This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
perlop: fix broken example by deleting it [perl #119667]
[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
30a15352
KW
299Like C<newSVpvn>, but takes a literal C<NUL>-terminated string instead of a
300string/length pair.
3fe05580 301
84bafc02 302=for apidoc Ama|SV*|newSVpvs_flags|const char* s|U32 flags
30a15352
KW
303Like C<newSVpvn_flags>, but takes a literal C<NUL>-terminated string instead of
304a string/length pair.
84bafc02 305
3fe05580 306=for apidoc Ama|SV*|newSVpvs_share|const char* s
30a15352
KW
307Like C<newSVpvn_share>, but takes a literal C<NUL>-terminated string instead of
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
30a15352
KW
311Like C<sv_catpvn_flags>, but takes a literal C<NUL>-terminated string instead
312of a string/length pair.
9dcc53ea
Z
313
314=for apidoc Am|void|sv_catpvs_nomg|SV* sv|const char* s
315Like C<sv_catpvn_nomg>, but takes a literal string instead of a
316string/length pair.
317
7b4e7469 318=for apidoc Am|void|sv_catpvs|SV* sv|const char* s
3fe05580
MHM
319Like C<sv_catpvn>, but takes a literal string instead of a string/length pair.
320
9dcc53ea
Z
321=for apidoc Am|void|sv_catpvs_mg|SV* sv|const char* s
322Like C<sv_catpvn_mg>, but takes a literal string instead of a
323string/length pair.
324
7b4e7469 325=for apidoc Am|void|sv_setpvs|SV* sv|const char* s
3fe05580
MHM
326Like C<sv_setpvn>, but takes a literal string instead of a string/length pair.
327
9dcc53ea
Z
328=for apidoc Am|void|sv_setpvs_mg|SV* sv|const char* s
329Like C<sv_setpvn_mg>, but takes a literal string instead of a
330string/length pair.
331
332=for apidoc Am|SV *|sv_setref_pvs|const char* s
333Like C<sv_setref_pvn>, but takes a literal string instead of a
334string/length pair.
335
3fe05580
MHM
336=head1 Memory Management
337
338=for apidoc Ama|char*|savepvs|const char* s
30a15352
KW
339Like C<savepvn>, but takes a literal C<NUL>-terminated string instead of a
340string/length pair.
3fe05580 341
9dcc53ea
Z
342=for apidoc Ama|char*|savesharedpvs|const char* s
343A version of C<savepvs()> which allocates the duplicate string in memory
344which is shared between threads.
345
3fe05580
MHM
346=head1 GV Functions
347
348=for apidoc Am|HV*|gv_stashpvs|const char* name|I32 create
349Like C<gv_stashpvn>, but takes a literal string instead of a string/length pair.
350
351=head1 Hash Manipulation Functions
352
353=for apidoc Am|SV**|hv_fetchs|HV* tb|const char* key|I32 lval
354Like C<hv_fetch>, but takes a literal string instead of a string/length pair.
355
356=for apidoc Am|SV**|hv_stores|HV* tb|const char* key|NULLOK SV* val
357Like C<hv_store>, but takes a literal string instead of a string/length pair
358and omits the hash parameter.
359
510966aa
Z
360=head1 Lexer interface
361
362=for apidoc Amx|void|lex_stuff_pvs|const char *pv|U32 flags
363
364Like L</lex_stuff_pvn>, but takes a literal string instead of a
365string/length pair.
366
3fe05580
MHM
367=cut
368*/
369
926bb54c
KW
370/* concatenating with "" ensures that only literal strings are accepted as
371 * argument */
26bfd924 372#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
2efa8cc7 373
926bb54c
KW
374/* note that STR_WITH_LEN() can't be used as argument to macros or functions
375 * that under some configurations might be macros, which means that it requires
376 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
ba3a79e7
GA
377 */
378
379/* STR_WITH_LEN() shortcuts */
380#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
381#define newSVpvs_flags(str,flags) \
382 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 383#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
384#define sv_catpvs_flags(sv, str, flags) \
385 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
386#define sv_catpvs_nomg(sv, str) \
387 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
388#define sv_catpvs(sv, str) \
389 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
390#define sv_catpvs_mg(sv, str) \
391 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 392#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
393#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
394#define sv_setref_pvs(rv, classname, str) \
395 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 396#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
397#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
398#define gv_stashpvs(str, create) \
399 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
400#define gv_fetchpvs(namebeg, add, sv_type) \
401 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
402#define gv_fetchpvn(namebeg, len, add, sv_type) \
403 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
404#define sv_catxmlpvs(dsv, str, utf8) \
405 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
a038e571
NC
406#define hv_fetchs(hv,key,lval) \
407 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
408 (lval) ? (HV_FETCH_JUST_SV | HV_FETCH_LVALUE) \
409 : HV_FETCH_JUST_SV, NULL, 0))
410
411#define hv_stores(hv,key,val) \
412 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
413 (HV_FETCH_ISSTORE|HV_FETCH_JUST_SV), (val), 0))
ba3a79e7 414
510966aa
Z
415#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
416
b96d8cd9
NC
417#define get_cvs(str, flags) \
418 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 419
954c1994 420/*
ccfc67b7
JH
421=head1 Miscellaneous Functions
422
954c1994 423=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
424Test two C<NUL>-terminated strings to see if they are different. Returns true
425or false.
954c1994
GS
426
427=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
428Test two C<NUL>-terminated strings to see if they are equal. Returns true or
429false.
954c1994
GS
430
431=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
432Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
433second, C<s2>. Returns true or false.
954c1994
GS
434
435=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
436Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
437equal to the second, C<s2>. Returns true or false.
954c1994
GS
438
439=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
440Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
441the second, C<s2>. Returns true or false.
954c1994
GS
442
443=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
444Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
445or equal to the second, C<s2>. Returns true or false.
954c1994
GS
446
447=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
448Test two C<NUL>-terminated strings to see if they are different. The C<len>
449parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
450wrapper for C<strncmp>).
451
452=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
453Test two C<NUL>-terminated strings to see if they are equal. The C<len>
454parameter indicates the number of bytes to compare. Returns true or false. (A
455wrapper for C<strncmp>).
954c1994 456
bd18bd40
KW
457=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
458Test two buffers (which may contain embedded C<NUL> characters, to see if they
459are equal. The C<len> parameter indicates the number of bytes to compare.
460Returns zero if equal, or non-zero if non-equal.
461
462=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
463Test two buffers (which may contain embedded C<NUL> characters, to see if they
464are not equal. The C<len> parameter indicates the number of bytes to compare.
465Returns zero if non-equal, or non-zero if equal.
466
954c1994
GS
467=cut
468*/
469
8d063cd8
LW
470#define strNE(s1,s2) (strcmp(s1,s2))
471#define strEQ(s1,s2) (!strcmp(s1,s2))
472#define strLT(s1,s2) (strcmp(s1,s2) < 0)
473#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
474#define strGT(s1,s2) (strcmp(s1,s2) > 0)
475#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
476#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
477#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
378cc40b 478
36477c24
PP
479#ifdef HAS_MEMCMP
480# define memNE(s1,s2,l) (memcmp(s1,s2,l))
481# define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
482#else
483# define memNE(s1,s2,l) (bcmp(s1,s2,l))
484# define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
485#endif
486
568a785a 487#define memEQs(s1, l, s2) \
8e6497b6 488 (sizeof(s2)-1 == l && memEQ(s1, ("" s2 ""), (sizeof(s2)-1)))
568a785a
NC
489#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
490
bbce6d69
PP
491/*
492 * Character classes.
493 *
494 * Unfortunately, the introduction of locales means that we
495 * can't trust isupper(), etc. to tell the truth. And when
496 * it comes to /\w+/ with tainting enabled, we *must* be able
497 * to trust our character classes.
498 *
499 * Therefore, the default tests in the text of Perl will be
500 * independent of locale. Any code that wants to depend on
501 * the current locale will use the tests that begin with "lc".
502 */
503
2304df62
AD
504#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
505# ifndef CTYPE256
506# define CTYPE256
507# endif
508#endif
509
954c1994 510/*
ccfc67b7 511
dcccc8ff 512=head1 Character classification
243effed
KW
513This section is about functions (really macros) that classify characters
514into types, such as punctuation versus alphabetic, etc. Most of these are
515analogous to regular expression character classes. (See
516L<perlrecharclass/POSIX Character Classes>.) There are several variants for
517each class. (Not all macros have all variants; each item below lists the
518ones valid for it.) None are affected by C<use bytes>, and only the ones
519with C<LC> in the name are affected by the current locale.
