classification> for an explanation of
633variants
d0da05db 634C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
243effed
KW
635C<isDIGIT_LC_utf8>.
636
637=for apidoc Am|bool|isGRAPH|char ch
638Returns a boolean indicating whether the specified character is a
6aff1f14 639graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff
KW
640See the L<top of this section|/Character classification> for an explanation of
641variants
d0da05db 642C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
243effed 643C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
954c1994 644
0c82b6df 645=for apidoc Am|bool|isLOWER|char ch
2787a470 646Returns a boolean indicating whether the specified character is a
6aff1f14 647lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
648See the L<top of this section|/Character classification> for an explanation of
649variants
d0da05db 650C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>, C<isLOWER_LC>,
243effed 651C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
0c82b6df 652
c99e91e9 653=for apidoc Am|bool|isOCTAL|char ch
2787a470 654Returns a boolean indicating whether the specified character is an
6aff1f14 655octal digit, [0-7].
243effed
KW
656The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
657C<isOCTAL>.
658
659=for apidoc Am|bool|isPUNCT|char ch
660Returns a boolean indicating whether the specified character is a
6aff1f14
KW
661punctuation character, analogous to C<m/[[:punct:]]/>.
662Note that the definition of what is punctuation isn't as
243effed
KW
663straightforward as one might desire. See L<perlrecharclass/POSIX Character
664Classes> for details.
dcccc8ff
KW
665See the L<top of this section|/Character classification> for an explanation of
666variants
d0da05db 667C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
243effed 668C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
c99e91e9 669
0c82b6df 670=for apidoc Am|bool|isSPACE|char ch
2787a470 671Returns a boolean indicating whether the specified character is a
6aff1f14 672whitespace character. This is analogous
398d098a 673to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 674this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
675locale forms of this macro (the ones with C<LC> in their names) matched
676precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
677in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
678(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
679See the L<top of this section|/Character classification> for an explanation of
680variants
d0da05db 681C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>, C<isSPACE_LC>,
243effed 682C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
0c82b6df 683
398d098a
KW
684=for apidoc Am|bool|isPSXSPC|char ch
685(short for Posix Space)
779cf272
KW
686Starting in 5.18, this is identical in all its forms to the
687corresponding C<isSPACE()> macros.
398d098a
KW
688The locale forms of this macro are identical to their corresponding
689C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
690non-locale forms differ from their C<isSPACE()> forms only in that the
691C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
692Otherwise they are identical. Thus this macro is analogous to what
693C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 694See the L<top of this section|/Character classification> for an explanation of
d0da05db 695variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
b7d90381 696C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
398d098a 697
954c1994 698=for apidoc Am|bool|isUPPER|char ch
2787a470 699Returns a boolean indicating whether the specified character is an
6aff1f14 700uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff
KW
701See the L<top of this section|/Character classification> for an explanation of
702variants
d0da05db 703C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>, C<isUPPER_LC>,
243effed 704C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
954c1994 705
243effed 706=for apidoc Am|bool|isPRINT|char ch
8eea39dd 707Returns a boolean indicating whether the specified character is a
6aff1f14 708printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
709See the L<top of this section|/Character classification> for an explanation of
710variants
d0da05db 711C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>, C<isPRINT_LC>,
243effed
KW
712C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
713
714=for apidoc Am|bool|isWORDCHAR|char ch
715Returns a boolean indicating whether the specified character is a character
716that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
717in a regular expression. A word character is an alphabetic character, a
718decimal digit, a connecting punctuation character (such as an underscore), or
719a "mark" character that attaches to one of those (like some sort of accent).
720C<isALNUM()> is a synonym provided for backward compatibility, even though a
721word character includes more than the standard C language meaning of
722alphanumeric.
dcccc8ff
KW
723See the L<top of this section|/Character classification> for an explanation of
724variants
d0da05db 725C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and C<isWORDCHAR_utf8>.
1a83413c
KW
726C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
727described there, but additionally include the platform's native underscore.
8a58bdcf
KW
728
729=for apidoc Am|bool|isXDIGIT|char ch
730Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
731digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
732and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
733See the L<top of this section|/Character classification> for an explanation of
734variants
d0da05db 735C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
243effed
KW
736C<isXDIGIT_LC_utf8>.
737
3c3ecf18
KW
738=for apidoc Am|bool|isIDFIRST|char ch
739Returns a boolean indicating whether the specified character can be the first
740character of an identifier. This is very close to, but not quite the same as
741the official Unicode property C<XID_Start>. The difference is that this
742returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
743See the L<top of this section|/Character classification> for an explanation of
744variants
d0da05db 745C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
3c3ecf18
KW
746C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
747
748=for apidoc Am|bool|isIDCONT|char ch
749Returns a boolean indicating whether the specified character can be the
750second or succeeding character of an identifier. This is very close to, but
751not quite the same as the official Unicode property C<XID_Continue>. The
752difference is that this returns true only if the input character also matches
dcccc8ff
KW
753L</isWORDCHAR>. See the L<top of this section|/Character classification> for
754an
d0da05db 755explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
3c3ecf18
KW
756C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
757C<isIDCONT_LC_utf8>.
758
243effed 759=head1 Miscellaneous Functions
8eea39dd 760
95a59cab 761=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 762Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
763Behaviour is only well defined when isXDIGIT(*str) is true.
764
e7c1e6c1
KW
765=head1 Character case changing
766
1f607577
KW
767=for apidoc Am|U8|toUPPER|U8 ch
768Converts the specified character to uppercase. If the input is anything but an
769ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 770C<toUPPER_A> is equivalent.
954c1994 771
d0da05db
KW
772=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
773Converts the code point C<cp> to its uppercase version, and
774stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
775point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
776that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
777bytes since the uppercase version may be longer than the original character.
778
779The first code point of the uppercased version is returned
780(but note, as explained just above, that there may be more.)
781
782=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
783Converts the UTF-8 encoded character at C<p> to its uppercase version, and
784stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
785that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
786bytes since the uppercase version may be longer than the original character.
787
788The first code point of the uppercased version is returned
789(but note, as explained just above, that there may be more.)
790
791The input character at C<p> is assumed to be well-formed.
792
25200305
KW
793=for apidoc Am|U8|toFOLD|U8 ch
794Converts the specified character to foldcase. If the input is anything but an
795ASCII uppercase character, that input character itself is returned. Variant
796C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 797Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 798
d0da05db
KW
799=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
800Converts the code point C<cp> to its foldcase version, and
801stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
802point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
803that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
804bytes since the foldcase version may be longer than the original character.
805
806The first code point of the foldcased version is returned
807(but note, as explained just above, that there may be more.)
