This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
charclass_invlists.h: Add regen entry
[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;
6b8eaf93
JH
176#ifdef PERL_CORE
177# ifdef HAS_QUAD
8175356b
JH
178typedef I64TYPE I64;
179typedef U64TYPE U64;
6b8eaf93
JH
180# endif
181#endif /* PERL_CORE */
8175356b 182
9c39b21c
JH
183/* INT64_C/UINT64_C are C99 from <stdint.h> (so they will not be
184 * available in strict C89 mode), but they are nice, so let's define
185 * them if necessary. */
0d3bd0a9 186#if defined(HAS_QUAD)
9c39b21c
JH
187# undef PeRl_INT64_C
188# undef PeRl_UINT64_C
189/* Prefer the native integer types (int and long) over long long
190 * (which is not C89) and Win32-specific __int64. */
191# if QUADKIND == QUAD_IS_INT && INTSIZE == 8
192# define PeRl_INT64_C(c) (c)
193# define PeRl_UINT64_C(c) CAT2(c,U)
194# endif
195# if QUADKIND == QUAD_IS_LONG && LONGSIZE == 8
196# define PeRl_INT64_C(c) CAT2(c,L)
197# define PeRl_UINT64_C(c) CAT2(c,UL)
198# endif
199# if QUADKIND == QUAD_IS_LONG_LONG && defined(HAS_LONG_LONG)
200# define PeRl_INT64_C(c) CAT2(c,LL)
201# define PeRl_UINT64_C(c) CAT2(c,ULL)
202# endif
203# if QUADKIND == QUAD_IS___INT64
204# define PeRl_INT64_C(c) CAT2(c,I64)
205# define PeRl_UINT64_C(c) CAT2(c,UI64)
206# endif
207# ifndef PeRl_INT64_C
208# define PeRl_INT64_C(c) ((I64TYPE)(c)) /* last resort */
209# define PeRl_UINT64_C(c) ((U64TYPE)(c))
210# endif
211/* In OS X the INT64_C/UINT64_C are defined with LL/ULL, which will
212 * not fly with C89-pedantic gcc, so let's undefine them first so that
213 * we can redefine them with our native integer preferring versions. */
8eafc6eb 214# if defined(PERL_DARWIN) && defined(PERL_GCC_PEDANTIC)
9c39b21c
JH
215# undef INT64_C
216# undef UINT64_C
217# endif
218# ifndef INT64_C
219# define INT64_C(c) PeRl_INT64_C(c)
220# endif
221# ifndef UINT64_C
222# define UINT64_C(c) PeRl_UINT64_C(c)
223# endif
e8c95190 224#endif
dc750f2e 225
d8668976 226#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 227
5ff3f7a4
GS
228/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
229 Please search CHAR_MAX in perl.h for further details. */
230#define U8_MAX UINT8_MAX
231#define U8_MIN UINT8_MIN
232
5ff3f7a4
GS
233#define I16_MAX INT16_MAX
234#define I16_MIN INT16_MIN
235#define U16_MAX UINT16_MAX
236#define U16_MIN UINT16_MIN
237
5ff3f7a4
GS
238#define I32_MAX INT32_MAX
239#define I32_MIN INT32_MIN
0e983133
GS
240#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
241# define U32_MAX UINT32_MAX
242#else
243# define U32_MAX 4294967295U
244#endif
5ff3f7a4
GS
245#define U32_MIN UINT32_MIN
246
247#else
248
5c9fa16e
KA
249/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
250 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
251#define U8_MAX PERL_UCHAR_MAX
252#define U8_MIN PERL_UCHAR_MIN
79072805 253
27d4fb96
PP
254#define I16_MAX PERL_SHORT_MAX
255#define I16_MIN PERL_SHORT_MIN
256#define U16_MAX PERL_USHORT_MAX
257#define U16_MIN PERL_USHORT_MIN
79072805 258
c4f23d77 259#if LONGSIZE > 4
27d4fb96
PP
260# define I32_MAX PERL_INT_MAX
261# define I32_MIN PERL_INT_MIN
262# define U32_MAX PERL_UINT_MAX
263# define U32_MIN PERL_UINT_MIN
79072805 264#else
27d4fb96
PP
265# define I32_MAX PERL_LONG_MAX
266# define I32_MIN PERL_LONG_MIN
267# define U32_MAX PERL_ULONG_MAX
268# define U32_MIN PERL_ULONG_MIN
79072805
LW
269#endif
270
5ff3f7a4
GS
271#endif
272
58a9a5d5 273/* log(2) is pretty close to 0.30103, just in case anyone is grepping for it */
fc36a67e
PP
274#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
275#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
276#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
277
88794300 278/* Unused by core; should be deprecated */
ff68c719 279#define Ctl(ch) ((ch) & 037)
8d063cd8 280
4650c663
KW
281/* This is a helper macro to avoid preprocessor issues, replaced by nothing
282 * unless under DEBUGGING, where it expands to an assert of its argument,
283 * followed by a comma (hence the comma operator). If we just used a straight
284 * assert(), we would get a comma with nothing before it when not DEBUGGING */
0f092d08
KW
285#ifdef DEBUGGING
286# define __ASSERT_(statement) assert(statement),
287#else
288# define __ASSERT_(statement)
289#endif
290
3fe05580
MHM
291/*
292=head1 SV-Body Allocation
293
294=for apidoc Ama|SV*|newSVpvs|const char* s
30a15352
KW
295Like C<newSVpvn>, but takes a literal C<NUL>-terminated string instead of a
296string/length pair.
3fe05580 297
84bafc02 298=for apidoc Ama|SV*|newSVpvs_flags|const char* s|U32 flags
30a15352
KW
299Like C<newSVpvn_flags>, but takes a literal C<NUL>-terminated string instead of
300a string/length pair.
84bafc02 301
3fe05580 302=for apidoc Ama|SV*|newSVpvs_share|const char* s
30a15352
KW
303Like C<newSVpvn_share>, but takes a literal C<NUL>-terminated string instead of
304a string/length pair and omits the hash parameter.
3fe05580 305
9dcc53ea 306=for apidoc Am|void|sv_catpvs_flags|SV* sv|const char* s|I32 flags
30a15352
KW
307Like C<sv_catpvn_flags>, but takes a literal C<NUL>-terminated string instead
308of a string/length pair.
9dcc53ea
Z
309
310=for apidoc Am|void|sv_catpvs_nomg|SV* sv|const char* s
311Like C<sv_catpvn_nomg>, but takes a literal string instead of a
312string/length pair.
313
7b4e7469 314=for apidoc Am|void|sv_catpvs|SV* sv|const char* s
3fe05580
MHM
315Like C<sv_catpvn>, but takes a literal string instead of a string/length pair.
316
9dcc53ea
Z
317=for apidoc Am|void|sv_catpvs_mg|SV* sv|const char* s
318Like C<sv_catpvn_mg>, but takes a literal string instead of a
319string/length pair.
320
7b4e7469 321=for apidoc Am|void|sv_setpvs|SV* sv|const char* s
3fe05580
MHM
322Like C<sv_setpvn>, but takes a literal string instead of a string/length pair.
323
9dcc53ea
Z
324=for apidoc Am|void|sv_setpvs_mg|SV* sv|const char* s
325Like C<sv_setpvn_mg>, but takes a literal string instead of a
326string/length pair.
327
328=for apidoc Am|SV *|sv_setref_pvs|const char* s
329Like C<sv_setref_pvn>, but takes a literal string instead of a
330string/length pair.
331
3fe05580
MHM
332=head1 Memory Management
333
334=for apidoc Ama|char*|savepvs|const char* s
30a15352
KW
335Like C<savepvn>, but takes a literal C<NUL>-terminated string instead of a
336string/length pair.
3fe05580 337
9dcc53ea
Z
338=for apidoc Ama|char*|savesharedpvs|const char* s
339A version of C<savepvs()> which allocates the duplicate string in memory
340which is shared between threads.
341
3fe05580
MHM
342=head1 GV Functions
343
344=for apidoc Am|HV*|gv_stashpvs|const char* name|I32 create
345Like C<gv_stashpvn>, but takes a literal string instead of a string/length pair.
346
347=head1 Hash Manipulation Functions
348
349=for apidoc Am|SV**|hv_fetchs|HV* tb|const char* key|I32 lval
350Like C<hv_fetch>, but takes a literal string instead of a string/length pair.
351
352=for apidoc Am|SV**|hv_stores|HV* tb|const char* key|NULLOK SV* val
353Like C<hv_store>, but takes a literal string instead of a string/length pair
354and omits the hash parameter.
355
510966aa
Z
356=head1 Lexer interface
357
358=for apidoc Amx|void|lex_stuff_pvs|const char *pv|U32 flags
359
360Like L</lex_stuff_pvn>, but takes a literal string instead of a
361string/length pair.