520
521The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
522C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
523represented by that octet is (or on non-ASCII platforms, corresponds to) an
524ASCII character in the named class based on platform, Unicode, and Perl rules.
525If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 526
6aff1f14
KW
527Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
528with no suffix C<"_A">.
4b9734bf
KW
529
530Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
531onto the platform. That is, the code points that are ASCII are unaffected,
532since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
533as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
534true when called with the code point 0xDF, which is a word character in both
4650c663 535ASCII and EBCDIC (though it represents different characters in each).
243effed
KW
536
537Variant C<isFOO_uni> is like the C<isFOO_L1> variant, but accepts any UV code
538point as input. If the code point is larger than 255, Unicode rules are used
539to determine if it is in the character class. For example,
6aff1f14
KW
540C<isWORDCHAR_uni(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
541WITH MACRON in Unicode, and is a word character.
243effed
KW
542
543Variant C<isFOO_utf8> is like C<isFOO_uni>, but the input is a pointer to a
dc6b0978
KW
544(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
545possibly containing embedded C<NUL> characters). The classification of just
546the first (possibly multi-byte) character in the string is tested.
243effed 547
31f05a37
KW
548Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
549result is based on the current locale, which is what C<LC> in the name stands
550for. If Perl can determine that the current locale is a UTF-8 locale, it uses
551the published Unicode rules; otherwise, it uses the C library function that
552gives the named classification. For example, C<isDIGIT_LC()> when not in a
553UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
554returned if the input won't fit into an octet. On some platforms where the C
555library function is known to be defective, Perl changes its result to follow
556the POSIX standard's rules.
243effed
KW
557
558Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
559returns the same as C<isFOO_LC> for input code points less than 256, and
560returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
561
b7d90381 562Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer
dc6b0978
KW
563to a (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
564possibly containing embedded C<NUL> characters). The classification of just
565the first (possibly multi-byte) character in the string is tested.
ccfc67b7 566
954c1994 567=for apidoc Am|bool|isALPHA|char ch
2787a470 568Returns a boolean indicating whether the specified character is an
6aff1f14 569alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
570See the L<top of this section|/Character classification> for an explanation of
571variants
e5ad6aba 572C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uni>, C<isALPHA_utf8>, C<isALPHA_LC>,
243effed 573C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
8a58bdcf 574
15861f94
KW
575=for apidoc Am|bool|isALPHANUMERIC|char ch
576Returns a boolean indicating whether the specified character is a either an
6aff1f14 577alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
578See the L<top of this section|/Character classification> for an explanation of
579variants
15861f94 580C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uni>,
e5ad6aba 581C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
15861f94
KW
582C<isALPHANUMERIC_LC_utf8>.
583
8a58bdcf
KW
584=for apidoc Am|bool|isASCII|char ch
585Returns a boolean indicating whether the specified character is one of the 128
243effed 586characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 587On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
588character corresponds to an ASCII character. Variants C<isASCII_A()> and
589C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
590See the L<top of this section|/Character classification> for an explanation of
591variants
243effed
KW
592C<isASCII_uni>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
593C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
594library routine C<isascii()>. In these cases, the variants whose names contain
595C<LC> are the same as the corresponding ones without.
596
d98532ea
KW
597Also note, that because all ASCII characters are UTF-8 invariant (meaning they
598have the exact same representation (always a single byte) whether encoded in
599UTF-8 or not), C<isASCII> will give the correct results when called with any
600byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
601work properly on any string encoded or not in UTF-8.
602
243effed
KW
603=for apidoc Am|bool|isBLANK|char ch
604Returns a boolean indicating whether the specified character is a
6aff1f14 605character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
606See the L<top of this section|/Character classification> for an explanation of
607variants
e5ad6aba 608C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uni>, C<isBLANK_utf8>, C<isBLANK_LC>,
243effed
KW
609C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
610platforms do not have the C library routine C<isblank()>. In these cases, the
611variants whose names contain C<LC> are the same as the corresponding ones
612without.
613
614=for apidoc Am|bool|isCNTRL|char ch
615Returns a boolean indicating whether the specified character is a
6aff1f14 616control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
617See the L<top of this section|/Character classification> for an explanation of
618variants
e5ad6aba 619C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uni>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
6aff1f14
KW
620C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
621On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 622
954c1994 623=for apidoc Am|bool|isDIGIT|char ch
2787a470 624Returns a boolean indicating whether the specified character is a
6aff1f14 625digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 626Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
627See the L<top of this section|/Character classification> for an explanation of
628variants
e5ad6aba 629C<isDIGIT_uni>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
243effed
KW
630C<isDIGIT_LC_utf8>.
631
632=for apidoc Am|bool|isGRAPH|char ch
633Returns a boolean indicating whether the specified character is a
6aff1f14 634graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff
KW
635See the L<top of this section|/Character classification> for an explanation of
636variants
e5ad6aba 637C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uni>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
243effed 638C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
954c1994 639
0c82b6df 640=for apidoc Am|bool|isLOWER|char ch
2787a470 641Returns a boolean indicating whether the specified character is a
6aff1f14 642lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
643See the L<top of this section|/Character classification> for an explanation of
644variants
e5ad6aba 645C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uni>, C<isLOWER_utf8>, C<isLOWER_LC>,
243effed 646C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
0c82b6df 647
c99e91e9 648=for apidoc Am|bool|isOCTAL|char ch
2787a470 649Returns a boolean indicating whether the specified character is an
6aff1f14 650octal digit, [0-7].
243effed
KW
651The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
652C<isOCTAL>.
653
654=for apidoc Am|bool|isPUNCT|char ch
655Returns a boolean indicating whether the specified character is a
6aff1f14
KW
656punctuation character, analogous to C<m/[[:punct:]]/>.
657Note that the definition of what is punctuation isn't as
243effed
KW
658straightforward as one might desire. See L<perlrecharclass/POSIX Character
659Classes> for details.
dcccc8ff
KW
660See the L<top of this section|/Character classification> for an explanation of
661variants
e5ad6aba 662C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uni>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
243effed 663C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
c99e91e9 664
0c82b6df 665=for apidoc Am|bool|isSPACE|char ch
2787a470 666Returns a boolean indicating whether the specified character is a
6aff1f14 667whitespace character. This is analogous
398d098a 668to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 669this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
670locale forms of this macro (the ones with C<LC> in their names) matched
671precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
672in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
673(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
674See the L<top of this section|/Character classification> for an explanation of
675variants
e5ad6aba 676C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uni>, C<isSPACE_utf8>, C<isSPACE_LC>,
243effed 677C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
0c82b6df 678
398d098a
KW
679=for apidoc Am|bool|isPSXSPC|char ch
680(short for Posix Space)
779cf272
KW
681Starting in 5.18, this is identical in all its forms to the
682corresponding C<isSPACE()> macros.
398d098a
KW
683The locale forms of this macro are identical to their corresponding
684C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
685non-locale forms differ from their C<isSPACE()> forms only in that the
686C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
687Otherwise they are identical. Thus this macro is analogous to what
688C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 689See the L<top of this section|/Character classification> for an explanation of
b7d90381
KW
690variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uni>, C<isPSXSPC_utf8>,
691C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
398d098a 692
954c1994 693=for apidoc Am|bool|isUPPER|char ch
2787a470 694Returns a boolean indicating whether the specified character is an
6aff1f14 695uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff
KW
696See the L<top of this section|/Character classification> for an explanation of
697variants
e5ad6aba 698C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uni>, C<isUPPER_utf8>, C<isUPPER_LC>,
243effed 699C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
954c1994 700
243effed 701=for apidoc Am|bool|isPRINT|char ch
8eea39dd 702Returns a boolean indicating whether the specified character is a
6aff1f14 703printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
704See the L<top of this section|/Character classification> for an explanation of
705variants
e5ad6aba 706C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uni>, C<isPRINT_utf8>, C<isPRINT_LC>,
243effed
KW
707C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
708
709=for apidoc Am|bool|isWORDCHAR|char ch
710Returns a boolean indicating whether the specified character is a character
711that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
712in a regular expression. A word character is an alphabetic character, a
713decimal digit, a connecting punctuation character (such as an underscore), or
714a "mark" character that attaches to one of those (like some sort of accent).
715C<isALNUM()> is a synonym provided for backward compatibility, even though a
716word character includes more than the standard C language meaning of
717alphanumeric.
dcccc8ff
KW
718See the L<top of this section|/Character classification> for an explanation of
719variants
1a83413c
KW
720C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uni>, and C<isWORDCHAR_utf8>.
721C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
722described there, but additionally include the platform's native underscore.