808
809=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
810Converts the UTF-8 encoded character at C<p> to its foldcase version, and
811stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
812that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
813bytes since the foldcase version may be longer than the original character.
814
815The first code point of the foldcased version is returned
816(but note, as explained just above, that there may be more.)
817
818The input character at C<p> is assumed to be well-formed.
819
820=for apidoc Am|U8|toLOWER|U8 ch
821Converts the specified character to lowercase. If the input is anything but an
822ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 823C<toLOWER_A> is equivalent.
954c1994 824
1f607577 825=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
826Converts the specified Latin1 character to lowercase. The results are
827undefined if the input doesn't fit in a byte.
1f607577
KW
828
829=for apidoc Am|U8|toLOWER_LC|U8 ch
830Converts the specified character to lowercase using the current locale's rules,
831if possible; otherwise returns the input character itself.
832
d0da05db
KW
833=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
834Converts the code point C<cp> to its lowercase version, and
835stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
836point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
837that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
838bytes since the lowercase version may be longer than the original character.
839
840The first code point of the lowercased version is returned
841(but note, as explained just above, that there may be more.)
842
843=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
844Converts the UTF-8 encoded character at C<p> to its lowercase version, and
845stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
846that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
847bytes since the lowercase version may be longer than the original character.
848
849The first code point of the lowercased version is returned
850(but note, as explained just above, that there may be more.)
851
852The input character at C<p> is assumed to be well-formed.
853
25200305
KW
854=for apidoc Am|U8|toTITLE|U8 ch
855Converts the specified character to titlecase. If the input is anything but an
856ASCII lowercase character, that input character itself is returned. Variant
b7d90381 857C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 858range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 859not a concept used in locale handling, so there is no functionality for that.)
25200305 860
d0da05db
KW
861=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
862Converts the code point C<cp> to its titlecase version, and
863stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
864point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
865that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
866bytes since the titlecase version may be longer than the original character.
867
868The first code point of the titlecased version is returned
869(but note, as explained just above, that there may be more.)
870
871=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
872Converts the UTF-8 encoded character at C<p> to its titlecase version, and
873stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
874that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
875bytes since the titlecase version may be longer than the original character.
876
877The first code point of the titlecased version is returned
878(but note, as explained just above, that there may be more.)
879
880The input character at C<p> is assumed to be well-formed.
881
954c1994 882=cut
353c9b6f 883
d0da05db 884XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
25200305 885really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 886
8a58bdcf 887Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
888patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
889
954c1994
GS
890*/
891
7c062697
KW
892/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
893 * is known only in the perl core, and this macro can be called from outside
894 * that */
895#ifdef HAS_QUAD
896# define WIDEST_UTYPE U64TYPE
897#else
898# define WIDEST_UTYPE U32
899#endif
900
3912bc88
KW
901/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
902 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
903 * bits of information are lost even on a 64-bit machine, but to get the
904 * compiler to optimize it out if possible. This is because Configure makes
905 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
906 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
907 *
908 * For Coverity, be always true, because otherwise Coverity thinks
909 * it finds several expressions that are always true, independent
910 * of operands. Well, they are, but that is kind of the point.
220c71bf 911 */
7e75d1a1 912#ifndef __COVERITY__
6c5b02ac
KW
913 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
914 * pointer) */
915#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
916 || !(((WIDEST_UTYPE)(c | 0)) & ~0xFF))
7e75d1a1
JH
917#else
918#define FITS_IN_8_BITS(c) (1)
919#endif
cf301eb7 920
41f43cc2 921#ifdef EBCDIC
b6340bd0 922# ifndef _ALL_SOURCE
0852beac
KW
923 /* The native libc isascii() et.al. functions return the wrong results
924 * on at least z/OS unless this is defined. */
b6340bd0
KW
925# error _ALL_SOURCE should probably be defined
926# endif
41f43cc2 927#else
0852beac
KW
928 /* There is a simple definition of ASCII for ASCII platforms. But the
929 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59
LM
930 * other macros below.
931 * The '| 0' part ensures that c is an integer (and not e.g. a pointer) */
932# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
933#endif
934
c2da0b36
KW
935/* The lower 3 bits in both the ASCII and EBCDIC representations of '0' are 0,
936 * and the 8 possible permutations of those bits exactly comprise the 8 octal
937 * digits */
3d33b3e3 938#define isOCTAL_A(c) cBOOL(FITS_IN_8_BITS(c) && (0xF8 & (c)) == '0')
c2da0b36 939
9fb1bf9d 940#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 941
a500dc72
KW
942/* Character class numbers. For internal core Perl use only. The ones less
943 * than 32 are used in PL_charclass[] and the ones up through the one that
944 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
945 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
946 * their actual definitions are here. If that file has a name not used here,
947 * it won't compile.
1709d539
KW
948 *
949 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
950 * frequency of their use. This gives a slight edge to exiting a loop earlier
951 * (in reginclass() in regexec.c) */
1709d539
KW
952# define _CC_WORDCHAR 0 /* \w and [:word:] */
953# define _CC_DIGIT 1 /* \d and [:digit:] */
954# define _CC_ALPHA 2 /* [:alpha:] */
955# define _CC_LOWER 3 /* [:lower:] */
956# define _CC_UPPER 4 /* [:upper:] */
957# define _CC_PUNCT 5 /* [:punct:] */
958# define _CC_PRINT 6 /* [:print:] */
15861f94 959# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 960# define _CC_GRAPH 8 /* [:graph:] */
b0d691b2 961# define _CC_CASED 9 /* [:lower:] and [:upper:] under /i */
1709d539 962
b0d691b2 963#define _FIRST_NON_SWASH_CC 10
1709d539
KW
964/* The character classes above are implemented with swashes. The second group
965 * (just below) contains the ones implemented without. These are also sorted
966 * in rough order of the frequency of their use, except that \v should be last,
967 * as it isn't a real Posix character class, and some (small) inefficiencies in
968 * regular expression handling would be introduced by putting it in the middle
969 * of those that are. Also, cntrl and ascii come after the others as it may be
970 * useful to group these which have no members that match above Latin1, (or
971 * above ASCII in the latter case) */
972
779cf272 973# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
974# define _CC_BLANK 11 /* [:blank:] */
975# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
976# define _CC_CNTRL 13 /* [:cntrl:] */
977# define _CC_ASCII 14 /* [:ascii:] */
978# define _CC_VERTSPACE 15 /* \v */
1709d539 979
a0947d7b
KW
980# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
981
1709d539 982/* The members of the third group below do not need to be coordinated with data
3ffc8c70 983 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
984# define _CC_IDFIRST 16
985# define _CC_CHARNAME_CONT 17
986# define _CC_NONLATIN1_FOLD 18
987# define _CC_NONLATIN1_SIMPLE_FOLD 19
988# define _CC_QUOTEMETA 20
989# define _CC_NON_FINAL_FOLD 21
990# define _CC_IS_IN_SOME_FOLD 22
991# define _CC_MNEMONIC_CNTRL 23
073c22b3
KW
992
993/* This next group is only used on EBCDIC platforms, so theoretically could be
994 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
995# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
996# define _CC_UTF8_IS_START 29
997# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
998# define _CC_UTF8_IS_CONTINUATION 31
999/* Unused: 24-27
f4cdb42c
KW
1000 * If more bits are needed, one could add a second word for non-64bit
1001 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1002 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1003 * is used only for optimization (as of this writing), and differs in the
1004 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1005 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1006 * would just cause /i regexes which match them to run less efficiently.