362
3fe05580
MHM
363=cut
364*/
365
926bb54c
KW
366/* concatenating with "" ensures that only literal strings are accepted as
367 * argument */
26bfd924 368#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
2efa8cc7 369
926bb54c
KW
370/* note that STR_WITH_LEN() can't be used as argument to macros or functions
371 * that under some configurations might be macros, which means that it requires
372 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
ba3a79e7
GA
373 */
374
375/* STR_WITH_LEN() shortcuts */
376#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
377#define newSVpvs_flags(str,flags) \
378 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 379#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
380#define sv_catpvs_flags(sv, str, flags) \
381 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
382#define sv_catpvs_nomg(sv, str) \
383 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
384#define sv_catpvs(sv, str) \
385 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
386#define sv_catpvs_mg(sv, str) \
387 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 388#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
389#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
390#define sv_setref_pvs(rv, classname, str) \
391 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 392#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
393#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
394#define gv_stashpvs(str, create) \
395 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
396#define gv_fetchpvs(namebeg, add, sv_type) \
397 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
398#define gv_fetchpvn(namebeg, len, add, sv_type) \
399 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
400#define sv_catxmlpvs(dsv, str, utf8) \
401 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
a038e571
NC
402#define hv_fetchs(hv,key,lval) \
403 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
404 (lval) ? (HV_FETCH_JUST_SV | HV_FETCH_LVALUE) \
405 : HV_FETCH_JUST_SV, NULL, 0))
406
407#define hv_stores(hv,key,val) \
408 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
409 (HV_FETCH_ISSTORE|HV_FETCH_JUST_SV), (val), 0))
ba3a79e7 410
510966aa
Z
411#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
412
b96d8cd9
NC
413#define get_cvs(str, flags) \
414 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 415
954c1994 416/*
ccfc67b7
JH
417=head1 Miscellaneous Functions
418
954c1994
GS
419=for apidoc Am|bool|strNE|char* s1|char* s2
420Test two strings to see if they are different. Returns true or
421false.
422
423=for apidoc Am|bool|strEQ|char* s1|char* s2
424Test two strings to see if they are equal. Returns true or false.
425
426=for apidoc Am|bool|strLT|char* s1|char* s2
427Test two strings to see if the first, C<s1>, is less than the second,
428C<s2>. Returns true or false.
429
430=for apidoc Am|bool|strLE|char* s1|char* s2
431Test two strings to see if the first, C<s1>, is less than or equal to the
432second, C<s2>. Returns true or false.
433
434=for apidoc Am|bool|strGT|char* s1|char* s2
435Test two strings to see if the first, C<s1>, is greater than the second,
436C<s2>. Returns true or false.
437
438=for apidoc Am|bool|strGE|char* s1|char* s2
439Test two strings to see if the first, C<s1>, is greater than or equal to
440the second, C<s2>. Returns true or false.
441
442=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
443Test two strings to see if they are different. The C<len> parameter
72d33970 444indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
445wrapper for C<strncmp>).
446
447=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
448Test two strings to see if they are equal. The C<len> parameter indicates
72d33970 449the number of bytes to compare. Returns true or false. (A wrapper for
954c1994
GS
450C<strncmp>).
451
452=cut
453*/
454
8d063cd8
LW
455#define strNE(s1,s2) (strcmp(s1,s2))
456#define strEQ(s1,s2) (!strcmp(s1,s2))
457#define strLT(s1,s2) (strcmp(s1,s2) < 0)
458#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
459#define strGT(s1,s2) (strcmp(s1,s2) > 0)
460#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
461#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
462#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
378cc40b 463
36477c24
PP
464#ifdef HAS_MEMCMP
465# define memNE(s1,s2,l) (memcmp(s1,s2,l))
466# define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
467#else
468# define memNE(s1,s2,l) (bcmp(s1,s2,l))
469# define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
470#endif
471
568a785a 472#define memEQs(s1, l, s2) \
8e6497b6 473 (sizeof(s2)-1 == l && memEQ(s1, ("" s2 ""), (sizeof(s2)-1)))
568a785a
NC
474#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
475
bbce6d69
PP
476/*
477 * Character classes.
478 *
479 * Unfortunately, the introduction of locales means that we
480 * can't trust isupper(), etc. to tell the truth. And when
481 * it comes to /\w+/ with tainting enabled, we *must* be able
482 * to trust our character classes.
483 *
484 * Therefore, the default tests in the text of Perl will be
485 * independent of locale. Any code that wants to depend on
486 * the current locale will use the tests that begin with "lc".
487 */
488
2304df62
AD
489#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
490# ifndef CTYPE256
491# define CTYPE256
492# endif
493#endif
494
954c1994 495/*
ccfc67b7 496
dcccc8ff 497=head1 Character classification
243effed
KW
498This section is about functions (really macros) that classify characters
499into types, such as punctuation versus alphabetic, etc. Most of these are
500analogous to regular expression character classes. (See
501L<perlrecharclass/POSIX Character Classes>.) There are several variants for
502each class. (Not all macros have all variants; each item below lists the
503ones valid for it.) None are affected by C<use bytes>, and only the ones
504with C<LC> in the name are affected by the current locale.
505
506The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
507C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
508represented by that octet is (or on non-ASCII platforms, corresponds to) an
509ASCII character in the named class based on platform, Unicode, and Perl rules.
510If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 511
6aff1f14
KW
512Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
513with no suffix C<"_A">.
4b9734bf
KW
514
515Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
516onto the platform. That is, the code points that are ASCII are unaffected,
517since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
518as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
519true when called with the code point 0xDF, which is a word character in both
4650c663 520ASCII and EBCDIC (though it represents different characters in each).
243effed
KW
521
522Variant C<isFOO_uni> is like the C<isFOO_L1> variant, but accepts any UV code
523point as input. If the code point is larger than 255, Unicode rules are used
524to determine if it is in the character class. For example,
6aff1f14
KW
525C<isWORDCHAR_uni(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
526WITH MACRON in Unicode, and is a word character.
243effed
KW
527
528Variant C<isFOO_utf8> is like C<isFOO_uni>, but the input is a pointer to a
529(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
6aff1f14
KW
530classification of just the first (possibly multi-byte) character in the string
531is tested.
243effed 532
31f05a37
KW
533Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
534result is based on the current locale, which is what C<LC> in the name stands
535for. If Perl can determine that the current locale is a UTF-8 locale, it uses
536the published Unicode rules; otherwise, it uses the C library function that
537gives the named classification. For example, C<isDIGIT_LC()> when not in a
538UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
539returned if the input won't fit into an octet. On some platforms where the C
540library function is known to be defective, Perl changes its result to follow
541the POSIX standard's rules.
243effed
KW
542
543Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
544returns the same as C<isFOO_LC> for input code points less than 256, and
545returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
546
547Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer to a
548(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
6aff1f14
KW
549classification of just the first (possibly multi-byte) character in the string
550is tested.
ccfc67b7 551
954c1994 552=for apidoc Am|bool|isALPHA|char ch
2787a470 553Returns a boolean indicating whether the specified character is an
6aff1f14 554alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
555See the L<top of this section|/Character classification> for an explanation of
556variants
e5ad6aba 557C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uni>, C<isALPHA_utf8>, C<isALPHA_LC>,
243effed 558C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
8a58bdcf 559
15861f94
KW
560=for apidoc Am|bool|isALPHANUMERIC|char ch
561Returns a boolean indicating whether the specified character is a either an
6aff1f14 562alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
563See the L<top of this section|/Character classification> for an explanation of
564variants
15861f94 565C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uni>,
e5ad6aba 566C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
15861f94
KW
567C<isALPHANUMERIC_LC_utf8>.
568
8a58bdcf
KW
569=for apidoc Am|bool|isASCII|char ch
570Returns a boolean indicating whether the specified character is one of the 128
243effed 571characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 572On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
573character corresponds to an ASCII character. Variants C<isASCII_A()> and
574C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
575See the L<top of this section|/Character classification> for an explanation of
576variants
243effed
KW
577C<isASCII_uni>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
578C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
579library routine C<isascii()>. In these cases, the variants whose names contain
580C<LC> are the same as the corresponding ones without.
581
d98532ea
KW
582Also note, that because all ASCII characters are UTF-8 invariant (meaning they
583have the exact same representation (always a single byte) whether encoded in
584UTF-8 or not), C<isASCII> will give the correct results when called with any
585byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
586work properly on any string encoded or not in UTF-8.
587
243effed
KW
588=for apidoc Am|bool|isBLANK|char ch
589Returns a boolean indicating whether the specified character is a
6aff1f14 590character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
591See the L<top of this section|/Character classification> for an explanation of
592variants
e5ad6aba 593C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uni>, C<isBLANK_utf8>, C<isBLANK_LC>,
243effed
KW
594C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
595platforms do not have the C library routine C<isblank()>. In these cases, the
596variants whose names contain C<LC> are the same as the corresponding ones
597without.
598
599=for apidoc Am|bool|isCNTRL|char ch
600Returns a boolean indicating whether the specified character is a
6aff1f14 601control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
602See the L<top of this section|/Character classification> for an explanation of
603variants
e5ad6aba 604C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uni>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
6aff1f14
KW
605C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
606On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 607
954c1994 608=for apidoc Am|bool|isDIGIT|char ch
2787a470 609Returns a boolean indicating whether the specified character is a
6aff1f14 610digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 611Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
612See the L<top of this section|/Character classification> for an explanation of
613variants
e5ad6aba 614C<isDIGIT_uni>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
243effed
KW
615C<isDIGIT_LC_utf8>.
616
617=for apidoc Am|bool|isGRAPH|char ch
618Returns a boolean indicating whether the specified character is a
6aff1f14 619graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff
KW
620See the L<top of this section|/Character classification> for an explanation of
621variants
e5ad6aba 622C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uni>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
243effed 623C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
954c1994 624
0c82b6df 625=for apidoc Am|bool|isLOWER|char ch
2787a470 626Returns a boolean indicating whether the specified character is a
6aff1f14 627lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
628See the L<top of this section|/Character classification> for an explanation of
629variants
e5ad6aba 630C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uni>, C<isLOWER_utf8>, C<isLOWER_LC>,
243effed 631C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
0c82b6df 632
c99e91e9 633=for apidoc Am|bool|isOCTAL|char ch
2787a470 634Returns a boolean indicating whether the specified character is an
6aff1f14 635octal digit, [0-7].