8a58bdcf
KW
723
724=for apidoc Am|bool|isXDIGIT|char ch
725Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
726digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
727and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
728See the L<top of this section|/Character classification> for an explanation of
729variants
243effed
KW
730C<isXDIGIT_uni>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
731C<isXDIGIT_LC_utf8>.
732
3c3ecf18
KW
733=for apidoc Am|bool|isIDFIRST|char ch
734Returns a boolean indicating whether the specified character can be the first
735character of an identifier. This is very close to, but not quite the same as
736the official Unicode property C<XID_Start>. The difference is that this
737returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
738See the L<top of this section|/Character classification> for an explanation of
739variants
3c3ecf18
KW
740C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uni>, C<isIDFIRST_utf8>,
741C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
742
743=for apidoc Am|bool|isIDCONT|char ch
744Returns a boolean indicating whether the specified character can be the
745second or succeeding character of an identifier. This is very close to, but
746not quite the same as the official Unicode property C<XID_Continue>. The
747difference is that this returns true only if the input character also matches
dcccc8ff
KW
748L</isWORDCHAR>. See the L<top of this section|/Character classification> for
749an
3c3ecf18
KW
750explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uni>,
751C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
752C<isIDCONT_LC_utf8>.
753
243effed 754=head1 Miscellaneous Functions
8eea39dd 755
95a59cab 756=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 757Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
758Behaviour is only well defined when isXDIGIT(*str) is true.
759
e7c1e6c1
KW
760=head1 Character case changing
761
1f607577
KW
762=for apidoc Am|U8|toUPPER|U8 ch
763Converts the specified character to uppercase. If the input is anything but an
764ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 765C<toUPPER_A> is equivalent.
954c1994 766
1f607577
KW
767=for apidoc Am|UV|toUPPER_uni|UV cp|U8* s|STRLEN* lenp
768Converts the Unicode code point C<cp> to its uppercase version, and
769stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
770that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
771bytes since the uppercase version may be longer than the original character.
772
773The first code point of the uppercased version is returned
774(but note, as explained just above, that there may be more.)
775
776=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
777Converts the UTF-8 encoded character at C<p> to its uppercase version, and
778stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
779that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
780bytes since the uppercase version may be longer than the original character.
781
782The first code point of the uppercased version is returned
783(but note, as explained just above, that there may be more.)
784
785The input character at C<p> is assumed to be well-formed.
786
25200305
KW
787=for apidoc Am|U8|toFOLD|U8 ch
788Converts the specified character to foldcase. If the input is anything but an
789ASCII uppercase character, that input character itself is returned. Variant
790C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
791Latin1 range, as the full generality of L</toFOLD_uni> is needed there.)
792
1f607577
KW
793=for apidoc Am|UV|toFOLD_uni|UV cp|U8* s|STRLEN* lenp
794Converts the Unicode code point C<cp> to its foldcase version, and
795stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
796that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
797bytes since the foldcase version may be longer than the original character.
798
799The first code point of the foldcased version is returned
800(but note, as explained just above, that there may be more.)
801
802=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
803Converts the UTF-8 encoded character at C<p> to its foldcase version, and
804stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
805that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
806bytes since the foldcase version may be longer than the original character.
807
808The first code point of the foldcased version is returned
809(but note, as explained just above, that there may be more.)
810
811The input character at C<p> is assumed to be well-formed.
812
813=for apidoc Am|U8|toLOWER|U8 ch
814Converts the specified character to lowercase. If the input is anything but an
815ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 816C<toLOWER_A> is equivalent.
954c1994 817
1f607577 818=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
819Converts the specified Latin1 character to lowercase. The results are
820undefined if the input doesn't fit in a byte.
1f607577
KW
821
822=for apidoc Am|U8|toLOWER_LC|U8 ch
823Converts the specified character to lowercase using the current locale's rules,
824if possible; otherwise returns the input character itself.
825
826=for apidoc Am|UV|toLOWER_uni|UV cp|U8* s|STRLEN* lenp
827Converts the Unicode code point C<cp> to its lowercase version, and
828stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
829that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
830bytes since the lowercase version may be longer than the original character.
831
832The first code point of the lowercased version is returned
833(but note, as explained just above, that there may be more.)
834
835=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
836Converts the UTF-8 encoded character at C<p> to its lowercase version, and
837stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
838that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
839bytes since the lowercase version may be longer than the original character.
840
841The first code point of the lowercased version is returned
842(but note, as explained just above, that there may be more.)
843
844The input character at C<p> is assumed to be well-formed.
845
25200305
KW
846=for apidoc Am|U8|toTITLE|U8 ch
847Converts the specified character to titlecase. If the input is anything but an
848ASCII lowercase character, that input character itself is returned. Variant
b7d90381
KW
849C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
850range, as the full generality of L</toTITLE_uni> is needed there. Titlecase is
851not a concept used in locale handling, so there is no functionality for that.)
25200305 852
1f607577
KW
853=for apidoc Am|UV|toTITLE_uni|UV cp|U8* s|STRLEN* lenp
854Converts the Unicode code point C<cp> to its titlecase version, and
855stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
856that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
857bytes since the titlecase version may be longer than the original character.
858
859The first code point of the titlecased version is returned
860(but note, as explained just above, that there may be more.)
861
862=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
863Converts the UTF-8 encoded character at C<p> to its titlecase version, and
864stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
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
871The input character at C<p> is assumed to be well-formed.
872
954c1994 873=cut
353c9b6f 874
1f607577 875XXX Still undocumented isVERTWS_uni and _utf8; it's unclear what their names
25200305 876really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 877
8a58bdcf 878Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
879patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
880
954c1994
GS
881*/
882
7c062697
KW
883/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
884 * is known only in the perl core, and this macro can be called from outside
885 * that */
886#ifdef HAS_QUAD
887# define WIDEST_UTYPE U64TYPE
888#else
889# define WIDEST_UTYPE U32
890#endif
891
3912bc88
KW
892/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
893 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
894 * bits of information are lost even on a 64-bit machine, but to get the
895 * compiler to optimize it out if possible. This is because Configure makes
896 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
897 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
898 *
899 * For Coverity, be always true, because otherwise Coverity thinks
900 * it finds several expressions that are always true, independent
901 * of operands. Well, they are, but that is kind of the point.
220c71bf 902 */
7e75d1a1 903#ifndef __COVERITY__
220c71bf 904#define FITS_IN_8_BITS(c) ((sizeof(c) == 1) || !(((WIDEST_UTYPE)(c)) & ~0xFF))
7e75d1a1
JH
905#else
906#define FITS_IN_8_BITS(c) (1)
907#endif
cf301eb7 908
41f43cc2 909#ifdef EBCDIC
b6340bd0 910# ifndef _ALL_SOURCE
0852beac
KW
911 /* The native libc isascii() et.al. functions return the wrong results
912 * on at least z/OS unless this is defined. */
b6340bd0
KW
913# error _ALL_SOURCE should probably be defined
914# endif
41f43cc2 915#else
0852beac
KW
916 /* There is a simple definition of ASCII for ASCII platforms. But the
917 * EBCDIC one isn't so simple, so is defined using table look-up like the
918 * other macros below */
41f43cc2
KW
919# define isASCII(c) ((WIDEST_UTYPE)(c) < 128)
920#endif
921
c2da0b36
KW
922/* The lower 3 bits in both the ASCII and EBCDIC representations of '0' are 0,
923 * and the 8 possible permutations of those bits exactly comprise the 8 octal
924 * digits */
3d33b3e3 925#define isOCTAL_A(c) cBOOL(FITS_IN_8_BITS(c) && (0xF8 & (c)) == '0')
c2da0b36 926
9fb1bf9d 927#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 928
a500dc72
KW
929/* Character class numbers. For internal core Perl use only. The ones less
930 * than 32 are used in PL_charclass[] and the ones up through the one that
931 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
932 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
933 * their actual definitions are here. If that file has a name not used here,
934 * it won't compile.