1007 * Similarly the EBCDIC-only bits are used just for speed, and could be
1008 * replaced by other means */
96ac0975 1009
3a371f2f
KW
1010#if defined(PERL_CORE) || defined(PERL_EXT)
1011/* An enum version of the character class numbers, to help compilers
1012 * optimize */
1013typedef enum {
3a371f2f 1014 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1015 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1016 _CC_ENUM_ASCII = _CC_ASCII,
1017 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1018 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1019 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1020 _CC_ENUM_DIGIT = _CC_DIGIT,
1021 _CC_ENUM_GRAPH = _CC_GRAPH,
1022 _CC_ENUM_LOWER = _CC_LOWER,
1023 _CC_ENUM_PRINT = _CC_PRINT,
1024 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1025 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1026 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1027 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1028 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1029 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1030} _char_class_number;
1031#endif
1032
63c61c3f 1033#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1034#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1035
b7d90381
KW
1036#if defined(PERL_IN_UTF8_C) \
1037 || defined(PERL_IN_REGCOMP_C) \
1038 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1039# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1040 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1041 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1042 #error Need to adjust order of swash_property_names[]
1043# endif
1044
b44daeeb
KW
1045/* This is declared static in each of the few files that this is #defined for
1046 * to keep them from being publicly accessible. Hence there is a small amount
1047 * of wasted space */
1048
922e8cb4
KW
1049static const char* const swash_property_names[] = {
1050 "XPosixWord",
1051 "XPosixDigit",
1052 "XPosixAlpha",
1053 "XPosixLower",
1054 "XPosixUpper",
1055 "XPosixPunct",
1056 "XPosixPrint",
1057 "XPosixAlnum",
b0d691b2
KW
1058 "XPosixGraph",
1059 "Cased"
922e8cb4
KW
1060};
1061#endif
1062
6635f04f 1063START_EXTERN_C
96ac0975
NC
1064# ifdef DOINIT
1065EXTCONST U32 PL_charclass[] = {
1066# include "l1_char_class_tab.h"
1067};
1068
1069# else /* ! DOINIT */
1070EXTCONST U32 PL_charclass[];
1071# endif
6635f04f 1072END_EXTERN_C
96ac0975 1073
265c1f46 1074 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1075# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1076
1077 /* For internal core Perl use only: the base macro for defining macros like
1078 * isALPHA */
ff7ecfc3 1079# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1080 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1081
f4cdb42c
KW
1082 /* The mask for the _A versions of the macros; it just adds in the bit for
1083 * ASCII. */
1084# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1085
4650c663
KW
1086 /* For internal core Perl use only: the base macro for defining macros like
1087 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1088 * the ASCII bit are present */
b7d90381
KW
1089# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1090 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1091 == _CC_mask_A(classnum)))
f4cdb42c 1092
f4cdb42c 1093# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1094# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1095# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1096# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1097# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1098# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1099# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1100# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1101# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1102# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1103# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1104# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1105# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1106 */
d95f8b6a 1107# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1108# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1109# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1110# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1111
1112 /* continuation character for legal NAME in \N{NAME} */
1113# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1114
1115# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1116# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1117# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1118# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1119# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1120# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1121# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1122# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1123# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1124# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1125
0852beac
KW
1126# ifdef EBCDIC
1127# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1128# endif
1129
f12c0118
KW
1130 /* Participates in a single-character fold with a character above 255 */
1131# 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)))
1132
1133 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1134# 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 1135
4eeeb416 1136# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1137# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1138 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1139# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1140 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1141# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1142 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1143#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1144
1145 /* If we don't have perl.h, we are compiling a utility program. Below we
1146 * hard-code various macro definitions that wouldn't otherwise be available
0852beac
KW
1147 * to it. Most are coded based on first principals. First some ones common
1148 * to both ASCII and EBCDIC */
1149# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1150# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1151# define isSPACE_A(c) (isBLANK_A(c) \
1152 || (c) == '\n' \
1153 || (c) == '\r' \
1154 || (c) == '\v' \
1155 || (c) == '\f')
1156# ifdef EBCDIC /* There are gaps between 'i' and 'j'; 'r' and 's'. Same
1157 for uppercase. This is ordered to exclude most things
1158 early */
1159# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1160 && ((c) <= 'i' \
1161 || ((c) >= 'j' && (c) <= 'r') \
1162 || (c) >= 's'))
1163# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1164 && ((c) <= 'I' \
1165 || ((c) >= 'J' && (c) <= 'R') \
1166 || (c) >= 'S'))
1167# else /* ASCII platform. */
9fb1bf9d 1168# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z')
2181da29 1169# define isUPPER_A(c) ((c) <= 'Z' && (c) >= 'A')
0852beac
KW
1170# endif
1171
1172 /* Some more ASCII, non-ASCII common definitions */
a4d7a999
KW
1173# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1174# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1175# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac
KW
1176# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1177# define isXDIGIT_A(c) (isDIGIT_A(c) \
1178 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1179 || ((c) <= 'F' && (c) >= 'A'))
3ded5eb0 1180
0852beac
KW
1181# ifdef EBCDIC
1182# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1183 || (c) == '#' || (c) == '$' || (c) == '%' \
1184 || (c) == '&' || (c) == '\'' || (c) == '(' \
1185 || (c) == ')' || (c) == '*' || (c) == '+' \
1186 || (c) == ',' || (c) == '.' || (c) == '/' \
1187 || (c) == ':' || (c) == ';' || (c) == '<' \
1188 || (c) == '=' || (c) == '>' || (c) == '?' \
1189 || (c) == '@' || (c) == '[' || (c) == '\\' \
1190 || (c) == ']' || (c) == '^' || (c) == '_' \
1191 || (c) == '`' || (c) == '{' || (c) == '|' \
1192 || (c) == '}' || (c) == '~')
1193# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1194# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1195
1196# ifdef QUESTION_MARK_CTRL
1197# define _isQMC(c) ((c) == QUESTION_MARK_CTRL)
1198# else
1199# define _isQMC(c) 0
1200# endif
1201
1202 /* I (khw) can't think of a way to define all the ASCII controls
1203 * without resorting to a libc (locale-sensitive) call. But we know
1204 * that all controls but the question-mark one are in the range 0-0x3f.