243effed
KW
636The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
637C<isOCTAL>.
638
639=for apidoc Am|bool|isPUNCT|char ch
640Returns a boolean indicating whether the specified character is a
6aff1f14
KW
641punctuation character, analogous to C<m/[[:punct:]]/>.
642Note that the definition of what is punctuation isn't as
243effed
KW
643straightforward as one might desire. See L<perlrecharclass/POSIX Character
644Classes> for details.
dcccc8ff
KW
645See the L<top of this section|/Character classification> for an explanation of
646variants
e5ad6aba 647C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uni>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
243effed 648C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
c99e91e9 649
0c82b6df 650=for apidoc Am|bool|isSPACE|char ch
2787a470 651Returns a boolean indicating whether the specified character is a
6aff1f14 652whitespace character. This is analogous
398d098a
KW
653to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
654(experimentally), this also matches what C<m/[[:space:]]/> does.
a04e6aad 655("Experimentally" means that this change may be backed out in 5.22 if
398d098a
KW
656field experience indicates that it was unwise.) Prior to 5.18, only the
657locale forms of this macro (the ones with C<LC> in their names) matched
658precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
659in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
660(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
661See the L<top of this section|/Character classification> for an explanation of
662variants
e5ad6aba 663C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uni>, C<isSPACE_utf8>, C<isSPACE_LC>,
243effed 664C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
0c82b6df 665
398d098a
KW
666=for apidoc Am|bool|isPSXSPC|char ch
667(short for Posix Space)
668Starting in 5.18, this is identical (experimentally) in all its forms to the
669corresponding C<isSPACE()> macros. ("Experimentally" means that this change
a04e6aad 670may be backed out in 5.22 if field experience indicates that it
398d098a
KW
671was unwise.)
672The locale forms of this macro are identical to their corresponding
673C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
674non-locale forms differ from their C<isSPACE()> forms only in that the
675C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
676Otherwise they are identical. Thus this macro is analogous to what
677C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff
KW
678See the L<top of this section|/Character classification> for an explanation of
679variants
e5ad6aba 680C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uni>, C<isPSXSPC_utf8>, C<isPSXSPC_LC>,
398d098a
KW
681C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
682
954c1994 683=for apidoc Am|bool|isUPPER|char ch
2787a470 684Returns a boolean indicating whether the specified character is an
6aff1f14 685uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff
KW
686See the L<top of this section|/Character classification> for an explanation of
687variants
e5ad6aba 688C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uni>, C<isUPPER_utf8>, C<isUPPER_LC>,
243effed 689C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
954c1994 690
243effed 691=for apidoc Am|bool|isPRINT|char ch
8eea39dd 692Returns a boolean indicating whether the specified character is a
6aff1f14 693printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
694See the L<top of this section|/Character classification> for an explanation of
695variants
e5ad6aba 696C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uni>, C<isPRINT_utf8>, C<isPRINT_LC>,
243effed
KW
697C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
698
699=for apidoc Am|bool|isWORDCHAR|char ch
700Returns a boolean indicating whether the specified character is a character
701that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
702in a regular expression. A word character is an alphabetic character, a
703decimal digit, a connecting punctuation character (such as an underscore), or
704a "mark" character that attaches to one of those (like some sort of accent).
705C<isALNUM()> is a synonym provided for backward compatibility, even though a
706word character includes more than the standard C language meaning of
707alphanumeric.
dcccc8ff
KW
708See the L<top of this section|/Character classification> for an explanation of
709variants
1a83413c
KW
710C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uni>, and C<isWORDCHAR_utf8>.
711C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
712described there, but additionally include the platform's native underscore.
8a58bdcf
KW
713
714=for apidoc Am|bool|isXDIGIT|char ch
715Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
716digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
717and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
718See the L<top of this section|/Character classification> for an explanation of
719variants
243effed
KW
720C<isXDIGIT_uni>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
721C<isXDIGIT_LC_utf8>.
722
3c3ecf18
KW
723=for apidoc Am|bool|isIDFIRST|char ch
724Returns a boolean indicating whether the specified character can be the first
725character of an identifier. This is very close to, but not quite the same as
726the official Unicode property C<XID_Start>. The difference is that this
727returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
728See the L<top of this section|/Character classification> for an explanation of
729variants
3c3ecf18
KW
730C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uni>, C<isIDFIRST_utf8>,
731C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
732
733=for apidoc Am|bool|isIDCONT|char ch
734Returns a boolean indicating whether the specified character can be the
735second or succeeding character of an identifier. This is very close to, but
736not quite the same as the official Unicode property C<XID_Continue>. The
737difference is that this returns true only if the input character also matches
dcccc8ff
KW
738L</isWORDCHAR>. See the L<top of this section|/Character classification> for
739an
3c3ecf18
KW
740explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uni>,
741C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
742C<isIDCONT_LC_utf8>.
743
243effed 744=head1 Miscellaneous Functions
8eea39dd 745
95a59cab 746=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 747Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
748Behaviour is only well defined when isXDIGIT(*str) is true.
749
e7c1e6c1
KW
750=head1 Character case changing
751
1f607577
KW
752=for apidoc Am|U8|toUPPER|U8 ch
753Converts the specified character to uppercase. If the input is anything but an
754ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 755C<toUPPER_A> is equivalent.
954c1994 756
1f607577
KW
757=for apidoc Am|UV|toUPPER_uni|UV cp|U8* s|STRLEN* lenp
758Converts the Unicode code point C<cp> to its uppercase version, and
759stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
760that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
761bytes since the uppercase version may be longer than the original character.
762
763The first code point of the uppercased version is returned
764(but note, as explained just above, that there may be more.)
765
766=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
767Converts the UTF-8 encoded character at C<p> to its uppercase version, and
768stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
769that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
770bytes since the uppercase version may be longer than the original character.
771
772The first code point of the uppercased version is returned
773(but note, as explained just above, that there may be more.)
774
775The input character at C<p> is assumed to be well-formed.
776
25200305
KW
777=for apidoc Am|U8|toFOLD|U8 ch
778Converts the specified character to foldcase. If the input is anything but an
779ASCII uppercase character, that input character itself is returned. Variant
780C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
781Latin1 range, as the full generality of L</toFOLD_uni> is needed there.)
782
1f607577
KW
783=for apidoc Am|UV|toFOLD_uni|UV cp|U8* s|STRLEN* lenp
784Converts the Unicode code point C<cp> to its foldcase version, and
785stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
786that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
787bytes since the foldcase version may be longer than the original character.
788
789The first code point of the foldcased version is returned
790(but note, as explained just above, that there may be more.)
791
792=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
793Converts the UTF-8 encoded character at C<p> to its foldcase version, and
794stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
795that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
796bytes since the foldcase version may be longer than the original character.
797
798The first code point of the foldcased version is returned
799(but note, as explained just above, that there may be more.)
800
801The input character at C<p> is assumed to be well-formed.
802
803=for apidoc Am|U8|toLOWER|U8 ch
804Converts the specified character to lowercase. If the input is anything but an
805ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 806C<toLOWER_A> is equivalent.
954c1994 807
1f607577
KW
808=for apidoc Am|U8|toLOWER_L1|U8 ch
809Converts the specified Latin1 character to lowercase. The results are undefined if
810the input doesn't fit in a byte.
811
812=for apidoc Am|U8|toLOWER_LC|U8 ch
813Converts the specified character to lowercase using the current locale's rules,
814if possible; otherwise returns the input character itself.
815
816=for apidoc Am|UV|toLOWER_uni|UV cp|U8* s|STRLEN* lenp
817Converts the Unicode code point C<cp> to its lowercase version, and
818stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
819that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
820bytes since the lowercase version may be longer than the original character.
821
822The first code point of the lowercased version is returned
823(but note, as explained just above, that there may be more.)
824
825=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
826Converts the UTF-8 encoded character at C<p> to its lowercase version, and
827stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
828that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
829bytes since the lowercase version may be longer than the original character.
830
831The first code point of the lowercased version is returned
832(but note, as explained just above, that there may be more.)
833
834The input character at C<p> is assumed to be well-formed.
835
25200305
KW
836=for apidoc Am|U8|toTITLE|U8 ch
837Converts the specified character to titlecase. If the input is anything but an
838ASCII lowercase character, that input character itself is returned. Variant
839C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1 range,
840as the full generality of L</toTITLE_uni> is needed there. Titlecase is not a
841concept used in locale handling, so there is no functionality for that.)
842
1f607577
KW
843=for apidoc Am|UV|toTITLE_uni|UV cp|U8* s|STRLEN* lenp
844Converts the Unicode code point C<cp> to its titlecase version, and
845stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
846that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
847bytes since the titlecase version may be longer than the original character.
848
849The first code point of the titlecased version is returned
850(but note, as explained just above, that there may be more.)
851
852=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
853Converts the UTF-8 encoded character at C<p> to its titlecase version, and
854stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
855that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
856bytes since the titlecase version may be longer than the original character.
857
858The first code point of the titlecased version is returned
859(but note, as explained just above, that there may be more.)
860
861The input character at C<p> is assumed to be well-formed.