1709d539
KW
935 *
936 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
937 * frequency of their use. This gives a slight edge to exiting a loop earlier
938 * (in reginclass() in regexec.c) */
1709d539
KW
939# define _CC_WORDCHAR 0 /* \w and [:word:] */
940# define _CC_DIGIT 1 /* \d and [:digit:] */
941# define _CC_ALPHA 2 /* [:alpha:] */
942# define _CC_LOWER 3 /* [:lower:] */
943# define _CC_UPPER 4 /* [:upper:] */
944# define _CC_PUNCT 5 /* [:punct:] */
945# define _CC_PRINT 6 /* [:print:] */
15861f94 946# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 947# define _CC_GRAPH 8 /* [:graph:] */
b0d691b2 948# define _CC_CASED 9 /* [:lower:] and [:upper:] under /i */
1709d539 949
b0d691b2 950#define _FIRST_NON_SWASH_CC 10
1709d539
KW
951/* The character classes above are implemented with swashes. The second group
952 * (just below) contains the ones implemented without. These are also sorted
953 * in rough order of the frequency of their use, except that \v should be last,
954 * as it isn't a real Posix character class, and some (small) inefficiencies in
955 * regular expression handling would be introduced by putting it in the middle
956 * of those that are. Also, cntrl and ascii come after the others as it may be
957 * useful to group these which have no members that match above Latin1, (or
958 * above ASCII in the latter case) */
959
779cf272 960# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
961# define _CC_BLANK 11 /* [:blank:] */
962# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
963# define _CC_CNTRL 13 /* [:cntrl:] */
964# define _CC_ASCII 14 /* [:ascii:] */
965# define _CC_VERTSPACE 15 /* \v */
1709d539 966
a0947d7b
KW
967# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
968
1709d539 969/* The members of the third group below do not need to be coordinated with data
3ffc8c70 970 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
971# define _CC_IDFIRST 16
972# define _CC_CHARNAME_CONT 17
973# define _CC_NONLATIN1_FOLD 18
974# define _CC_NONLATIN1_SIMPLE_FOLD 19
975# define _CC_QUOTEMETA 20
976# define _CC_NON_FINAL_FOLD 21
977# define _CC_IS_IN_SOME_FOLD 22
978# define _CC_MNEMONIC_CNTRL 23
073c22b3
KW
979
980/* This next group is only used on EBCDIC platforms, so theoretically could be
981 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
982# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
983# define _CC_UTF8_IS_START 29
984# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
985# define _CC_UTF8_IS_CONTINUATION 31
986/* Unused: 24-27
f4cdb42c
KW
987 * If more bits are needed, one could add a second word for non-64bit
988 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
989 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
990 * is used only for optimization (as of this writing), and differs in the
991 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 992 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
993 * would just cause /i regexes which match them to run less efficiently.
994 * Similarly the EBCDIC-only bits are used just for speed, and could be
995 * replaced by other means */
96ac0975 996
3a371f2f
KW
997#if defined(PERL_CORE) || defined(PERL_EXT)
998/* An enum version of the character class numbers, to help compilers
999 * optimize */
1000typedef enum {
3a371f2f 1001 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1002 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1003 _CC_ENUM_ASCII = _CC_ASCII,
1004 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1005 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1006 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1007 _CC_ENUM_DIGIT = _CC_DIGIT,
1008 _CC_ENUM_GRAPH = _CC_GRAPH,
1009 _CC_ENUM_LOWER = _CC_LOWER,
1010 _CC_ENUM_PRINT = _CC_PRINT,
1011 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1012 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1013 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1014 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1015 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1016 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1017} _char_class_number;
1018#endif
1019
63c61c3f 1020#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1021#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1022
b7d90381
KW
1023#if defined(PERL_IN_UTF8_C) \
1024 || defined(PERL_IN_REGCOMP_C) \
1025 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1026# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1027 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1028 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1029 #error Need to adjust order of swash_property_names[]
1030# endif
1031
b44daeeb
KW
1032/* This is declared static in each of the few files that this is #defined for
1033 * to keep them from being publicly accessible. Hence there is a small amount
1034 * of wasted space */
1035
922e8cb4
KW
1036static const char* const swash_property_names[] = {
1037 "XPosixWord",
1038 "XPosixDigit",
1039 "XPosixAlpha",
1040 "XPosixLower",
1041 "XPosixUpper",
1042 "XPosixPunct",
1043 "XPosixPrint",
1044 "XPosixAlnum",
b0d691b2
KW
1045 "XPosixGraph",
1046 "Cased"
922e8cb4
KW
1047};
1048#endif
1049
6635f04f 1050START_EXTERN_C
96ac0975
NC
1051# ifdef DOINIT
1052EXTCONST U32 PL_charclass[] = {
1053# include "l1_char_class_tab.h"
1054};
1055
1056# else /* ! DOINIT */
1057EXTCONST U32 PL_charclass[];
1058# endif
6635f04f 1059END_EXTERN_C
96ac0975 1060
265c1f46 1061 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1062# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1063
1064 /* For internal core Perl use only: the base macro for defining macros like
1065 * isALPHA */
ff7ecfc3 1066# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1067 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1068
f4cdb42c
KW
1069 /* The mask for the _A versions of the macros; it just adds in the bit for
1070 * ASCII. */
1071# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1072
4650c663
KW
1073 /* For internal core Perl use only: the base macro for defining macros like
1074 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1075 * the ASCII bit are present */
b7d90381
KW
1076# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1077 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1078 == _CC_mask_A(classnum)))
f4cdb42c 1079
f4cdb42c 1080# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1081# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1082# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1083# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1084# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1085# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1086# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1087# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1088# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1089# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1090# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1091# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1092# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1093 */
d95f8b6a 1094# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1095# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1096# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1097# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1098
1099 /* continuation character for legal NAME in \N{NAME} */
1100# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1101
1102# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1103# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1104# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1105# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1106# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1107# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1108# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1109# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1110# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1111# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1112
0852beac
KW
1113# ifdef EBCDIC
1114# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1115# endif
1116
f12c0118
KW
1117 /* Participates in a single-character fold with a character above 255 */
1118# 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)))
1119
1120 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1121# 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 1122
4eeeb416 1123# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1124# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1125 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1126# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1127 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1128# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1129 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1130#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1131
1132 /* If we don't have perl.h, we are compiling a utility program. Below we
1133 * hard-code various macro definitions that wouldn't otherwise be available
0852beac
KW
1134 * to it. Most are coded based on first principals. First some ones common
1135 * to both ASCII and EBCDIC */
1136# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1137# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1138# define isSPACE_A(c) (isBLANK_A(c) \
1139 || (c) == '\n' \
1140 || (c) == '\r' \
1141 || (c) == '\v' \
1142 || (c) == '\f')
1143# ifdef EBCDIC /* There are gaps between 'i' and 'j'; 'r' and 's'. Same
1144 for uppercase. This is ordered to exclude most things
1145 early */
1146# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1147 && ((c) <= 'i' \
1148 || ((c) >= 'j' && (c) <= 'r') \
1149 || (c) >= 's'))
1150# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1151 && ((c) <= 'I' \
1152 || ((c) >= 'J' && (c) <= 'R') \
1153 || (c) >= 'S'))
1154# else /* ASCII platform. */
9fb1bf9d 1155# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z')
2181da29 1156# define isUPPER_A(c) ((c) <= 'Z' && (c) >= 'A')
0852beac
KW
1157# endif
1158
1159 /* Some more ASCII, non-ASCII common definitions */
a4d7a999
KW
1160# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1161# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1162# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac
KW
1163# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1164# define isXDIGIT_A(c) (isDIGIT_A(c) \
1165 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1166 || ((c) <= 'F' && (c) >= 'A'))
3ded5eb0 1167
0852beac
KW
1168# ifdef EBCDIC
1169# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1170 || (c) == '#' || (c) == '$' || (c) == '%' \
1171 || (c) == '&' || (c) == '\'' || (c) == '(' \
1172 || (c) == ')' || (c) == '*' || (c) == '+' \
1173 || (c) == ',' || (c) == '.' || (c) == '/' \
1174 || (c) == ':' || (c) == ';' || (c) == '<' \
1175 || (c) == '=' || (c) == '>' || (c) == '?' \
1176 || (c) == '@' || (c) == '[' || (c) == '\\' \
1177 || (c) == ']' || (c) == '^' || (c) == '_' \
1178 || (c) == '`' || (c) == '{' || (c) == '|' \
1179 || (c) == '}' || (c) == '~')
1180# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1181# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1182
1183# ifdef QUESTION_MARK_CTRL
1184# define _isQMC(c) ((c) == QUESTION_MARK_CTRL)
1185# else
1186# define _isQMC(c) 0
1187# endif
1188
1189 /* I (khw) can't think of a way to define all the ASCII controls
1190 * without resorting to a libc (locale-sensitive) call. But we know
1191 * that all controls but the question-mark one are in the range 0-0x3f.