1205 * This makes sure that all the controls that have names are included,
1206 * and all controls that are also considered ASCII in the locale. This
1207 * may include more or fewer than what it actually should, but the
1208 * wrong ones are less-important controls, so likely won't impact
1209 * things (keep in mind that this is compiled only if perl.h isn't
1210 * available). The question mark control is included if available */
1211# define isCNTRL_A(c) (((c) < 0x40 && isascii(c)) \
1212 || (c) == '\0' || (c) == '\a' || (c) == '\b' \
1213 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1214 || (c) == '\t' || (c) == '\v' || _isQMC(c))
1215
1216# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1217# else /* ASCII platform; things are simpler, and isASCII has already
1218 been defined */
1219# define isGRAPH_A(c) (((c) > ' ' && (c) < 127))
1220# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1221# define isPUNCT_A(c) (isGRAPH_A(c) && (! isALPHANUMERIC_A(c)))
1222# define isCNTRL_A(c) (isASCII(c) && (! isPRINT_A(c)))
1223# endif
1224
3ffc8c70 1225 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1226 * during debugging, and it seems best to keep them. We may be called
1227 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1228 * do anything anyway, so make it not a problem */
1229# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1230# define NATIVE_TO_LATIN1(ch) (ch)
1231# endif
3ded5eb0
KW
1232# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1233# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1234# define isBLANK_L1(c) (isBLANK_A(c) \
1235 || (FITS_IN_8_BITS(c) \
1236 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1237# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1238# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1239# define isLOWER_L1(c) (isLOWER_A(c) \
1240 || (FITS_IN_8_BITS(c) \
1241 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1242 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1243 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1244 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1245 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1246# define isPRINT_L1(c) (isPRINT_A(c) \
1247 || (FITS_IN_8_BITS(c) \
1248 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1249# define isPUNCT_L1(c) (isPUNCT_A(c) \
1250 || (FITS_IN_8_BITS(c) \
1251 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1252 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1253 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1254 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1255 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1256 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1257 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1258# define isSPACE_L1(c) (isSPACE_A(c) \
1259 || (FITS_IN_8_BITS(c) \
1260 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1261 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1262# define isUPPER_L1(c) (isUPPER_A(c) \
1263 || (FITS_IN_8_BITS(c) \
1264 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1265 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1266 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1267# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1268# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1269# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1270 || isBLANK_L1(c) \
1271 || (c) == '-' \
1272 || (c) == '(' \
1273 || (c) == ')')
1274 /* The following are not fully accurate in the above-ASCII range. I (khw)
1275 * don't think it's necessary to be so for the purposes where this gets
1276 * compiled */
1277# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1278# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1279
1280 /* And these aren't accurate at all. They are useful only for above
1281 * Latin1, which utilities and bootstrapping don't deal with */
1282# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1283# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1284# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1285
1286 /* Many of the macros later in this file are defined in terms of these. By
1287 * implementing them with a function, which converts the class number into
1288 * a call to the desired macro, all of the later ones work. However, that
1289 * function won't be actually defined when building a utility program (no
1290 * perl.h), and so a compiler error will be generated if one is attempted
1291 * to be used. And the above-Latin1 code points require Unicode tables to
1292 * be present, something unlikely to be the case when bootstrapping */
1293# define _generic_isCC(c, classnum) \
1294 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1295# define _generic_isCC_A(c, classnum) \
1296 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1297#endif /* End of no perl.h H_PERL */
8a58bdcf 1298
e66b99e9
KW
1299#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1300#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1301#define isASCII_A(c) isASCII(c)
1302#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1303#define isBLANK(c) isBLANK_A(c)
1304#define isCNTRL(c) isCNTRL_A(c)
1305#define isDIGIT(c) isDIGIT_A(c)
1306#define isGRAPH(c) isGRAPH_A(c)
1307#define isIDFIRST(c) isIDFIRST_A(c)
1308#define isLOWER(c) isLOWER_A(c)
1309#define isPRINT(c) isPRINT_A(c)
779cf272 1310#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1311#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1312#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1313#define isPUNCT(c) isPUNCT_A(c)
1314#define isSPACE(c) isSPACE_A(c)
1315#define isUPPER(c) isUPPER_A(c)
1316#define isWORDCHAR(c) isWORDCHAR_A(c)
1317#define isXDIGIT(c) isXDIGIT_A(c)
1318
1319/* ASCII casing. These could also be written as
1320 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1321 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1322 which uses table lookup and mask instead of subtraction. (This would
1323 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1324#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1325#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1326
25200305
KW
1327/* In the ASCII range, these are equivalent to what they're here defined to be.