862
954c1994 863=cut
353c9b6f 864
1f607577 865XXX Still undocumented isVERTWS_uni and _utf8; it's unclear what their names
25200305 866really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 867
8a58bdcf 868Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
869patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
870
954c1994
GS
871*/
872
7c062697
KW
873/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
874 * is known only in the perl core, and this macro can be called from outside
875 * that */
876#ifdef HAS_QUAD
877# define WIDEST_UTYPE U64TYPE
878#else
879# define WIDEST_UTYPE U32
880#endif
881
3912bc88
KW
882/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
883 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
884 * bits of information are lost even on a 64-bit machine, but to get the
885 * compiler to optimize it out if possible. This is because Configure makes
886 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
887 * sizeof() guarantees that this evaluates to a constant true at compile time.
220c71bf
KW
888 */
889#define FITS_IN_8_BITS(c) ((sizeof(c) == 1) || !(((WIDEST_UTYPE)(c)) & ~0xFF))
cf301eb7 890
41f43cc2 891#ifdef EBCDIC
b6340bd0 892# ifndef _ALL_SOURCE
0852beac
KW
893 /* The native libc isascii() et.al. functions return the wrong results
894 * on at least z/OS unless this is defined. */
b6340bd0
KW
895# error _ALL_SOURCE should probably be defined
896# endif
41f43cc2 897#else
0852beac
KW
898 /* There is a simple definition of ASCII for ASCII platforms. But the
899 * EBCDIC one isn't so simple, so is defined using table look-up like the
900 * other macros below */
41f43cc2
KW
901# define isASCII(c) ((WIDEST_UTYPE)(c) < 128)
902#endif
903
c2da0b36
KW
904/* The lower 3 bits in both the ASCII and EBCDIC representations of '0' are 0,
905 * and the 8 possible permutations of those bits exactly comprise the 8 octal
906 * digits */
3d33b3e3 907#define isOCTAL_A(c) cBOOL(FITS_IN_8_BITS(c) && (0xF8 & (c)) == '0')
c2da0b36 908
9fb1bf9d 909#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 910
a500dc72
KW
911/* Character class numbers. For internal core Perl use only. The ones less
912 * than 32 are used in PL_charclass[] and the ones up through the one that
913 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
914 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
915 * their actual definitions are here. If that file has a name not used here,
916 * it won't compile.
1709d539
KW
917 *
918 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
919 * frequency of their use. This gives a slight edge to exiting a loop earlier
920 * (in reginclass() in regexec.c) */
1709d539
KW
921# define _CC_WORDCHAR 0 /* \w and [:word:] */
922# define _CC_DIGIT 1 /* \d and [:digit:] */
923# define _CC_ALPHA 2 /* [:alpha:] */
924# define _CC_LOWER 3 /* [:lower:] */
925# define _CC_UPPER 4 /* [:upper:] */
926# define _CC_PUNCT 5 /* [:punct:] */
927# define _CC_PRINT 6 /* [:print:] */
15861f94 928# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 929# define _CC_GRAPH 8 /* [:graph:] */
b0d691b2 930# define _CC_CASED 9 /* [:lower:] and [:upper:] under /i */
1709d539 931
b0d691b2 932#define _FIRST_NON_SWASH_CC 10
1709d539
KW
933/* The character classes above are implemented with swashes. The second group
934 * (just below) contains the ones implemented without. These are also sorted
935 * in rough order of the frequency of their use, except that \v should be last,
936 * as it isn't a real Posix character class, and some (small) inefficiencies in
937 * regular expression handling would be introduced by putting it in the middle
938 * of those that are. Also, cntrl and ascii come after the others as it may be
939 * useful to group these which have no members that match above Latin1, (or
940 * above ASCII in the latter case) */
941
b0d691b2
KW
942# define _CC_SPACE 10 /* \s */
943# define _CC_BLANK 11 /* [:blank:] */
944# define _CC_XDIGIT 12 /* [:xdigit:] */
945# define _CC_PSXSPC 13 /* [:space:] */
946# define _CC_CNTRL 14 /* [:cntrl:] */
947# define _CC_ASCII 15 /* [:ascii:] */
948# define _CC_VERTSPACE 16 /* \v */
1709d539 949
a0947d7b
KW
950# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
951
1709d539 952/* The members of the third group below do not need to be coordinated with data
3ffc8c70 953 * structures in regcomp.[ch] and regexec.c. */
f12c0118
KW
954# define _CC_IDFIRST 17
955# define _CC_CHARNAME_CONT 18
956# define _CC_NONLATIN1_FOLD 19
957# define _CC_NONLATIN1_SIMPLE_FOLD 20
958# define _CC_QUOTEMETA 21
959# define _CC_NON_FINAL_FOLD 22
960# define _CC_IS_IN_SOME_FOLD 23
8e35b056
KW
961# define _CC_MNEMONIC_CNTRL 24
962/* Unused: 25-31
f4cdb42c
KW
963 * If more bits are needed, one could add a second word for non-64bit
964 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
965 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
966 * is used only for optimization (as of this writing), and differs in the
967 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72
KW
968 * characters: the masculine and feminine ordinal indicators, so removing it
969 * would just cause /i regexes which match them to run less efficiently */
96ac0975 970
3a371f2f
KW
971#if defined(PERL_CORE) || defined(PERL_EXT)
972/* An enum version of the character class numbers, to help compilers
973 * optimize */
974typedef enum {
3a371f2f 975 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
976 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
977 _CC_ENUM_ASCII = _CC_ASCII,
978 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 979 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 980 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
981 _CC_ENUM_DIGIT = _CC_DIGIT,
982 _CC_ENUM_GRAPH = _CC_GRAPH,
983 _CC_ENUM_LOWER = _CC_LOWER,
984 _CC_ENUM_PRINT = _CC_PRINT,
e8d596e0 985 _CC_ENUM_PSXSPC = _CC_PSXSPC,
3a371f2f 986 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 987 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 988 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 989 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 990 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 991 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
992} _char_class_number;
993#endif
994
63c61c3f 995#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 996#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 997
3018b823 998#if defined(PERL_IN_UTF8_C) || defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_REGEXEC_C)
922e8cb4 999# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1000 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1001 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1002 #error Need to adjust order of swash_property_names[]
1003# endif
1004
b44daeeb
KW
1005/* This is declared static in each of the few files that this is #defined for
1006 * to keep them from being publicly accessible. Hence there is a small amount
1007 * of wasted space */
1008
922e8cb4
KW
1009static const char* const swash_property_names[] = {
1010 "XPosixWord",
1011 "XPosixDigit",
1012 "XPosixAlpha",
1013 "XPosixLower",
1014 "XPosixUpper",
1015 "XPosixPunct",
1016 "XPosixPrint",
1017 "XPosixAlnum",
b0d691b2
KW
1018 "XPosixGraph",
1019 "Cased"
922e8cb4
KW
1020};
1021#endif
1022
6635f04f 1023START_EXTERN_C
96ac0975
NC
1024# ifdef DOINIT
1025EXTCONST U32 PL_charclass[] = {
1026# include "l1_char_class_tab.h"
1027};
1028
1029# else /* ! DOINIT */
1030EXTCONST U32 PL_charclass[];
1031# endif
6635f04f 1032END_EXTERN_C
96ac0975 1033
265c1f46 1034 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1035# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1036
1037 /* For internal core Perl use only: the base macro for defining macros like
1038 * isALPHA */
ff7ecfc3 1039# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1040 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1041
f4cdb42c
KW
1042 /* The mask for the _A versions of the macros; it just adds in the bit for
1043 * ASCII. */
1044# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1045
4650c663
KW
1046 /* For internal core Perl use only: the base macro for defining macros like
1047 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1048 * the ASCII bit are present */
3ded5eb0 1049# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
f4cd282c 1050 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
f4cdb42c
KW
1051 == _CC_mask_A(classnum)))
1052
f4cdb42c 1053# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1054# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1055# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1056# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1057# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1058# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1059# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1060# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1061# define isPSXSPC_A(c) _generic_isCC_A(c, _CC_PSXSPC)
1062# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1063# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1064# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1065# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
96ca48da 1066# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits */
d95f8b6a 1067# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1068# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1069# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1070# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1071
1072 /* continuation character for legal NAME in \N{NAME} */
1073# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1074
1075# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1076# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1077# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1078# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1079# define isPSXSPC_L1(c) _generic_isCC(c, _CC_PSXSPC)
1080# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1081# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1082# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1083# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1084# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1085
0852beac
KW
1086# ifdef EBCDIC
1087# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1088# endif
1089
f12c0118
KW
1090 /* Participates in a single-character fold with a character above 255 */
1091# 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)))
1092
1093 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1094# 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 1095
4eeeb416 1096# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd
KW
1097# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1098 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926
KW
1099# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1100 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1101# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1102 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1103#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1104
1105 /* If we don't have perl.h, we are compiling a utility program. Below we
1106 * hard-code various macro definitions that wouldn't otherwise be available
0852beac
KW
1107 * to it. Most are coded based on first principals. First some ones common
1108 * to both ASCII and EBCDIC */
1109# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1110# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1111# define isSPACE_A(c) (isBLANK_A(c) \
1112 || (c) == '\n' \
1113 || (c) == '\r' \
1114 || (c) == '\v' \
1115 || (c) == '\f')
1116# ifdef EBCDIC /* There are gaps between 'i' and 'j'; 'r' and 's'. Same
1117 for uppercase. This is ordered to exclude most things
1118 early */
1119# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1120 && ((c) <= 'i' \
1121 || ((c) >= 'j' && (c) <= 'r') \
1122 || (c) >= 's'))
1123# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1124 && ((c) <= 'I' \
1125 || ((c) >= 'J' && (c) <= 'R') \
1126 || (c) >= 'S'))
1127# else /* ASCII platform. */
9fb1bf9d 1128# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z')
2181da29 1129# define isUPPER_A(c) ((c) <= 'Z' && (c) >= 'A')
0852beac
KW
1130# endif
1131
1132 /* Some more ASCII, non-ASCII common definitions */
a4d7a999
KW
1133# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1134# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1135# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac
KW
1136# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1137# define isXDIGIT_A(c) (isDIGIT_A(c) \
1138 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1139 || ((c) <= 'F' && (c) >= 'A'))
3ded5eb0 1140
0852beac
KW
1141# ifdef EBCDIC
1142# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1143 || (c) == '#' || (c) == '$' || (c) == '%' \
1144 || (c) == '&' || (c) == '\'' || (c) == '(' \
1145 || (c) == ')' || (c) == '*' || (c) == '+' \
1146 || (c) == ',' || (c) == '.' || (c) == '/' \
1147 || (c) == ':' || (c) == ';' || (c) == '<' \
1148 || (c) == '=' || (c) == '>' || (c) == '?' \
1149 || (c) == '@' || (c) == '[' || (c) == '\\' \
1150 || (c) == ']' || (c) == '^' || (c) == '_' \
1151 || (c) == '`' || (c) == '{' || (c) == '|' \
1152 || (c) == '}' || (c) == '~')
1153# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1154# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1155
1156# ifdef QUESTION_MARK_CTRL
1157# define _isQMC(c) ((c) == QUESTION_MARK_CTRL)
1158# else
1159# define _isQMC(c) 0
1160# endif
1161
1162 /* I (khw) can't think of a way to define all the ASCII controls
1163 * without resorting to a libc (locale-sensitive) call. But we know
1164 * that all controls but the question-mark one are in the range 0-0x3f.