1192 * This makes sure that all the controls that have names are included,
1193 * and all controls that are also considered ASCII in the locale. This
1194 * may include more or fewer than what it actually should, but the
1195 * wrong ones are less-important controls, so likely won't impact
1196 * things (keep in mind that this is compiled only if perl.h isn't
1197 * available). The question mark control is included if available */
1198# define isCNTRL_A(c) (((c) < 0x40 && isascii(c)) \
1199 || (c) == '\0' || (c) == '\a' || (c) == '\b' \
1200 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1201 || (c) == '\t' || (c) == '\v' || _isQMC(c))
1202
1203# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1204# else /* ASCII platform; things are simpler, and isASCII has already
1205 been defined */
1206# define isGRAPH_A(c) (((c) > ' ' && (c) < 127))
1207# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1208# define isPUNCT_A(c) (isGRAPH_A(c) && (! isALPHANUMERIC_A(c)))
1209# define isCNTRL_A(c) (isASCII(c) && (! isPRINT_A(c)))
1210# endif
1211
3ffc8c70 1212 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1213 * during debugging, and it seems best to keep them. We may be called
1214 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1215 * do anything anyway, so make it not a problem */
1216# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1217# define NATIVE_TO_LATIN1(ch) (ch)
1218# endif
3ded5eb0
KW
1219# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1220# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1221# define isBLANK_L1(c) (isBLANK_A(c) \
1222 || (FITS_IN_8_BITS(c) \
1223 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1224# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1225# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1226# define isLOWER_L1(c) (isLOWER_A(c) \
1227 || (FITS_IN_8_BITS(c) \
1228 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1229 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1230 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1231 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1232 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1233# define isPRINT_L1(c) (isPRINT_A(c) \
1234 || (FITS_IN_8_BITS(c) \
1235 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1236# define isPUNCT_L1(c) (isPUNCT_A(c) \
1237 || (FITS_IN_8_BITS(c) \
1238 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1239 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1240 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1241 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1242 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1243 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1244 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1245# define isSPACE_L1(c) (isSPACE_A(c) \
1246 || (FITS_IN_8_BITS(c) \
1247 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1248 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1249# define isUPPER_L1(c) (isUPPER_A(c) \
1250 || (FITS_IN_8_BITS(c) \
1251 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1252 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1253 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1254# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1255# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1256# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1257 || isBLANK_L1(c) \
1258 || (c) == '-' \
1259 || (c) == '(' \
1260 || (c) == ')')
1261 /* The following are not fully accurate in the above-ASCII range. I (khw)
1262 * don't think it's necessary to be so for the purposes where this gets
1263 * compiled */
1264# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1265# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1266
1267 /* And these aren't accurate at all. They are useful only for above
1268 * Latin1, which utilities and bootstrapping don't deal with */
1269# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1270# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1271# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1272
1273 /* Many of the macros later in this file are defined in terms of these. By
1274 * implementing them with a function, which converts the class number into
1275 * a call to the desired macro, all of the later ones work. However, that
1276 * function won't be actually defined when building a utility program (no
1277 * perl.h), and so a compiler error will be generated if one is attempted
1278 * to be used. And the above-Latin1 code points require Unicode tables to
1279 * be present, something unlikely to be the case when bootstrapping */
1280# define _generic_isCC(c, classnum) \
1281 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1282# define _generic_isCC_A(c, classnum) \
1283 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1284#endif /* End of no perl.h H_PERL */
8a58bdcf 1285
e66b99e9
KW
1286#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1287#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1288#define isASCII_A(c) isASCII(c)
1289#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1290#define isBLANK(c) isBLANK_A(c)
1291#define isCNTRL(c) isCNTRL_A(c)
1292#define isDIGIT(c) isDIGIT_A(c)
1293#define isGRAPH(c) isGRAPH_A(c)
1294#define isIDFIRST(c) isIDFIRST_A(c)
1295#define isLOWER(c) isLOWER_A(c)
1296#define isPRINT(c) isPRINT_A(c)
779cf272 1297#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1298#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1299#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1300#define isPUNCT(c) isPUNCT_A(c)
1301#define isSPACE(c) isSPACE_A(c)
1302#define isUPPER(c) isUPPER_A(c)
1303#define isWORDCHAR(c) isWORDCHAR_A(c)
1304#define isXDIGIT(c) isXDIGIT_A(c)
1305
1306/* ASCII casing. These could also be written as
1307 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1308 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1309 which uses table lookup and mask instead of subtraction. (This would
1310 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1311#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1312#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1313
25200305
KW
1314/* In the ASCII range, these are equivalent to what they're here defined to be.
1315 * But by creating these definitions, other code doesn't have to be aware of
1316 * this detail */
1317#define toFOLD(c) toLOWER(c)
25200305
KW
1318#define toTITLE(c) toUPPER(c)
1319
c753c8d3
KW
1320#define toLOWER_A(c) toLOWER(c)
1321#define toUPPER_A(c) toUPPER(c)
25200305
KW
1322#define toFOLD_A(c) toFOLD(c)
1323#define toTITLE_A(c) toTITLE(c)
1a0901db 1324
4650c663 1325/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1326#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1327 ? (c) \
f4cd282c 1328 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1329#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1330
1a0901db 1331/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1332 * all mapped to one of them, and these need special handling; returns the
1333 * input itself if is out-of-range */
b2bf251f 1334#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1335 ? (c) \
f4cd282c 1336 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1337#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1338
beab9ebe
KW
1339/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1340
1341/* For internal core Perl use only: the base macro for defining macros like
1342 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1343 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1344 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1345 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1346 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1347 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1348 * code points under locale rules, use the macros based on _generic_LC_uvchr
1349 * instead of this. */
beab9ebe
KW
1350#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1351 (! FITS_IN_8_BITS(c) \
1352 ? 0 \
31f05a37
KW
1353 : IN_UTF8_CTYPE_LOCALE \
1354 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1355 : cBOOL(non_utf8))
1356
1357/* For internal core Perl use only: a helper macro for defining macros like
1358 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1359 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1360 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1361 * _generic_LC_base for more info */
1362#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1363 _generic_LC_base(c,utf8_locale_classnum, \
1364 non_utf8_func( (_LC_CAST) (c)))
1365
1366/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1367 * 'c' is the platform's native underscore character */
1368#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1369 _generic_LC_base(c, utf8_locale_classnum, \
1370 (non_utf8_func( (_LC_CAST) (c)) \
1371 || (char)(c) == '_'))
1372
1373/* These next three are also for internal core Perl use only: case-change
1374 * helper macros */
1375#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1376 ? (c) \
31f05a37
KW
1377 : (IN_UTF8_CTYPE_LOCALE) \
1378 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1379 : (cast)function((cast)(c)))
beab9ebe 1380
31f05a37
KW
1381/* Note that the result can be larger than a byte in a UTF-8 locale. It
1382 * returns a single value, so can't adequately return the upper case of LATIN
1383 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1384 * values "SS"); instead it asserts against that under DEBUGGING, and
1385 * otherwise returns its input */
beab9ebe
KW
1386#define _generic_toUPPER_LC(c, function, cast) \
1387 (! FITS_IN_8_BITS(c) \
1388 ? (c) \
31f05a37 1389 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1390 ? (cast)function((cast)(c)) \
31f05a37
KW
1391 : ((((U8)(c)) == MICRO_SIGN) \
1392 ? GREEK_CAPITAL_LETTER_MU \
1393 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1394 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1395 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1396 ? (__ASSERT_(0) (c)) \
1397 : PL_mod_latin1_uc[ (U8) (c) ])))))
1398
1399/* Note that the result can be larger than a byte in a UTF-8 locale. It
1400 * returns a single value, so can't adequately return the fold case of LATIN
1401 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1402 * values "ss"); instead it asserts against that under DEBUGGING, and
1403 * otherwise returns its input */
beab9ebe 1404#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1405 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1406 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1407 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1408 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1409 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1410
84061b6a 1411/* Use the libc versions for these if available. */
f05550c0 1412#if defined(HAS_ISASCII)
84061b6a
KW
1413# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1414#else
1415# define isASCII_LC(c) isASCII(c)
1416#endif
1417
f05550c0 1418#if defined(HAS_ISBLANK)
84061b6a 1419# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1420#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1421# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1422#endif
1423
f05550c0 1424#define _LC_CAST U8
bbce6d69 1425
f05550c0 1426#ifdef WIN32
375f5f06
KW
1427 /* The Windows functions don't bother to follow the POSIX standard, which
1428 * for example says that something can't both be a printable and a control.