1328 * But by creating these definitions, other code doesn't have to be aware of
1329 * this detail */
1330#define toFOLD(c) toLOWER(c)
25200305
KW
1331#define toTITLE(c) toUPPER(c)
1332
c753c8d3
KW
1333#define toLOWER_A(c) toLOWER(c)
1334#define toUPPER_A(c) toUPPER(c)
25200305
KW
1335#define toFOLD_A(c) toFOLD(c)
1336#define toTITLE_A(c) toTITLE(c)
1a0901db 1337
4650c663 1338/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1339#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1340 ? (c) \
f4cd282c 1341 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1342#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1343
1a0901db 1344/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1345 * all mapped to one of them, and these need special handling; returns the
1346 * input itself if is out-of-range */
b2bf251f 1347#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1348 ? (c) \
f4cd282c 1349 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1350#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1351
beab9ebe
KW
1352/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1353
1354/* For internal core Perl use only: the base macro for defining macros like
1355 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1356 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1357 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1358 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1359 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1360 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1361 * code points under locale rules, use the macros based on _generic_LC_uvchr
1362 * instead of this. */
beab9ebe
KW
1363#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1364 (! FITS_IN_8_BITS(c) \
1365 ? 0 \
31f05a37
KW
1366 : IN_UTF8_CTYPE_LOCALE \
1367 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1368 : cBOOL(non_utf8))
1369
1370/* For internal core Perl use only: a helper macro for defining macros like
1371 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1372 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1373 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1374 * _generic_LC_base for more info */
1375#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1376 _generic_LC_base(c,utf8_locale_classnum, \
1377 non_utf8_func( (_LC_CAST) (c)))
1378
1379/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1380 * 'c' is the platform's native underscore character */
1381#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1382 _generic_LC_base(c, utf8_locale_classnum, \
1383 (non_utf8_func( (_LC_CAST) (c)) \
1384 || (char)(c) == '_'))
1385
1386/* These next three are also for internal core Perl use only: case-change
1387 * helper macros */
1388#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1389 ? (c) \
31f05a37
KW
1390 : (IN_UTF8_CTYPE_LOCALE) \
1391 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1392 : (cast)function((cast)(c)))
beab9ebe 1393
31f05a37
KW
1394/* Note that the result can be larger than a byte in a UTF-8 locale. It
1395 * returns a single value, so can't adequately return the upper case of LATIN
1396 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1397 * values "SS"); instead it asserts against that under DEBUGGING, and
1398 * otherwise returns its input */
beab9ebe
KW
1399#define _generic_toUPPER_LC(c, function, cast) \
1400 (! FITS_IN_8_BITS(c) \
1401 ? (c) \
31f05a37 1402 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1403 ? (cast)function((cast)(c)) \
31f05a37
KW
1404 : ((((U8)(c)) == MICRO_SIGN) \
1405 ? GREEK_CAPITAL_LETTER_MU \
1406 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1407 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1408 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1409 ? (__ASSERT_(0) (c)) \
1410 : PL_mod_latin1_uc[ (U8) (c) ])))))
1411
1412/* Note that the result can be larger than a byte in a UTF-8 locale. It
1413 * returns a single value, so can't adequately return the fold case of LATIN
1414 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1415 * values "ss"); instead it asserts against that under DEBUGGING, and
1416 * otherwise returns its input */
beab9ebe 1417#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1418 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1419 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1420 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1421 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1422 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1423
84061b6a 1424/* Use the libc versions for these if available. */
f05550c0 1425#if defined(HAS_ISASCII)
84061b6a
KW
1426# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1427#else
1428# define isASCII_LC(c) isASCII(c)
1429#endif
1430
f05550c0 1431#if defined(HAS_ISBLANK)
84061b6a 1432# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1433#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1434# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1435#endif
1436
f05550c0 1437#define _LC_CAST U8
bbce6d69 1438
f05550c0 1439#ifdef WIN32
375f5f06
KW
1440 /* The Windows functions don't bother to follow the POSIX standard, which
1441 * for example says that something can't both be a printable and a control.
1442 * But Windows treats the \t control as a printable, and does such things
1443 * as making superscripts into both digits and punctuation. This tames
1444 * these flaws by assuming that the definitions of both controls and space
1445 * are correct, and then making sure that other definitions don't have
1446 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1447 * Not all possible weirdnesses are checked for, just the ones that were
1448 * detected on actual Microsoft code pages */
1449
b7d90381
KW
1450# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1451# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1452
1453# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1454 && isALPHANUMERIC_LC(c))
1455# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1456 ! isPUNCT_LC(c))
1457# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1458 isALPHANUMERIC_LC(c))
1459# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1460# define isIDFIRST_LC(c) (((c) == '_') \
1461 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1462# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1463# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1464# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1465# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1466# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1467# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1468 && isALPHANUMERIC_LC(c))
f05550c0
BF
1469
1470# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1471# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1472# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1473
1474#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1475 /* For most other platforms */
beab9ebe 1476
f05550c0
BF
1477# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1478# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1479# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1480# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1481# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1482# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1483# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1484# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1485# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1486# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1487# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1488# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1489# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1490
1491
1492# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1493# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1494# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1495
1496#else /* The final fallback position */
1497
b7d90381
KW
1498# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1499# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1500# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1501# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1502# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1503# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1504# define isLOWER_LC(c) (isascii(c) && islower(c))
1505# define isPRINT_LC(c) (isascii(c) && isprint(c))
1506# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1507# define isSPACE_LC(c) (isascii(c) && isspace(c))
1508# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1509# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1510# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1511
1512# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1513# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1514# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1515
f05550c0 1516#endif
55204971 1517
eba68aa0
KW
1518#define isIDCONT(c) isWORDCHAR(c)
1519#define isIDCONT_A(c) isWORDCHAR_A(c)
1520#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1521#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1522#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1523
4650c663 1524/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1525 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1526 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1527 * classes where there is a macro or function 'above_latin1' that takes the
1528 * single argument 'c' and returns the desired value. These exist for those
1529 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1530 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1531 * for classes where that overhead is faster than a direct lookup.
d0da05db 1532 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1533 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1534 * have duplicate checks here, so could create versions of the macros that
1535 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1536
1537/* Note that all ignore 'use bytes' */
d0da05db 1538#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
cd500f2f
KW
1539 ? _generic_isCC(c, classnum) \
1540 : above_latin1(c))
d0da05db 1541#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
cd500f2f 1542 ? _generic_isCC(c, classnum) \
922e8cb4 1543 : _is_uni_FOO(classnum, c))
d0da05db
KW
1544#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1545#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1546#define isASCII_uvchr(c) isASCII(c)
1547#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1548#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1549#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1550#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1551#define isIDCONT_uvchr(c) _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1552#define isIDFIRST_uvchr(c) _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
1553#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1554#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1555
1556#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1557#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1558#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1559
1560#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1561#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1562#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1563#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1564
1565#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1566#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1567#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1568#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1569
1570/* For backwards compatibility, even though '_uni' should mean official Unicode
1571 * code points, in Perl it means native for those below 256 */
1572#define isALPHA_uni(c) isALPHA_uvchr(c)
1573#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1574#define isASCII_uni(c) isASCII_uvchr(c)
1575#define isBLANK_uni(c) isBLANK_uvchr(c)
1576#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1577#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1578#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1579#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1580#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1581#define isLOWER_uni(c) isLOWER_uvchr(c)
1582#define isPRINT_uni(c) isPRINT_uvchr(c)
1583#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1584#define isSPACE_uni(c) isSPACE_uvchr(c)
1585#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1586#define isUPPER_uni(c) isUPPER_uvchr(c)
1587#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1588#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1589#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1590#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1591#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1592#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1593#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1594
4650c663
KW
1595/* For internal core Perl use only: the base macros for defining macros like
1596 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1597 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1598 * simple class definitions; the other for more complex. These are like
d0da05db 1599 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1600#define _generic_LC_uvchr(latin1, above_latin1, c) \
1601 (c < 256 ? latin1(c) : above_latin1(c))
1602#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1603 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1604
1605#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1606#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1607 _CC_ALPHANUMERIC, c)
b7d90381
KW
1608#define isASCII_LC_uvchr(c) isASCII_LC(c)
1609#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1610 is_HORIZWS_cp_high, c)
feeab5a9 1611#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1612#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1613#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1614#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1615 _is_uni_perl_idcont, c)
b7d90381 1616#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1617 _is_uni_perl_idstart, c)
1618#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1619#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1620#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1621#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1622#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1623 is_XPERLSPACE_cp_high, c)
cd500f2f 1624#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1625#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1626 _CC_WORDCHAR, c)
b7d90381
KW
1627#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1628 is_XDIGIT_cp_high, c)
e712593e 1629
b7d90381 1630#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1631
4650c663
KW
1632/* For internal core Perl use only: the base macros for defining macros like
1633 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1634 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1635 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1636 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1637 * same representation whether utf8 or not. Note that it assumes that the utf8
1638 * has been validated, and ignores 'use bytes' */
cd500f2f 1639#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1640 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1641 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1642 ? _generic_isCC( \
a62b247b 1643 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
aea0c08c 1644 *((p)+1 )), \
94bb8c36 1645 classnum) \
252810d0 1646 : utf8)
b7d90381
KW
1647/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1648 * 'above_latin1' can be a macro */
cd500f2f
KW
1649#define _generic_func_utf8(classnum, above_latin1, p) \
1650 _generic_utf8(classnum, p, above_latin1(p))
4650c663 1651/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1652 * 'above_latin1' parameter */
922e8cb4 1653#define _generic_swash_utf8(classnum, p) \
cd500f2f 1654 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
922e8cb4 1655
cc8ab7c0 1656/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1657 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1658 * class is TRUE for. Hence it can skip the tests for this range.