1165 * This makes sure that all the controls that have names are included,
1166 * and all controls that are also considered ASCII in the locale. This
1167 * may include more or fewer than what it actually should, but the
1168 * wrong ones are less-important controls, so likely won't impact
1169 * things (keep in mind that this is compiled only if perl.h isn't
1170 * available). The question mark control is included if available */
1171# define isCNTRL_A(c) (((c) < 0x40 && isascii(c)) \
1172 || (c) == '\0' || (c) == '\a' || (c) == '\b' \
1173 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1174 || (c) == '\t' || (c) == '\v' || _isQMC(c))
1175
1176# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1177# else /* ASCII platform; things are simpler, and isASCII has already
1178 been defined */
1179# define isGRAPH_A(c) (((c) > ' ' && (c) < 127))
1180# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1181# define isPUNCT_A(c) (isGRAPH_A(c) && (! isALPHANUMERIC_A(c)))
1182# define isCNTRL_A(c) (isASCII(c) && (! isPRINT_A(c)))
1183# endif
1184
3ffc8c70 1185 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1186 * during debugging, and it seems best to keep them. We may be called
1187 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1188 * do anything anyway, so make it not a problem */
1189# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1190# define NATIVE_TO_LATIN1(ch) (ch)
1191# endif
3ded5eb0
KW
1192# define isPSXSPC_A(c) isSPACE_A(c) /* XXX Assumes SPACE matches '\v' */
1193# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1194# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1195# define isBLANK_L1(c) (isBLANK_A(c) \
1196 || (FITS_IN_8_BITS(c) \
1197 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1198# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1199# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1200# define isLOWER_L1(c) (isLOWER_A(c) \
1201 || (FITS_IN_8_BITS(c) \
1202 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1203 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1204 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1205 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1206 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1207# define isPRINT_L1(c) (isPRINT_A(c) \
1208 || (FITS_IN_8_BITS(c) \
1209 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1210# define isPSXSPC_L1(c) isSPACE_L1(c)
1211# define isPUNCT_L1(c) (isPUNCT_A(c) \
1212 || (FITS_IN_8_BITS(c) \
1213 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1214 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1215 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1216 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1217 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1218 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1219 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1220# define isSPACE_L1(c) (isSPACE_A(c) \
1221 || (FITS_IN_8_BITS(c) \
1222 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1223 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1224# define isUPPER_L1(c) (isUPPER_A(c) \
1225 || (FITS_IN_8_BITS(c) \
1226 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1227 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1228 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1229# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1230# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1231# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1232 || isBLANK_L1(c) \
1233 || (c) == '-' \
1234 || (c) == '(' \
1235 || (c) == ')')
1236 /* The following are not fully accurate in the above-ASCII range. I (khw)
1237 * don't think it's necessary to be so for the purposes where this gets
1238 * compiled */
1239# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1240# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1241
1242 /* And these aren't accurate at all. They are useful only for above
1243 * Latin1, which utilities and bootstrapping don't deal with */
1244# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1245# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1246# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1247
1248 /* Many of the macros later in this file are defined in terms of these. By
1249 * implementing them with a function, which converts the class number into
1250 * a call to the desired macro, all of the later ones work. However, that
1251 * function won't be actually defined when building a utility program (no
1252 * perl.h), and so a compiler error will be generated if one is attempted
1253 * to be used. And the above-Latin1 code points require Unicode tables to
1254 * be present, something unlikely to be the case when bootstrapping */
1255# define _generic_isCC(c, classnum) \
1256 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1257# define _generic_isCC_A(c, classnum) \
1258 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1259#endif /* End of no perl.h H_PERL */
8a58bdcf 1260
e66b99e9
KW
1261#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1262#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1263#define isASCII_A(c) isASCII(c)
1264#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1265#define isBLANK(c) isBLANK_A(c)
1266#define isCNTRL(c) isCNTRL_A(c)
1267#define isDIGIT(c) isDIGIT_A(c)
1268#define isGRAPH(c) isGRAPH_A(c)
1269#define isIDFIRST(c) isIDFIRST_A(c)
1270#define isLOWER(c) isLOWER_A(c)
1271#define isPRINT(c) isPRINT_A(c)
1272#define isPSXSPC(c) isPSXSPC_A(c)
1273#define isPUNCT(c) isPUNCT_A(c)
1274#define isSPACE(c) isSPACE_A(c)
1275#define isUPPER(c) isUPPER_A(c)
1276#define isWORDCHAR(c) isWORDCHAR_A(c)
1277#define isXDIGIT(c) isXDIGIT_A(c)
1278
1279/* ASCII casing. These could also be written as
1280 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1281 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1282 which uses table lookup and mask instead of subtraction. (This would
1283 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1284#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1285#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1286
25200305
KW
1287/* In the ASCII range, these are equivalent to what they're here defined to be.
1288 * But by creating these definitions, other code doesn't have to be aware of
1289 * this detail */
1290#define toFOLD(c) toLOWER(c)
25200305
KW
1291#define toTITLE(c) toUPPER(c)
1292
c753c8d3
KW
1293#define toLOWER_A(c) toLOWER(c)
1294#define toUPPER_A(c) toUPPER(c)
25200305
KW
1295#define toFOLD_A(c) toFOLD(c)
1296#define toTITLE_A(c) toTITLE(c)
1a0901db 1297
4650c663 1298/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1299#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1300 ? (c) \
f4cd282c 1301 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1302#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1303
1a0901db 1304/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1305 * all mapped to one of them, and these need special handling; returns the
1306 * input itself if is out-of-range */
b2bf251f 1307#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1308 ? (c) \
f4cd282c 1309 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1310#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1311
beab9ebe
KW
1312/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1313
1314/* For internal core Perl use only: the base macro for defining macros like
1315 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1316 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1317 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1318 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1319 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1320 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1321 * code points under locale rules, use the macros based on _generic_LC_uvchr
1322 * instead of this. */
beab9ebe
KW
1323#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1324 (! FITS_IN_8_BITS(c) \
1325 ? 0 \
31f05a37
KW
1326 : IN_UTF8_CTYPE_LOCALE \
1327 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1328 : cBOOL(non_utf8))
1329
1330/* For internal core Perl use only: a helper macro for defining macros like
1331 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1332 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1333 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1334 * _generic_LC_base for more info */
1335#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1336 _generic_LC_base(c,utf8_locale_classnum, \
1337 non_utf8_func( (_LC_CAST) (c)))
1338
1339/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1340 * 'c' is the platform's native underscore character */
1341#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1342 _generic_LC_base(c, utf8_locale_classnum, \
1343 (non_utf8_func( (_LC_CAST) (c)) \
1344 || (char)(c) == '_'))
1345
1346/* These next three are also for internal core Perl use only: case-change
1347 * helper macros */
1348#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1349 ? (c) \
31f05a37
KW
1350 : (IN_UTF8_CTYPE_LOCALE) \
1351 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1352 : (cast)function((cast)(c)))
beab9ebe 1353
31f05a37
KW
1354/* Note that the result can be larger than a byte in a UTF-8 locale. It
1355 * returns a single value, so can't adequately return the upper case of LATIN
1356 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1357 * values "SS"); instead it asserts against that under DEBUGGING, and
1358 * otherwise returns its input */
beab9ebe
KW
1359#define _generic_toUPPER_LC(c, function, cast) \
1360 (! FITS_IN_8_BITS(c) \
1361 ? (c) \
31f05a37 1362 : ((! IN_UTF8_CTYPE_LOCALE) \
d2a8cf7f 1363 ? (cast)function((cast)(c)) \
31f05a37
KW
1364 : ((((U8)(c)) == MICRO_SIGN) \
1365 ? GREEK_CAPITAL_LETTER_MU \
1366 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1367 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1368 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1369 ? (__ASSERT_(0) (c)) \
1370 : PL_mod_latin1_uc[ (U8) (c) ])))))
1371
1372/* Note that the result can be larger than a byte in a UTF-8 locale. It
1373 * returns a single value, so can't adequately return the fold case of LATIN
1374 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1375 * values "ss"); instead it asserts against that under DEBUGGING, and
1376 * otherwise returns its input */
beab9ebe 1377#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1378 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1379 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1380 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1381 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1382 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1383
84061b6a 1384/* Use the libc versions for these if available. */
f05550c0 1385#if defined(HAS_ISASCII)
84061b6a
KW
1386# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1387#else
1388# define isASCII_LC(c) isASCII(c)
1389#endif
1390
f05550c0 1391#if defined(HAS_ISBLANK)
84061b6a 1392# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1393#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1394# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1395#endif
1396
f05550c0 1397#define _LC_CAST U8
bbce6d69 1398
f05550c0 1399#ifdef WIN32
375f5f06
KW
1400 /* The Windows functions don't bother to follow the POSIX standard, which
1401 * for example says that something can't both be a printable and a control.