1429 * But Windows treats the \t control as a printable, and does such things
1430 * as making superscripts into both digits and punctuation. This tames
1431 * these flaws by assuming that the definitions of both controls and space
1432 * are correct, and then making sure that other definitions don't have
1433 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1434 * Not all possible weirdnesses are checked for, just the ones that were
1435 * detected on actual Microsoft code pages */
1436
b7d90381
KW
1437# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1438# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1439
1440# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1441 && isALPHANUMERIC_LC(c))
1442# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1443 ! isPUNCT_LC(c))
1444# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1445 isALPHANUMERIC_LC(c))
1446# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1447# define isIDFIRST_LC(c) (((c) == '_') \
1448 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1449# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1450# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1451# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1452# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1453# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1454# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1455 && isALPHANUMERIC_LC(c))
f05550c0
BF
1456
1457# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1458# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1459# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1460
1461#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1462 /* For most other platforms */
beab9ebe 1463
f05550c0
BF
1464# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1465# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1466# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1467# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1468# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1469# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1470# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1471# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1472# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1473# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1474# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1475# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1476# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1477
1478
1479# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1480# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1481# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1482
1483#else /* The final fallback position */
1484
b7d90381
KW
1485# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1486# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1487# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1488# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1489# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1490# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1491# define isLOWER_LC(c) (isascii(c) && islower(c))
1492# define isPRINT_LC(c) (isascii(c) && isprint(c))
1493# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1494# define isSPACE_LC(c) (isascii(c) && isspace(c))
1495# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1496# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1497# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1498
1499# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1500# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1501# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1502
f05550c0 1503#endif
55204971 1504
eba68aa0
KW
1505#define isIDCONT(c) isWORDCHAR(c)
1506#define isIDCONT_A(c) isWORDCHAR_A(c)
1507#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1508#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1509#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1510
4650c663
KW
1511/* For internal core Perl use only: the base macros for defining macros like
1512 * isALPHA_uni. 'c' is the code point to check. 'classnum' is the POSIX class
1513 * number defined earlier in this file. _generic_uni() is used for POSIX
1514 * classes where there is a macro or function 'above_latin1' that takes the
1515 * single argument 'c' and returns the desired value. These exist for those
1516 * classes which have simple definitions, avoiding the overhead of a hash
1517 * lookup or inversion list binary search. _generic_swash_uni() can be used
1518 * for classes where that overhead is faster than a direct lookup.
1519 * _generic_uni() won't compile if 'c' isn't unsigned, as it won't match the
1520 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1521 * have duplicate checks here, so could create versions of the macros that
1522 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1523
1524/* Note that all ignore 'use bytes' */
cd500f2f
KW
1525#define _generic_uni(classnum, above_latin1, c) ((c) < 256 \
1526 ? _generic_isCC(c, classnum) \
1527 : above_latin1(c))
1528#define _generic_swash_uni(classnum, c) ((c) < 256 \
1529 ? _generic_isCC(c, classnum) \
922e8cb4 1530 : _is_uni_FOO(classnum, c))
cd500f2f
KW
1531#define isALPHA_uni(c) _generic_swash_uni(_CC_ALPHA, c)
1532#define isALPHANUMERIC_uni(c) _generic_swash_uni(_CC_ALPHANUMERIC, c)
509fb054
KW
1533#define isASCII_uni(c) isASCII(c)
1534#define isBLANK_uni(c) _generic_uni(_CC_BLANK, is_HORIZWS_cp_high, c)
1535#define isCNTRL_uni(c) isCNTRL_L1(c) /* All controls are in Latin1 */
cd500f2f
KW
1536#define isDIGIT_uni(c) _generic_swash_uni(_CC_DIGIT, c)
1537#define isGRAPH_uni(c) _generic_swash_uni(_CC_GRAPH, c)
eba68aa0 1538#define isIDCONT_uni(c) _generic_uni(_CC_WORDCHAR, _is_uni_perl_idcont, c)
509fb054 1539#define isIDFIRST_uni(c) _generic_uni(_CC_IDFIRST, _is_uni_perl_idstart, c)
cd500f2f
KW
1540#define isLOWER_uni(c) _generic_swash_uni(_CC_LOWER, c)
1541#define isPRINT_uni(c) _generic_swash_uni(_CC_PRINT, c)
aea0c08c 1542
cd500f2f 1543#define isPUNCT_uni(c) _generic_swash_uni(_CC_PUNCT, c)
509fb054 1544#define isSPACE_uni(c) _generic_uni(_CC_SPACE, is_XPERLSPACE_cp_high, c)
779cf272
KW
1545#define isPSXSPC_uni(c) isSPACE_uni(c)
1546
cd500f2f 1547#define isUPPER_uni(c) _generic_swash_uni(_CC_UPPER, c)
509fb054 1548#define isVERTWS_uni(c) _generic_uni(_CC_VERTSPACE, is_VERTWS_cp_high, c)
cd500f2f 1549#define isWORDCHAR_uni(c) _generic_swash_uni(_CC_WORDCHAR, c)
509fb054 1550#define isXDIGIT_uni(c) _generic_uni(_CC_XDIGIT, is_XDIGIT_cp_high, c)
b18192e9 1551
b8d68ded 1552#define toFOLD_uni(c,s,l) to_uni_fold(c,s,l)
509fb054
KW
1553#define toLOWER_uni(c,s,l) to_uni_lower(c,s,l)
1554#define toTITLE_uni(c,s,l) to_uni_title(c,s,l)
1555#define toUPPER_uni(c,s,l) to_uni_upper(c,s,l)
a0ed51b3 1556
4650c663
KW
1557/* For internal core Perl use only: the base macros for defining macros like
1558 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1559 * point, not just 0-255. Like _generic_uni, there are two versions, one for
1560 * simple class definitions; the other for more complex. These are like
1561 * _generic_uni, so see it for more info. */
cd500f2f
KW
1562#define _generic_LC_uvchr(latin1, above_latin1, c) \
1563 (c < 256 ? latin1(c) : above_latin1(c))
1564#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1565 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1566
1567#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1568#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1569 _CC_ALPHANUMERIC, c)
b7d90381
KW
1570#define isASCII_LC_uvchr(c) isASCII_LC(c)
1571#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1572 is_HORIZWS_cp_high, c)
feeab5a9 1573#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1574#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1575#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1576#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1577 _is_uni_perl_idcont, c)
b7d90381 1578#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1579 _is_uni_perl_idstart, c)
1580#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1581#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1582#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1583#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1584#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1585 is_XPERLSPACE_cp_high, c)
cd500f2f 1586#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1587#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1588 _CC_WORDCHAR, c)
b7d90381
KW
1589#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1590 is_XDIGIT_cp_high, c)
e712593e 1591
b7d90381 1592#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1593
4650c663
KW
1594/* For internal core Perl use only: the base macros for defining macros like
1595 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1596 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1597 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1598 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1599 * same representation whether utf8 or not. Note that it assumes that the utf8
1600 * has been validated, and ignores 'use bytes' */
cd500f2f 1601#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1602 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1603 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1604 ? _generic_isCC( \
a62b247b 1605 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
aea0c08c 1606 *((p)+1 )), \
94bb8c36 1607 classnum) \
252810d0 1608 : utf8)
b7d90381
KW
1609/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1610 * 'above_latin1' can be a macro */
cd500f2f
KW
1611#define _generic_func_utf8(classnum, above_latin1, p) \
1612 _generic_utf8(classnum, p, above_latin1(p))
4650c663 1613/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1614 * 'above_latin1' parameter */
922e8cb4 1615#define _generic_swash_utf8(classnum, p) \
cd500f2f 1616 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
922e8cb4 1617
cc8ab7c0 1618/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1619 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1620 * class is TRUE for. Hence it can skip the tests for this range.
1621 * 'above_latin1' should include its arguments */
3ded5eb0 1622#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1623 (UTF8_IS_INVARIANT(*(p)) \
1624 ? _generic_isCC(*(p), classnum) \
1625 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1626 ? above_latin1 \
cc8ab7c0
KW
1627 : 0)
1628
84238efa
KW
1629/* NOTE that some of these macros have very similar ones in regcharclass.h.