1659 * 'above_latin1' should include its arguments */
3ded5eb0 1660#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1661 (UTF8_IS_INVARIANT(*(p)) \
1662 ? _generic_isCC(*(p), classnum) \
1663 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1664 ? above_latin1 \
cc8ab7c0
KW
1665 : 0)
1666
84238efa
KW
1667/* NOTE that some of these macros have very similar ones in regcharclass.h.
1668 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1669 * there, differing in name only by an underscore from the one here
1670 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1671 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1672 * points; the regcharclass.h ones are implemented as a series of
1673 * "if-else-if-else ..." */
1674
b7d90381
KW
1675#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1676#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1677#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
509fb054
KW
1678 utf8, the non-utf8 macro works
1679 */
b7d90381 1680#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2
KW
1681
1682#ifdef EBCDIC
1683 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1684 * more efficient macro instead of the more general one */
88794300 1685# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
e8fa43e2
KW
1686#else
1687# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1688#endif
1689
cd500f2f
KW
1690#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1691 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1692#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1693#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1694 _is_utf8_perl_idcont, p)
e5dcd934 1695
c11ff943
KW
1696/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1697 * IDFIRST is an alnum. See
c8362b00 1698 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1699 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1700 * This used to be not the XID version, but we decided to go with the more
1701 * modern Unicode definition */
b7d90381 1702#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1703 _is_utf8_perl_idstart, p)
e5dcd934 1704
b7d90381
KW
1705#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1706#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1707#define isPSXSPC_utf8(p) isSPACE_utf8(p)
1708#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1709#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1710#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1711#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1712#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1713#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
cd500f2f 1714 is_XDIGIT_high(p))
a0ed51b3 1715
7d2b61dd 1716#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1717#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1718#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1719#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1720
4650c663
KW
1721/* For internal core Perl use only: the base macros for defining macros like
1722 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1723 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1724 * 'macro' parameter */
94bb8c36
KW
1725#define _generic_LC_utf8(macro, p, utf8) \
1726 (UTF8_IS_INVARIANT(*(p)) \
1727 ? macro(*(p)) \
1728 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
a62b247b 1729 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
9201a572
KW
1730 : utf8)
1731
cd500f2f
KW
1732#define _generic_LC_swash_utf8(macro, classnum, p) \
1733 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1734#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1735 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1736
b7d90381 1737#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
cd500f2f 1738 _CC_ALPHANUMERIC, p)
b7d90381
KW
1739#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1740#define isASCII_LC_utf8(p) isASCII_LC(*p)
1741#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1742 is_HORIZWS_high, p)
1743#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1744#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1745#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1746#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1747 _is_utf8_perl_idcont, p)
1748#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1749 _is_utf8_perl_idstart, p)
1750#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1751#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1752#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1753#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1754#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1755 is_XPERLSPACE_high, p)
1756#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
cd500f2f
KW
1757#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1758 _CC_WORDCHAR, p)
b7d90381
KW
1759#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1760 is_XDIGIT_high, p)
aaa51d5e 1761
fbc19f27
KW
1762/* Macros for backwards compatibility and for completeness when the ASCII and
1763 * Latin1 values are identical */
b7d90381
KW
1764#define isALPHAU(c) isALPHA_L1(c)
1765#define isDIGIT_L1(c) isDIGIT_A(c)
1766#define isOCTAL(c) isOCTAL_A(c)
1767#define isOCTAL_L1(c) isOCTAL_A(c)
1768#define isXDIGIT_L1(c) isXDIGIT_A(c)
1769#define isALNUM(c) isWORDCHAR(c)
1770#define isALNUMU(c) isWORDCHAR_L1(c)
1771#define isALNUM_LC(c) isWORDCHAR_LC(c)
1772#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 1773#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
1774#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1775#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1776#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1777#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1778#define isALNUMC(c) isALPHANUMERIC(c)
1779#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1780#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 1781#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 1782#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 1783#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1784
2bd1cbf6
KW
1785/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1786 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1787 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1788 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1789 * the outlier from the block that contains the other controls, just like
1790 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1791 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
1792 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1793 * etc. */
2bd1cbf6 1794#ifndef EBCDIC
75763b3a 1795# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 1796#else
75763b3a
KW
1797# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1798 ((isPRINT_A(c)) \
1799 ? (UNLIKELY((c) == '?') \
1800 ? QUESTION_MARK_CTRL \
1801 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1802 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1803 ? '?' \
1804 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 1805#endif
bbce6d69 1806
dea28490
JJ
1807/* Line numbers are unsigned, 32 bits. */
1808typedef U32 line_t;
e5dcd934 1809#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 1810
91152fc1
DG
1811/* Helpful alias for version prescan */
1812#define is_LAX_VERSION(a,b) \
1813 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1814
1815#define is_STRICT_VERSION(a,b) \
1816 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1817
1818#define BADVERSION(a,b,c) \
1819 if (b) { \
1820 *b = c; \
1821 } \
1822 return a;
8c52afec 1823
cb27eebd
KW
1824/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1825 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1826 * which is advanced. The input is validated only by an assert() in DEBUGGING
1827 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1828 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1829#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1830 ? (c) \
1831 : ((c) + 9))))
902008b8
KW
1832#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1833 ? (*(s)++) \
1834 : (*(s)++ + 9))))
95a59cab 1835
cb27eebd
KW
1836/* Converts a character known to represent an octal digit (0-7) to its numeric
1837 * value. The input is validated only by an assert() in DEBUGGING builds. In
1838 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1839#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1840
305b8651
KW
1841/* Efficiently returns a boolean as to if two native characters are equivalent
1842 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1843 * the ALPHA in the name is to remind you of that. This is asserted() in
1844 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1845 * works (on valid input) for both non- and UTF-8-encoded bytes.