1402 * But Windows treats the \t control as a printable, and does such things
1403 * as making superscripts into both digits and punctuation. This tames
1404 * these flaws by assuming that the definitions of both controls and space
1405 * are correct, and then making sure that other definitions don't have
1406 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1407 * Not all possible weirdnesses are checked for, just the ones that were
1408 * detected on actual Microsoft code pages */
1409
f05550c0
BF
1410# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1411# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1412
1413# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) && isALPHANUMERIC_LC(c))
1414# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && ! isPUNCT_LC(c))
1415# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && isALPHANUMERIC_LC(c))
1416# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1417# define isIDFIRST_LC(c) (((c) == '_') || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1418# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1419# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1420# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1421# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1422# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1423# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) && isALPHANUMERIC_LC(c))
1424
1425# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1426# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1427# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1428
1429#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1430 /* For most other platforms */
beab9ebe 1431
f05550c0
BF
1432# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1433# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1434# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1435# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1436# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1437# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1438# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1439# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1440# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1441# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1442# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1443# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1444# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1445
1446
1447# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1448# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1449# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1450
1451#else /* The final fallback position */
1452
1453# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1454# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1455# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1456# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1457# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1458# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1459# define isLOWER_LC(c) (isascii(c) && islower(c))
1460# define isPRINT_LC(c) (isascii(c) && isprint(c))
1461# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1462# define isSPACE_LC(c) (isascii(c) && isspace(c))
1463# define isUPPER_LC(c) (isascii(c) && isupper(c))
1464# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1465# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1466
1467# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1468# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1469# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1470
f05550c0 1471#endif
55204971 1472
eba68aa0
KW
1473#define isIDCONT(c) isWORDCHAR(c)
1474#define isIDCONT_A(c) isWORDCHAR_A(c)
1475#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1476#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1477#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1478
4650c663
KW
1479/* For internal core Perl use only: the base macros for defining macros like
1480 * isALPHA_uni. 'c' is the code point to check. 'classnum' is the POSIX class
1481 * number defined earlier in this file. _generic_uni() is used for POSIX
1482 * classes where there is a macro or function 'above_latin1' that takes the
1483 * single argument 'c' and returns the desired value. These exist for those
1484 * classes which have simple definitions, avoiding the overhead of a hash
1485 * lookup or inversion list binary search. _generic_swash_uni() can be used
1486 * for classes where that overhead is faster than a direct lookup.
1487 * _generic_uni() won't compile if 'c' isn't unsigned, as it won't match the
1488 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1489 * have duplicate checks here, so could create versions of the macros that
1490 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1491
1492/* Note that all ignore 'use bytes' */
cd500f2f
KW
1493#define _generic_uni(classnum, above_latin1, c) ((c) < 256 \
1494 ? _generic_isCC(c, classnum) \
1495 : above_latin1(c))
1496#define _generic_swash_uni(classnum, c) ((c) < 256 \
1497 ? _generic_isCC(c, classnum) \
922e8cb4 1498 : _is_uni_FOO(classnum, c))
cd500f2f
KW
1499#define isALPHA_uni(c) _generic_swash_uni(_CC_ALPHA, c)
1500#define isALPHANUMERIC_uni(c) _generic_swash_uni(_CC_ALPHANUMERIC, c)
509fb054
KW
1501#define isASCII_uni(c) isASCII(c)
1502#define isBLANK_uni(c) _generic_uni(_CC_BLANK, is_HORIZWS_cp_high, c)
1503#define isCNTRL_uni(c) isCNTRL_L1(c) /* All controls are in Latin1 */
cd500f2f
KW
1504#define isDIGIT_uni(c) _generic_swash_uni(_CC_DIGIT, c)
1505#define isGRAPH_uni(c) _generic_swash_uni(_CC_GRAPH, c)
eba68aa0 1506#define isIDCONT_uni(c) _generic_uni(_CC_WORDCHAR, _is_uni_perl_idcont, c)
509fb054 1507#define isIDFIRST_uni(c) _generic_uni(_CC_IDFIRST, _is_uni_perl_idstart, c)
cd500f2f
KW
1508#define isLOWER_uni(c) _generic_swash_uni(_CC_LOWER, c)
1509#define isPRINT_uni(c) _generic_swash_uni(_CC_PRINT, c)
aea0c08c 1510
509fb054
KW
1511/* Posix and regular space are identical above Latin1 */
1512#define isPSXSPC_uni(c) _generic_uni(_CC_PSXSPC, is_XPERLSPACE_cp_high, c)
b18192e9 1513
cd500f2f 1514#define isPUNCT_uni(c) _generic_swash_uni(_CC_PUNCT, c)
509fb054 1515#define isSPACE_uni(c) _generic_uni(_CC_SPACE, is_XPERLSPACE_cp_high, c)
cd500f2f 1516#define isUPPER_uni(c) _generic_swash_uni(_CC_UPPER, c)
509fb054 1517#define isVERTWS_uni(c) _generic_uni(_CC_VERTSPACE, is_VERTWS_cp_high, c)
cd500f2f 1518#define isWORDCHAR_uni(c) _generic_swash_uni(_CC_WORDCHAR, c)
509fb054 1519#define isXDIGIT_uni(c) _generic_uni(_CC_XDIGIT, is_XDIGIT_cp_high, c)
b18192e9 1520
b8d68ded 1521#define toFOLD_uni(c,s,l) to_uni_fold(c,s,l)
509fb054
KW
1522#define toLOWER_uni(c,s,l) to_uni_lower(c,s,l)
1523#define toTITLE_uni(c,s,l) to_uni_title(c,s,l)
1524#define toUPPER_uni(c,s,l) to_uni_upper(c,s,l)
a0ed51b3 1525
4650c663
KW
1526/* For internal core Perl use only: the base macros for defining macros like
1527 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1528 * point, not just 0-255. Like _generic_uni, there are two versions, one for
1529 * simple class definitions; the other for more complex. These are like
1530 * _generic_uni, so see it for more info. */
cd500f2f
KW
1531#define _generic_LC_uvchr(latin1, above_latin1, c) \
1532 (c < 256 ? latin1(c) : above_latin1(c))
1533#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1534 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1535
1536#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1537#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1538 _CC_ALPHANUMERIC, c)
509fb054 1539#define isASCII_LC_uvchr(c) isASCII_LC(c)
cd500f2f 1540#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, is_HORIZWS_cp_high, c)
feeab5a9 1541#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1542#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1543#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
eba68aa0
KW
1544#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
1545 _is_uni_perl_idcont, c)
cd500f2f
KW
1546#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
1547 _is_uni_perl_idstart, c)
1548#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1549#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
509fb054
KW
1550#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c) /* space is identical to posix
1551 space under locale */
cd500f2f
KW
1552#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
1553#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1554 is_XPERLSPACE_cp_high, c)
cd500f2f
KW
1555#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
1556#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
1557 _CC_WORDCHAR, c)
1558#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, is_XDIGIT_cp_high, c)
e712593e 1559
36b00f93 1560#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1561
4650c663
KW
1562/* For internal core Perl use only: the base macros for defining macros like
1563 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1564 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1565 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1566 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1567 * same representation whether utf8 or not. Note that it assumes that the utf8
1568 * has been validated, and ignores 'use bytes' */
cd500f2f 1569#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1570 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1571 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1572 ? _generic_isCC( \
94bb8c36 1573 TWO_BYTE_UTF8_TO_NATIVE(*(p), \
aea0c08c 1574 *((p)+1 )), \
94bb8c36 1575 classnum) \
252810d0 1576 : utf8)
cd500f2f
KW
1577/* Like the above, but calls 'above_latin1(p)' to get the utf8 value. 'above_latin1'
1578 * can be a macro */
1579#define _generic_func_utf8(classnum, above_latin1, p) \
1580 _generic_utf8(classnum, p, above_latin1(p))
4650c663 1581/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1582 * 'above_latin1' parameter */
922e8cb4 1583#define _generic_swash_utf8(classnum, p) \
cd500f2f 1584 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
922e8cb4 1585
cc8ab7c0 1586/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1587 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1588 * class is TRUE for. Hence it can skip the tests for this range.