1630 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1631 * there, differing in name only by an underscore from the one here
1632 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1633 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1634 * points; the regcharclass.h ones are implemented as a series of
1635 * "if-else-if-else ..." */
1636
b7d90381
KW
1637#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1638#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1639#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
509fb054
KW
1640 utf8, the non-utf8 macro works
1641 */
b7d90381 1642#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2
KW
1643
1644#ifdef EBCDIC
1645 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1646 * more efficient macro instead of the more general one */
88794300 1647# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
e8fa43e2
KW
1648#else
1649# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1650#endif
1651
cd500f2f
KW
1652#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1653 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1654#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1655#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1656 _is_utf8_perl_idcont, p)
e5dcd934 1657
c11ff943
KW
1658/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1659 * IDFIRST is an alnum. See
c8362b00 1660 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1661 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1662 * This used to be not the XID version, but we decided to go with the more
1663 * modern Unicode definition */
b7d90381 1664#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1665 _is_utf8_perl_idstart, p)
e5dcd934 1666
b7d90381
KW
1667#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1668#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1669#define isPSXSPC_utf8(p) isSPACE_utf8(p)
1670#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1671#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1672#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1673#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1674#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1675#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
cd500f2f 1676 is_XDIGIT_high(p))
a0ed51b3 1677
7d2b61dd 1678#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1679#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1680#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1681#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1682
4650c663
KW
1683/* For internal core Perl use only: the base macros for defining macros like
1684 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1685 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1686 * 'macro' parameter */
94bb8c36
KW
1687#define _generic_LC_utf8(macro, p, utf8) \
1688 (UTF8_IS_INVARIANT(*(p)) \
1689 ? macro(*(p)) \
1690 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
a62b247b 1691 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
9201a572
KW
1692 : utf8)
1693
cd500f2f
KW
1694#define _generic_LC_swash_utf8(macro, classnum, p) \
1695 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1696#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1697 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1698
b7d90381 1699#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
cd500f2f 1700 _CC_ALPHANUMERIC, p)
b7d90381
KW
1701#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1702#define isASCII_LC_utf8(p) isASCII_LC(*p)
1703#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1704 is_HORIZWS_high, p)
1705#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1706#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1707#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1708#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1709 _is_utf8_perl_idcont, p)
1710#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1711 _is_utf8_perl_idstart, p)
1712#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1713#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1714#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1715#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1716#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1717 is_XPERLSPACE_high, p)
1718#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
cd500f2f
KW
1719#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1720 _CC_WORDCHAR, p)
b7d90381
KW
1721#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1722 is_XDIGIT_high, p)
aaa51d5e 1723
fbc19f27
KW
1724/* Macros for backwards compatibility and for completeness when the ASCII and
1725 * Latin1 values are identical */
b7d90381
KW
1726#define isALPHAU(c) isALPHA_L1(c)
1727#define isDIGIT_L1(c) isDIGIT_A(c)
1728#define isOCTAL(c) isOCTAL_A(c)
1729#define isOCTAL_L1(c) isOCTAL_A(c)
1730#define isXDIGIT_L1(c) isXDIGIT_A(c)
1731#define isALNUM(c) isWORDCHAR(c)
1732#define isALNUMU(c) isWORDCHAR_L1(c)
1733#define isALNUM_LC(c) isWORDCHAR_LC(c)
1734#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 1735#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
1736#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1737#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1738#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1739#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1740#define isALNUMC(c) isALPHANUMERIC(c)
1741#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1742#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 1743#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 1744#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 1745#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1746
2bd1cbf6
KW
1747/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1748 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1749 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1750 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1751 * the outlier from the block that contains the other controls, just like
1752 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1753 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
1754 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1755 * etc. */
2bd1cbf6 1756#ifndef EBCDIC
75763b3a 1757# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 1758#else
75763b3a
KW
1759# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1760 ((isPRINT_A(c)) \
1761 ? (UNLIKELY((c) == '?') \
1762 ? QUESTION_MARK_CTRL \
1763 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1764 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1765 ? '?' \
1766 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 1767#endif
bbce6d69 1768
dea28490
JJ
1769/* Line numbers are unsigned, 32 bits. */
1770typedef U32 line_t;
e5dcd934 1771#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 1772
91152fc1
DG
1773/* Helpful alias for version prescan */
1774#define is_LAX_VERSION(a,b) \
1775 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1776
1777#define is_STRICT_VERSION(a,b) \
1778 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1779
1780#define BADVERSION(a,b,c) \
1781 if (b) { \
1782 *b = c; \
1783 } \
1784 return a;
8c52afec 1785
cb27eebd
KW
1786/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1787 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1788 * which is advanced. The input is validated only by an assert() in DEBUGGING
1789 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1790 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1791#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1792 ? (c) \
1793 : ((c) + 9))))
902008b8
KW
1794#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1795 ? (*(s)++) \
1796 : (*(s)++ + 9))))
95a59cab 1797
cb27eebd
KW
1798/* Converts a character known to represent an octal digit (0-7) to its numeric
1799 * value. The input is validated only by an assert() in DEBUGGING builds. In
1800 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1801#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1802
305b8651
KW
1803/* Efficiently returns a boolean as to if two native characters are equivalent
1804 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1805 * the ALPHA in the name is to remind you of that. This is asserted() in
1806 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1807 * works (on valid input) for both non- and UTF-8-encoded bytes.
1808 *
1809 * When one of the inputs is a compile-time constant and gets folded by the
1810 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1811 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1812 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
1813 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1814 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1815 * just a single 0, in the bit position where the upper- and lowercase differ.
1816 * */
305b8651
KW
1817#define isALPHA_FOLD_EQ(c1, c2) \
1818 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1819 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1820#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1821
8e84507e 1822/*
ccfc67b7
JH
1823=head1 Memory Management
1824
a02a5408 1825=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
1826The XSUB-writer's interface to the C C<malloc> function.
1827
0d7b2759
MH
1828Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1829
c5008215
JC
1830In 5.9.3, Newx() and friends replace the older New() API, and drops
1831the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 1832themselves. This aid has been superseded by a new build option,
d10b4965 1833PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
1834there for use in XS modules supporting older perls.
1835
a02a5408 1836=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 1837The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 1838cast. See also C<L</Newx>>.
954c1994 1839
0d7b2759
MH
1840Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1841
a02a5408 1842=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 1843The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 1844memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 1845
0d7b2759
MH
1846Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1847
954c1994
GS
1848=for apidoc Am|void|Renew|void* ptr|int nitems|type
1849The XSUB-writer's interface to the C C<realloc> function.
1850
0d7b2759
MH
1851Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1852
954c1994
GS
1853=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1854The XSUB-writer's interface to the C C<realloc> function, with
1855cast.
1856
0d7b2759
MH
1857Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1858
49b8b560 1859=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
1860The XSUB-writer's interface to the C C<free> function.
1861
0d7b2759
MH
1862This should B<ONLY> be used on memory obtained using L<"Newx"> and friends.
1863
954c1994
GS
1864=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1865The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 1866source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1867C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 1868
e90e2364 1869=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 1870Like C<Move> but returns C<dest>. Useful
72d33970 1871for encouraging compilers to tail-call
e90e2364
NC
1872optimise.
1873
954c1994
GS
1874=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1875The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 1876source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1877C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 1878
e90e2364
NC
1879=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1880
796b6530 1881Like C<Copy> but returns C<dest>. Useful
72d33970 1882for encouraging compilers to tail-call
e90e2364
NC
1883optimise.
1884
954c1994
GS
1885=for apidoc Am|void|Zero|void* dest|int nitems|type
1886
1887The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1888destination, C<nitems> is the number of items, and C<type> is the type.
1889
e90e2364
NC
1890=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1891
72d33970
FC
1892Like C<Zero> but returns dest. Useful
1893for encouraging compilers to tail-call
e90e2364
NC
1894optimise.
1895
da5d8dbb 1896=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 1897This is an architecture-independent macro to copy one structure to another.
954c1994 1898
7e337ee0
JH
1899=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1900
1901Fill up memory with a byte pattern (a byte repeated over and over
1902again) that hopefully catches attempts to access uninitialized memory.
1903
1904=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1905
1906PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1907
1c12ffb4 1908=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
1909
1910PoisonWith(0xEF) for catching access to freed memory.