1846 *
1847 * When one of the inputs is a compile-time constant and gets folded by the
1848 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1849 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1850 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
1851 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1852 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1853 * just a single 0, in the bit position where the upper- and lowercase differ.
1854 * */
305b8651
KW
1855#define isALPHA_FOLD_EQ(c1, c2) \
1856 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1857 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1858#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1859
8e84507e 1860/*
ccfc67b7
JH
1861=head1 Memory Management
1862
a02a5408 1863=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
1864The XSUB-writer's interface to the C C<malloc> function.
1865
596f7718 1866Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1867
c5008215
JC
1868In 5.9.3, Newx() and friends replace the older New() API, and drops
1869the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 1870themselves. This aid has been superseded by a new build option,
d10b4965 1871PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
1872there for use in XS modules supporting older perls.
1873
a02a5408 1874=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 1875The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 1876cast. See also C<L</Newx>>.
954c1994 1877
596f7718 1878Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1879
a02a5408 1880=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 1881The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 1882memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 1883
596f7718 1884Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1885
954c1994
GS
1886=for apidoc Am|void|Renew|void* ptr|int nitems|type
1887The XSUB-writer's interface to the C C<realloc> function.
1888
596f7718 1889Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1890
954c1994
GS
1891=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1892The XSUB-writer's interface to the C C<realloc> function, with
1893cast.
1894
596f7718 1895Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1896
49b8b560 1897=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
1898The XSUB-writer's interface to the C C<free> function.
1899
596f7718 1900This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 1901
954c1994
GS
1902=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1903The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 1904source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1905C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 1906
e90e2364 1907=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 1908Like C<Move> but returns C<dest>. Useful
72d33970 1909for encouraging compilers to tail-call
e90e2364
NC
1910optimise.
1911
954c1994
GS
1912=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1913The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 1914source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1915C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 1916
e90e2364
NC
1917=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1918
796b6530 1919Like C<Copy> but returns C<dest>. Useful
72d33970 1920for encouraging compilers to tail-call
e90e2364
NC
1921optimise.
1922
954c1994
GS
1923=for apidoc Am|void|Zero|void* dest|int nitems|type
1924
1925The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1926destination, C<nitems> is the number of items, and C<type> is the type.
1927
e90e2364
NC
1928=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1929
72d33970
FC
1930Like C<Zero> but returns dest. Useful
1931for encouraging compilers to tail-call
e90e2364
NC
1932optimise.
1933
da5d8dbb 1934=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 1935This is an architecture-independent macro to copy one structure to another.
954c1994 1936
7e337ee0
JH
1937=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1938
1939Fill up memory with a byte pattern (a byte repeated over and over
1940again) that hopefully catches attempts to access uninitialized memory.
1941
1942=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1943
1944PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1945
1c12ffb4 1946=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
1947
1948PoisonWith(0xEF) for catching access to freed memory.
1949
9965345d
JH
1950=for apidoc Am|void|Poison|void* dest|int nitems|type
1951
7e337ee0 1952PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
1953
1954=cut */
954c1994 1955
561b68a9
SH
1956/* Maintained for backwards-compatibility only. Use newSV() instead. */
1957#ifndef PERL_CORE
ff06c60c 1958#define NEWSV(x,len) newSV(len)
561b68a9 1959#endif
ff06c60c 1960
19a94d75
JH
1961#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1962
e6bdf523 1963
27d5b266 1964#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
1965
1966/* This expression will be constant-folded at compile time. It checks
1967 * whether or not the type of the count n is so small (e.g. U8 or U16, or
1968 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
1969 * As well as avoiding the need for a run-time check in some cases, it's
1970 * designed to avoid compiler warnings like:
1971 * comparison is always false due to limited range of data type
73e8ff00
DM
1972 * It's mathematically equivalent to
1973 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
1974 */
1975
1976# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
004073ba 1977 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
e6bdf523 1978
88f9f128 1979/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
1980 * compiler warnings. We *want* to write the expression as
1981 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
1982 * (for some compile-time constant C), but even when the LHS
1983 * constant-folds to false at compile-time, g++ insists on emitting
1984 * warnings about the RHS (e.g. "comparison is always false"), so instead
1985 * we write it as
e6bdf523 1986 *
d98e5cde 1987 * (cond ? n : X) > C
88f9f128 1988 *
d98e5cde
DM
1989 * where X is a constant with X > C always false. Choosing a value for X
1990 * is tricky. If 0, some compilers will complain about 0 > C always being
1991 * false; if 1, Coverity complains when n happens to be the constant value
1992 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
1993 * for X and hope that nothing else whines.
e6bdf523
DM
1994 */
1995
1996# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
1997 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
1998 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
1999
2000# define MEM_WRAP_CHECK(n,t) \
2001 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2002 && (croak_memory_wrap(),0))
2003
2004# define MEM_WRAP_CHECK_1(n,t,a) \
2005 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2006 && (Perl_croak_nocontext("%s",(a)),0))
2007
8b44ba4c 2008#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2009
d1decf2b 2010#define PERL_STRLEN_ROUNDUP(n) ((void)(((n) > MEM_SIZE_MAX - 2 * PERL_STRLEN_ROUNDUP_QUANTUM) ? (croak_memory_wrap(),0):0),((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266
JH
2011#else
2012
410319be
NC
2013#define MEM_WRAP_CHECK(n,t)
2014#define MEM_WRAP_CHECK_1(n,t,a)
2015#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
2016#define MEM_WRAP_CHECK_(n,t)
2017
6fff79ce 2018#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 2019
1936d2a7 2020#endif
8b44ba4c 2021
fe4f188c 2022#ifdef PERL_MEM_LOG
46c6c7e2 2023/*
9f653bb5 2024 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2025 * go through functions, which are handy for debugging breakpoints, but
2026 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2027 * line number, and C function name if available) passed in. This info can
2028 * then be used for logging the calls, for which one gets a sample
73d1d973 2029 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2030 *
46c6c7e2 2031 * Known problems:
94e892a6 2032 * - not all memory allocs get logged, only those
46c6c7e2 2033 * that go through Newx() and derivatives (while all
94e892a6 2034 * Safefrees do get logged)
46c6c7e2
JH
2035 * - __FILE__ and __LINE__ do not work everywhere
2036 * - __func__ or __FUNCTION__ even less so
2037 * - I think more goes on after the perlio frees but
2038 * the thing is that STDERR gets closed (as do all
2039 * the file descriptors)
2040 * - no deeper calling stack than the caller of the Newx()
2041 * or the kind, but do I look like a C reflection/introspection
2042 * utility to you?