1589 * 'above_latin1' should include its arguments */
3ded5eb0 1590#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1591 (UTF8_IS_INVARIANT(*(p)) \
1592 ? _generic_isCC(*(p), classnum) \
1593 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1594 ? above_latin1 \
cc8ab7c0
KW
1595 : 0)
1596
84238efa
KW
1597/* NOTE that some of these macros have very similar ones in regcharclass.h.
1598 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1599 * there, differing in name only by an underscore from the one here
1600 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1601 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1602 * points; the regcharclass.h ones are implemented as a series of
1603 * "if-else-if-else ..." */
1604
922e8cb4
KW
1605#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1606#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
509fb054
KW
1607#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
1608 utf8, the non-utf8 macro works
1609 */
cd500f2f 1610#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2
KW
1611
1612#ifdef EBCDIC
1613 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1614 * more efficient macro instead of the more general one */
88794300 1615# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
e8fa43e2
KW
1616#else
1617# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1618#endif
1619
cd500f2f
KW
1620#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1621 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1622#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1623#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1624 _is_utf8_perl_idcont, p)
e5dcd934 1625
c11ff943
KW
1626/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1627 * IDFIRST is an alnum. See
c8362b00 1628 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1629 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1630 * This used to be not the XID version, but we decided to go with the more
1631 * modern Unicode definition */
cd500f2f 1632#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1633 _is_utf8_perl_idstart, p)
e5dcd934 1634
922e8cb4
KW
1635#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1636#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
aea0c08c 1637
509fb054 1638/* Posix and regular space are identical above Latin1 */
cd500f2f 1639#define isPSXSPC_utf8(p) _generic_func_utf8(_CC_PSXSPC, is_XPERLSPACE_high, p)
aea0c08c 1640
922e8cb4 1641#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
cd500f2f 1642#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
922e8cb4 1643#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
cd500f2f 1644#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
922e8cb4 1645#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
cd500f2f
KW
1646#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
1647 is_XDIGIT_high(p))
a0ed51b3 1648
7d2b61dd 1649#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1650#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1651#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1652#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1653
4650c663
KW
1654/* For internal core Perl use only: the base macros for defining macros like
1655 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1656 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1657 * 'macro' parameter */
94bb8c36
KW
1658#define _generic_LC_utf8(macro, p, utf8) \
1659 (UTF8_IS_INVARIANT(*(p)) \
1660 ? macro(*(p)) \
1661 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1662 ? macro(TWO_BYTE_UTF8_TO_NATIVE(*(p), *((p)+1))) \
9201a572
KW
1663 : utf8)
1664
cd500f2f
KW
1665#define _generic_LC_swash_utf8(macro, classnum, p) \
1666 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1667#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1668 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1669
cd500f2f
KW
1670#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
1671 _CC_ALPHANUMERIC, p)
1672#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
509fb054 1673#define isASCII_LC_utf8(p) isASCII_LC(*p)
cd500f2f
KW
1674#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, is_HORIZWS_high, p)
1675#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1676#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1677#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
eba68aa0 1678#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, _is_utf8_perl_idcont, p)
cd500f2f
KW
1679#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, _is_utf8_perl_idstart, p)
1680#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1681#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
509fb054
KW
1682#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p) /* space is identical to posix
1683 space under locale */
cd500f2f
KW
1684#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1685#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, is_XPERLSPACE_high, p)
1686#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
1687#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1688 _CC_WORDCHAR, p)
1689#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, is_XDIGIT_high, p)
aaa51d5e 1690
fbc19f27
KW
1691/* Macros for backwards compatibility and for completeness when the ASCII and
1692 * Latin1 values are identical */
fbc19f27
KW
1693#define isALPHAU(c) isALPHA_L1(c)
1694#define isDIGIT_L1(c) isDIGIT_A(c)
1695#define isOCTAL(c) isOCTAL_A(c)
1696#define isOCTAL_L1(c) isOCTAL_A(c)
1697#define isXDIGIT_L1(c) isXDIGIT_A(c)
2e28f0b9
KW
1698#define isALNUM(c) isWORDCHAR(c)
1699#define isALNUMU(c) isWORDCHAR_L1(c)
1700#define isALNUM_LC(c) isWORDCHAR_LC(c)
1701#define isALNUM_uni(c) isWORDCHAR_uni(c)
1702#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
1703#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1704#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
15861f94
KW
1705#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1706#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1707#define isALNUMC(c) isALPHANUMERIC(c)
1708#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1709#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
1710#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
1711#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
1712#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1713
2bd1cbf6
KW
1714/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1715 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1716 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1717 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1718 * the outlier from the block that contains the other controls, just like
1719 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1720 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
1721 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1722 * etc. */
2bd1cbf6 1723#ifndef EBCDIC
75763b3a 1724# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 1725#else
75763b3a
KW
1726# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1727 ((isPRINT_A(c)) \
1728 ? (UNLIKELY((c) == '?') \
1729 ? QUESTION_MARK_CTRL \
1730 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1731 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1732 ? '?' \
1733 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 1734#endif
bbce6d69 1735
dea28490
JJ
1736/* Line numbers are unsigned, 32 bits. */
1737typedef U32 line_t;
e5dcd934 1738#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 1739
91152fc1
DG
1740/* Helpful alias for version prescan */
1741#define is_LAX_VERSION(a,b) \
1742 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1743
1744#define is_STRICT_VERSION(a,b) \
1745 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1746
1747#define BADVERSION(a,b,c) \
1748 if (b) { \
1749 *b = c; \
1750 } \
1751 return a;
8c52afec 1752
cb27eebd
KW
1753/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1754 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1755 * which is advanced. The input is validated only by an assert() in DEBUGGING
1756 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1757 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1758#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1759 ? (c) \
1760 : ((c) + 9))))
902008b8
KW
1761#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1762 ? (*(s)++) \
1763 : (*(s)++ + 9))))
95a59cab 1764
cb27eebd
KW
1765/* Converts a character known to represent an octal digit (0-7) to its numeric
1766 * value. The input is validated only by an assert() in DEBUGGING builds. In
1767 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1768#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1769
305b8651
KW
1770/* Efficiently returns a boolean as to if two native characters are equivalent
1771 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1772 * the ALPHA in the name is to remind you of that. This is asserted() in
1773 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1774 * works (on valid input) for both non- and UTF-8-encoded bytes.
1775 *
1776 * When one of the inputs is a compile-time constant and gets folded by the
1777 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1778 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1779 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
1780 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1781 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1782 * just a single 0, in the bit position where the upper- and lowercase differ.
1783 * */
305b8651
KW
1784#define isALPHA_FOLD_EQ(c1, c2) \
1785 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1786 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1787#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1788
8e84507e 1789/*
ccfc67b7
JH
1790=head1 Memory Management
1791
a02a5408 1792=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
1793The XSUB-writer's interface to the C C<malloc> function.
1794
0d7b2759
MH
1795Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1796
c5008215
JC
1797In 5.9.3, Newx() and friends replace the older New() API, and drops
1798the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 1799themselves. This aid has been superseded by a new build option,
d10b4965 1800PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
1801there for use in XS modules supporting older perls.
1802
a02a5408 1803=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 1804The XSUB-writer's interface to the C C<malloc> function, with
c5008215 1805cast. See also C<Newx>.
954c1994 1806
0d7b2759
MH
1807Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1808
a02a5408 1809=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 1810The XSUB-writer's interface to the C C<malloc> function. The allocated
c5008215 1811memory is zeroed with C<memzero>. See also C<Newx>.
a02a5408 1812
0d7b2759
MH
1813Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1814
954c1994
GS
1815=for apidoc Am|void|Renew|void* ptr|int nitems|type
1816The XSUB-writer's interface to the C C<realloc> function.
1817
0d7b2759
MH
1818Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1819
954c1994
GS
1820=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1821The XSUB-writer's interface to the C C<realloc> function, with
1822cast.
1823
0d7b2759
MH
1824Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1825
49b8b560 1826=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
1827The XSUB-writer's interface to the C C<free> function.
1828
0d7b2759
MH
1829This should B<ONLY> be used on memory obtained using L<"Newx"> and friends.
1830
954c1994
GS
1831=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1832The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c
KW
1833source, C<dest> is the destination, C<nitems> is the number of items, and
1834C<type> is the type. Can do overlapping moves. See also C<Copy>.
954c1994 1835
e90e2364 1836=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
72d33970
FC
1837Like C<Move> but returns dest. Useful
1838for encouraging compilers to tail-call
e90e2364
NC
1839optimise.
1840
954c1994
GS
1841=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1842The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c
KW
1843source, C<dest> is the destination, C<nitems> is the number of items, and
1844C<type> is the type. May fail on overlapping copies. See also C<Move>.
954c1994 1845
e90e2364
NC
1846=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1847
72d33970
FC
1848Like C<Copy> but returns dest. Useful
1849for encouraging compilers to tail-call
e90e2364
NC
1850optimise.
1851
954c1994
GS
1852=for apidoc Am|void|Zero|void* dest|int nitems|type
1853
1854The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1855destination, C<nitems> is the number of items, and C<type> is the type.
1856
e90e2364
NC
1857=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1858
72d33970
FC
1859Like C<Zero> but returns dest. Useful
1860for encouraging compilers to tail-call
e90e2364
NC
1861optimise.