1911
9965345d
JH
1912=for apidoc Am|void|Poison|void* dest|int nitems|type
1913
7e337ee0 1914PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
1915
1916=cut */
954c1994 1917
561b68a9
SH
1918/* Maintained for backwards-compatibility only. Use newSV() instead. */
1919#ifndef PERL_CORE
ff06c60c 1920#define NEWSV(x,len) newSV(len)
561b68a9 1921#endif
ff06c60c 1922
19a94d75
JH
1923#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1924
e6bdf523 1925
27d5b266 1926#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
1927
1928/* This expression will be constant-folded at compile time. It checks
1929 * whether or not the type of the count n is so small (e.g. U8 or U16, or
1930 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
1931 * As well as avoiding the need for a run-time check in some cases, it's
1932 * designed to avoid compiler warnings like:
1933 * comparison is always false due to limited range of data type
73e8ff00
DM
1934 * It's mathematically equivalent to
1935 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
1936 */
1937
1938# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
73e8ff00
DM
1939 ( sizeof(MEM_SIZE) < sizeof(n) \
1940 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 1941
88f9f128 1942/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
1943 * compiler warnings. We *want* to write the expression as
1944 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
1945 * (for some compile-time constant C), but even when the LHS
1946 * constant-folds to false at compile-time, g++ insists on emitting
1947 * warnings about the RHS (e.g. "comparison is always false"), so instead
1948 * we write it as
e6bdf523 1949 *
d98e5cde 1950 * (cond ? n : X) > C
88f9f128 1951 *
d98e5cde
DM
1952 * where X is a constant with X > C always false. Choosing a value for X
1953 * is tricky. If 0, some compilers will complain about 0 > C always being
1954 * false; if 1, Coverity complains when n happens to be the constant value
1955 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
1956 * for X and hope that nothing else whines.
e6bdf523
DM
1957 */
1958
1959# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
1960 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
1961 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
1962
1963# define MEM_WRAP_CHECK(n,t) \
1964 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1965 && (croak_memory_wrap(),0))
1966
1967# define MEM_WRAP_CHECK_1(n,t,a) \
1968 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1969 && (Perl_croak_nocontext("%s",(a)),0))
1970
8b44ba4c 1971#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 1972
d1decf2b 1973#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
1974#else
1975
410319be
NC
1976#define MEM_WRAP_CHECK(n,t)
1977#define MEM_WRAP_CHECK_1(n,t,a)
1978#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
1979#define MEM_WRAP_CHECK_(n,t)
1980
6fff79ce 1981#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 1982
1936d2a7 1983#endif
8b44ba4c 1984
fe4f188c 1985#ifdef PERL_MEM_LOG
46c6c7e2 1986/*
9f653bb5 1987 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
1988 * go through functions, which are handy for debugging breakpoints, but
1989 * which more importantly get the immediate calling environment (file and
e352bcff
JH
1990 * line number, and C function name if available) passed in. This info can
1991 * then be used for logging the calls, for which one gets a sample
73d1d973 1992 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 1993 *
46c6c7e2 1994 * Known problems:
94e892a6 1995 * - not all memory allocs get logged, only those
46c6c7e2 1996 * that go through Newx() and derivatives (while all
94e892a6 1997 * Safefrees do get logged)
46c6c7e2
JH
1998 * - __FILE__ and __LINE__ do not work everywhere
1999 * - __func__ or __FUNCTION__ even less so
2000 * - I think more goes on after the perlio frees but
2001 * the thing is that STDERR gets closed (as do all
2002 * the file descriptors)
2003 * - no deeper calling stack than the caller of the Newx()
2004 * or the kind, but do I look like a C reflection/introspection
2005 * utility to you?
2006 * - the function prototypes for the logging functions
2007 * probably should maybe be somewhere else than handy.h
2008 * - one could consider inlining (macrofying) the logging
2009 * for speed, but I am too lazy
2010 * - one could imagine recording the allocations in a hash,
2011 * (keyed by the allocation address?), and maintain that
2012 * through reallocs and frees, but how to do that without
2013 * any News() happening...?
73d1d973 2014 * - lots of -Ddefines to get useful/controllable output
b953482e 2015 * - lots of ENV reads
46c6c7e2
JH
2016 */
2017
fc6bde6f 2018PERL_CALLCONV Malloc_t Perl_mem_log_alloc(const UV n, const UV typesize, const char *type_name, Malloc_t newalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 2019
fc6bde6f 2020PERL_CALLCONV Malloc_t Perl_mem_log_realloc(const UV n, const UV typesize, const char *type_name, Malloc_t oldalloc, Malloc_t newalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 2021
fc6bde6f 2022PERL_CALLCONV Malloc_t Perl_mem_log_free(Malloc_t oldalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 2023
0b0ab801 2024# ifdef PERL_CORE
73d1d973 2025# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2026enum mem_log_type {
2027 MLT_ALLOC,
2028 MLT_REALLOC,
d7a2c63c
MHM
2029 MLT_FREE,
2030 MLT_NEW_SV,
2031 MLT_DEL_SV
0b0ab801
MHM
2032};
2033# endif
12754f92 2034# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2035void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2036void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2037# endif
0b0ab801
MHM
2038# endif
2039
fe4f188c
JH
2040#endif
2041
2042#ifdef PERL_MEM_LOG
d1401ee9
MHM
2043#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2044#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2045#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2046#endif
2047
2048#ifndef MEM_LOG_ALLOC
2049#define MEM_LOG_ALLOC(n,t,a) (a)
2050#endif
2051#ifndef MEM_LOG_REALLOC
2052#define MEM_LOG_REALLOC(n,t,v,a) (a)
2053#endif
2054#ifndef MEM_LOG_FREE
2055#define MEM_LOG_FREE(a) (a)
2056#endif
2057
d1401ee9
MHM
2058#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2059#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2060#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2061
2062#ifndef PERL_CORE
a02a5408
JC
2063/* pre 5.9.x compatibility */
2064#define New(x,v,n,t) Newx(v,n,t)
2065#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2066#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2067#endif
a02a5408 2068
ff68c719 2069#define Renew(v,n,t) \
d1401ee9 2070 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2071#define Renewc(v,n,t,c) \
d1401ee9 2072 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2073
2074#ifdef PERL_POISON
2075#define Safefree(d) \
06c0cc96 2076 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2077#else
fe4f188c 2078#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2079#endif
55497cff 2080
8b44ba4c
NC
2081#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2082#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2083#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2084
8b44ba4c
NC
2085#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2086#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2087#ifdef HAS_MEMSET
8b44ba4c 2088#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2089#else
8b44ba4c
NC
2090/* Using bzero(), which returns void. */
2091#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2092#endif
2093
7e337ee0
JH
2094#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2095#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2096#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2097#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2098
caa674f3
DD
2099#ifdef PERL_POISON
2100# define PERL_POISON_EXPR(x) x
2101#else
2102# define PERL_POISON_EXPR(x)
2103#endif
2104
2304df62 2105#ifdef USE_STRUCT_COPY
ff68c719 2106#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2107#else
2108#define StructCopy(s,d,t) Copy(s,d,1,t)
2109#endif
2cc61e15 2110
c3caa5c3
JH
2111/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2112 * want your zero-based indices to be less than but not equal to).
2113 *
2114 * C_ARRAY_END is one past the last: half-open/half-closed range,
2115 * not last-inclusive range. */
622913ab 2116#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2117#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2118
2cc61e15
DD
2119#ifdef NEED_VA_COPY
2120# ifdef va_copy
2121# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2122# else
a1866d1b
JH
2123# if defined(__va_copy)
2124# define Perl_va_copy(s, d) __va_copy(d, s)
2125# else
2126# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2127# endif
2cc61e15
DD
2128# endif
2129#endif
2130
472d47bc
SB
2131/* convenience debug macros */
2132#ifdef USE_ITHREADS
2133#define pTHX_FORMAT "Perl interpreter: 0x%p"
2134#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2135#define pTHX_VALUE_ (void *)my_perl,
2136#define pTHX_VALUE (void *)my_perl
2137#define pTHX__VALUE_ ,(void *)my_perl,
2138#define pTHX__VALUE ,(void *)my_perl
472d47bc 2139#else
3609ea0d 2140#define pTHX_FORMAT
472d47bc 2141#define pTHX__FORMAT
3609ea0d 2142#define pTHX_VALUE_
472d47bc 2143#define pTHX_VALUE
3609ea0d 2144#define pTHX__VALUE_
472d47bc
SB
2145#define pTHX__VALUE
2146#endif /* USE_ITHREADS */
3609ea0d 2147
2acdbac1
NC
2148/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2149 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2150 CPAN::Unpack show any users outside the core. */
2151#ifdef PERL_CORE
b7d90381
KW
2152# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2153 "Use of " s " is deprecated")
2acdbac1
NC
2154#endif
2155
dfff4baf
BF
2156/* Internal macros to deal with gids and uids */
2157#ifdef PERL_CORE
2158
2159# if Uid_t_size > IVSIZE
2160# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2161# define SvUID(sv) SvNV(sv)
2162# else
2163# if Uid_t_sign <= 0
2164# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2165# define SvUID(sv) SvIV(sv)
2166# else
2167# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2168# define SvUID(sv) SvUV(sv)
2169# endif
2170# endif /* Uid_t_size */
2171
2172# if Gid_t_size > IVSIZE
2173# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2174# define SvGID(sv) SvNV(sv)
2175# else
2176# if Gid_t_sign <= 0
2177# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2178# define SvGID(sv) SvIV(sv)
2179# else
2180# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2181# define SvGID(sv) SvUV(sv)
2182# endif
2183# endif /* Gid_t_size */
2184
2185#endif
2186
9d745869
KW
2187#endif /* HANDY_H */
2188
e9a8c099 2189/*
14d04a33 2190 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2191 */