2043 * - the function prototypes for the logging functions
2044 * probably should maybe be somewhere else than handy.h
2045 * - one could consider inlining (macrofying) the logging
2046 * for speed, but I am too lazy
2047 * - one could imagine recording the allocations in a hash,
2048 * (keyed by the allocation address?), and maintain that
2049 * through reallocs and frees, but how to do that without
2050 * any News() happening...?
73d1d973 2051 * - lots of -Ddefines to get useful/controllable output
b953482e 2052 * - lots of ENV reads
46c6c7e2
JH
2053 */
2054
0b0ab801 2055# ifdef PERL_CORE
73d1d973 2056# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2057enum mem_log_type {
2058 MLT_ALLOC,
2059 MLT_REALLOC,
d7a2c63c
MHM
2060 MLT_FREE,
2061 MLT_NEW_SV,
2062 MLT_DEL_SV
0b0ab801
MHM
2063};
2064# endif
12754f92 2065# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2066void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2067void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2068# endif
0b0ab801
MHM
2069# endif
2070
fe4f188c
JH
2071#endif
2072
2073#ifdef PERL_MEM_LOG
d1401ee9
MHM
2074#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2075#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2076#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2077#endif
2078
2079#ifndef MEM_LOG_ALLOC
2080#define MEM_LOG_ALLOC(n,t,a) (a)
2081#endif
2082#ifndef MEM_LOG_REALLOC
2083#define MEM_LOG_REALLOC(n,t,v,a) (a)
2084#endif
2085#ifndef MEM_LOG_FREE
2086#define MEM_LOG_FREE(a) (a)
2087#endif
2088
d1401ee9
MHM
2089#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2090#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2091#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2092
2093#ifndef PERL_CORE
a02a5408
JC
2094/* pre 5.9.x compatibility */
2095#define New(x,v,n,t) Newx(v,n,t)
2096#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2097#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2098#endif
a02a5408 2099
ff68c719 2100#define Renew(v,n,t) \
d1401ee9 2101 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2102#define Renewc(v,n,t,c) \
d1401ee9 2103 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2104
2105#ifdef PERL_POISON
2106#define Safefree(d) \
06c0cc96 2107 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2108#else
fe4f188c 2109#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2110#endif
55497cff 2111
8b44ba4c
NC
2112#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2113#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2114#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2115
8b44ba4c
NC
2116#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2117#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2118#ifdef HAS_MEMSET
8b44ba4c 2119#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2120#else
8b44ba4c
NC
2121/* Using bzero(), which returns void. */
2122#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2123#endif
2124
7e337ee0
JH
2125#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2126#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2127#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2128#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2129
caa674f3
DD
2130#ifdef PERL_POISON
2131# define PERL_POISON_EXPR(x) x
2132#else
2133# define PERL_POISON_EXPR(x)
2134#endif
2135
2304df62 2136#ifdef USE_STRUCT_COPY
ff68c719 2137#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2138#else
2139#define StructCopy(s,d,t) Copy(s,d,1,t)
2140#endif
2cc61e15 2141
c3caa5c3
JH
2142/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2143 * want your zero-based indices to be less than but not equal to).
2144 *
2145 * C_ARRAY_END is one past the last: half-open/half-closed range,
2146 * not last-inclusive range. */
622913ab 2147#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2148#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2149
2cc61e15
DD
2150#ifdef NEED_VA_COPY
2151# ifdef va_copy
2152# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2153# else
a1866d1b
JH
2154# if defined(__va_copy)
2155# define Perl_va_copy(s, d) __va_copy(d, s)
2156# else
2157# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2158# endif
2cc61e15
DD
2159# endif
2160#endif
2161
472d47bc
SB
2162/* convenience debug macros */
2163#ifdef USE_ITHREADS
2164#define pTHX_FORMAT "Perl interpreter: 0x%p"
2165#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2166#define pTHX_VALUE_ (void *)my_perl,
2167#define pTHX_VALUE (void *)my_perl
2168#define pTHX__VALUE_ ,(void *)my_perl,
2169#define pTHX__VALUE ,(void *)my_perl
472d47bc 2170#else
3609ea0d 2171#define pTHX_FORMAT
472d47bc 2172#define pTHX__FORMAT
3609ea0d 2173#define pTHX_VALUE_
472d47bc 2174#define pTHX_VALUE
3609ea0d 2175#define pTHX__VALUE_
472d47bc
SB
2176#define pTHX__VALUE
2177#endif /* USE_ITHREADS */
3609ea0d 2178
2acdbac1
NC
2179/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2180 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2181 CPAN::Unpack show any users outside the core. */
2182#ifdef PERL_CORE
b7d90381
KW
2183# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2184 "Use of " s " is deprecated")
2acdbac1
NC
2185#endif
2186
dfff4baf
BF
2187/* Internal macros to deal with gids and uids */
2188#ifdef PERL_CORE
2189
2190# if Uid_t_size > IVSIZE
2191# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2192# define SvUID(sv) SvNV(sv)
2193# else
2194# if Uid_t_sign <= 0
2195# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2196# define SvUID(sv) SvIV(sv)
2197# else
2198# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2199# define SvUID(sv) SvUV(sv)
2200# endif
2201# endif /* Uid_t_size */
2202
2203# if Gid_t_size > IVSIZE
2204# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2205# define SvGID(sv) SvNV(sv)
2206# else
2207# if Gid_t_sign <= 0
2208# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2209# define SvGID(sv) SvIV(sv)
2210# else
2211# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2212# define SvGID(sv) SvUV(sv)
2213# endif
2214# endif /* Gid_t_size */
2215
2216#endif
2217
9d745869
KW
2218#endif /* HANDY_H */
2219
e9a8c099 2220/*
14d04a33 2221 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2222 */