1862
da5d8dbb 1863=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 1864This is an architecture-independent macro to copy one structure to another.
954c1994 1865
7e337ee0
JH
1866=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1867
1868Fill up memory with a byte pattern (a byte repeated over and over
1869again) that hopefully catches attempts to access uninitialized memory.
1870
1871=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1872
1873PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1874
1c12ffb4 1875=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
1876
1877PoisonWith(0xEF) for catching access to freed memory.
1878
9965345d
JH
1879=for apidoc Am|void|Poison|void* dest|int nitems|type
1880
7e337ee0 1881PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
1882
1883=cut */
954c1994 1884
561b68a9
SH
1885/* Maintained for backwards-compatibility only. Use newSV() instead. */
1886#ifndef PERL_CORE
ff06c60c 1887#define NEWSV(x,len) newSV(len)
561b68a9 1888#endif
ff06c60c 1889
19a94d75
JH
1890#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1891
2720f793
JH
1892/* The +0.0 in MEM_WRAP_CHECK_ is an attempt to foil
1893 * overly eager compilers that will bleat about e.g.
1894 * (U16)n > (size_t)~0/sizeof(U16) always being false. */
27d5b266 1895#ifdef PERL_MALLOC_WRAP
f019c49e 1896#define MEM_WRAP_CHECK(n,t) \
5d9574c1 1897 (void)(UNLIKELY(sizeof(t) > 1 && ((MEM_SIZE)(n)+0.0) > MEM_SIZE_MAX/sizeof(t)) && (croak_memory_wrap(),0))
27d5b266 1898#define MEM_WRAP_CHECK_1(n,t,a) \
5d9574c1 1899 (void)(UNLIKELY(sizeof(t) > 1 && ((MEM_SIZE)(n)+0.0) > MEM_SIZE_MAX/sizeof(t)) && (Perl_croak_nocontext("%s",(a)),0))
8b44ba4c 1900#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 1901
d1decf2b 1902#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
1903#else
1904
410319be
NC
1905#define MEM_WRAP_CHECK(n,t)
1906#define MEM_WRAP_CHECK_1(n,t,a)
1907#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
1908#define MEM_WRAP_CHECK_(n,t)
1909
6fff79ce 1910#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 1911
1936d2a7 1912#endif
8b44ba4c 1913
fe4f188c 1914#ifdef PERL_MEM_LOG
46c6c7e2 1915/*
9f653bb5 1916 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
1917 * go through functions, which are handy for debugging breakpoints, but
1918 * which more importantly get the immediate calling environment (file and
e352bcff
JH
1919 * line number, and C function name if available) passed in. This info can
1920 * then be used for logging the calls, for which one gets a sample
73d1d973 1921 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 1922 *
46c6c7e2 1923 * Known problems:
94e892a6 1924 * - not all memory allocs get logged, only those
46c6c7e2 1925 * that go through Newx() and derivatives (while all
94e892a6 1926 * Safefrees do get logged)
46c6c7e2
JH
1927 * - __FILE__ and __LINE__ do not work everywhere
1928 * - __func__ or __FUNCTION__ even less so
1929 * - I think more goes on after the perlio frees but
1930 * the thing is that STDERR gets closed (as do all
1931 * the file descriptors)
1932 * - no deeper calling stack than the caller of the Newx()
1933 * or the kind, but do I look like a C reflection/introspection
1934 * utility to you?
1935 * - the function prototypes for the logging functions
1936 * probably should maybe be somewhere else than handy.h
1937 * - one could consider inlining (macrofying) the logging
1938 * for speed, but I am too lazy
1939 * - one could imagine recording the allocations in a hash,
1940 * (keyed by the allocation address?), and maintain that
1941 * through reallocs and frees, but how to do that without
1942 * any News() happening...?
73d1d973 1943 * - lots of -Ddefines to get useful/controllable output
b953482e 1944 * - lots of ENV reads
46c6c7e2
JH
1945 */
1946
12754f92 1947PERL_EXPORT_C 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 1948
12754f92 1949PERL_EXPORT_C 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 1950
12754f92 1951PERL_EXPORT_C Malloc_t Perl_mem_log_free(Malloc_t oldalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 1952
0b0ab801 1953# ifdef PERL_CORE
73d1d973 1954# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
1955enum mem_log_type {
1956 MLT_ALLOC,
1957 MLT_REALLOC,
d7a2c63c
MHM
1958 MLT_FREE,
1959 MLT_NEW_SV,
1960 MLT_DEL_SV
0b0ab801
MHM
1961};
1962# endif
12754f92 1963# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
1964void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
1965void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 1966# endif
0b0ab801
MHM
1967# endif
1968
fe4f188c
JH
1969#endif
1970
1971#ifdef PERL_MEM_LOG
d1401ee9
MHM
1972#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
1973#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 1974#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
1975#endif
1976
1977#ifndef MEM_LOG_ALLOC
1978#define MEM_LOG_ALLOC(n,t,a) (a)
1979#endif
1980#ifndef MEM_LOG_REALLOC
1981#define MEM_LOG_REALLOC(n,t,v,a) (a)
1982#endif
1983#ifndef MEM_LOG_FREE
1984#define MEM_LOG_FREE(a) (a)
1985#endif
1986
d1401ee9
MHM
1987#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
1988#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
1989#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
1990
1991#ifndef PERL_CORE
a02a5408
JC
1992/* pre 5.9.x compatibility */
1993#define New(x,v,n,t) Newx(v,n,t)
1994#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 1995#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 1996#endif
a02a5408 1997
ff68c719 1998#define Renew(v,n,t) \
d1401ee9 1999 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2000#define Renewc(v,n,t,c) \
d1401ee9 2001 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2002
2003#ifdef PERL_POISON
2004#define Safefree(d) \
06c0cc96 2005 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2006#else
fe4f188c 2007#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2008#endif
55497cff 2009
8b44ba4c
NC
2010#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2011#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2012#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2013
8b44ba4c
NC
2014#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2015#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2016#ifdef HAS_MEMSET
8b44ba4c 2017#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2018#else
8b44ba4c
NC
2019/* Using bzero(), which returns void. */
2020#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2021#endif
2022
7e337ee0
JH
2023#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2024#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2025#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2026#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2027
caa674f3
DD
2028#ifdef PERL_POISON
2029# define PERL_POISON_EXPR(x) x
2030#else
2031# define PERL_POISON_EXPR(x)
2032#endif
2033
2304df62 2034#ifdef USE_STRUCT_COPY
ff68c719 2035#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2036#else
2037#define StructCopy(s,d,t) Copy(s,d,1,t)
2038#endif
2cc61e15 2039
c3caa5c3
JH
2040/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2041 * want your zero-based indices to be less than but not equal to).
2042 *
2043 * C_ARRAY_END is one past the last: half-open/half-closed range,
2044 * not last-inclusive range. */
622913ab 2045#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2046#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2047
2cc61e15
DD
2048#ifdef NEED_VA_COPY
2049# ifdef va_copy
2050# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2051# else
a1866d1b
JH
2052# if defined(__va_copy)
2053# define Perl_va_copy(s, d) __va_copy(d, s)
2054# else
2055# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2056# endif
2cc61e15
DD
2057# endif
2058#endif
2059
472d47bc
SB
2060/* convenience debug macros */
2061#ifdef USE_ITHREADS
2062#define pTHX_FORMAT "Perl interpreter: 0x%p"
2063#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2064#define pTHX_VALUE_ (void *)my_perl,
2065#define pTHX_VALUE (void *)my_perl
2066#define pTHX__VALUE_ ,(void *)my_perl,
2067#define pTHX__VALUE ,(void *)my_perl
472d47bc 2068#else
3609ea0d 2069#define pTHX_FORMAT
472d47bc 2070#define pTHX__FORMAT
3609ea0d 2071#define pTHX_VALUE_
472d47bc 2072#define pTHX_VALUE
3609ea0d 2073#define pTHX__VALUE_
472d47bc
SB
2074#define pTHX__VALUE
2075#endif /* USE_ITHREADS */
3609ea0d 2076
2acdbac1
NC
2077/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2078 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2079 CPAN::Unpack show any users outside the core. */
2080#ifdef PERL_CORE
d1d15184 2081# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), "Use of " s " is deprecated")
2acdbac1
NC
2082#endif
2083
dfff4baf
BF
2084/* Internal macros to deal with gids and uids */
2085#ifdef PERL_CORE
2086
2087# if Uid_t_size > IVSIZE
2088# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2089# define SvUID(sv) SvNV(sv)
2090# else
2091# if Uid_t_sign <= 0
2092# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2093# define SvUID(sv) SvIV(sv)
2094# else
2095# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2096# define SvUID(sv) SvUV(sv)
2097# endif
2098# endif /* Uid_t_size */
2099
2100# if Gid_t_size > IVSIZE
2101# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2102# define SvGID(sv) SvNV(sv)
2103# else
2104# if Gid_t_sign <= 0
2105# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2106# define SvGID(sv) SvIV(sv)
2107# else
2108# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2109# define SvGID(sv) SvUV(sv)
2110# endif
2111# endif /* Gid_t_size */
2112
2113#endif
2114
9d745869
KW
2115#endif /* HANDY_H */
2116
e9a8c099
MHM
2117/*
2118 * Local variables:
2119 * c-indentation-style: bsd
2120 * c-basic-offset: 4
14d04a33 2121 * indent-tabs-mode: nil
e9a8c099
MHM
2122 * End:
2123 *
14d04a33 2124 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2125 */