This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
lib/locale.t: Add some \Q
[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
ff68c719 278#define Ctl(ch) ((ch) & 037)
8d063cd8 279
4650c663
KW
280/* This is a helper macro to avoid preprocessor issues, replaced by nothing
281 * unless under DEBUGGING, where it expands to an assert of its argument,
282 * followed by a comma (hence the comma operator). If we just used a straight
283 * assert(), we would get a comma with nothing before it when not DEBUGGING */
0f092d08
KW
284#ifdef DEBUGGING
285# define __ASSERT_(statement) assert(statement),
286#else
287# define __ASSERT_(statement)
288#endif
289
3fe05580
MHM
290/*
291=head1 SV-Body Allocation
292
293=for apidoc Ama|SV*|newSVpvs|const char* s
30a15352
KW
294Like C<newSVpvn>, but takes a literal C<NUL>-terminated string instead of a
295string/length pair.
3fe05580 296
84bafc02 297=for apidoc Ama|SV*|newSVpvs_flags|const char* s|U32 flags
30a15352
KW
298Like C<newSVpvn_flags>, but takes a literal C<NUL>-terminated string instead of
299a string/length pair.
84bafc02 300
3fe05580 301=for apidoc Ama|SV*|newSVpvs_share|const char* s
30a15352
KW
302Like C<newSVpvn_share>, but takes a literal C<NUL>-terminated string instead of
303a string/length pair and omits the hash parameter.
3fe05580 304
9dcc53ea 305=for apidoc Am|void|sv_catpvs_flags|SV* sv|const char* s|I32 flags
30a15352
KW
306Like C<sv_catpvn_flags>, but takes a literal C<NUL>-terminated string instead
307of a string/length pair.
9dcc53ea
Z
308
309=for apidoc Am|void|sv_catpvs_nomg|SV* sv|const char* s
310Like C<sv_catpvn_nomg>, but takes a literal string instead of a
311string/length pair.
312
7b4e7469 313=for apidoc Am|void|sv_catpvs|SV* sv|const char* s
3fe05580
MHM
314Like C<sv_catpvn>, but takes a literal string instead of a string/length pair.
315
9dcc53ea
Z
316=for apidoc Am|void|sv_catpvs_mg|SV* sv|const char* s
317Like C<sv_catpvn_mg>, but takes a literal string instead of a
318string/length pair.
319
7b4e7469 320=for apidoc Am|void|sv_setpvs|SV* sv|const char* s
3fe05580
MHM
321Like C<sv_setpvn>, but takes a literal string instead of a string/length pair.
322
9dcc53ea
Z
323=for apidoc Am|void|sv_setpvs_mg|SV* sv|const char* s
324Like C<sv_setpvn_mg>, but takes a literal string instead of a
325string/length pair.
326
327=for apidoc Am|SV *|sv_setref_pvs|const char* s
328Like C<sv_setref_pvn>, but takes a literal string instead of a
329string/length pair.
330
3fe05580
MHM
331=head1 Memory Management
332
333=for apidoc Ama|char*|savepvs|const char* s
30a15352
KW
334Like C<savepvn>, but takes a literal C<NUL>-terminated string instead of a
335string/length pair.
3fe05580 336
9dcc53ea
Z
337=for apidoc Ama|char*|savesharedpvs|const char* s
338A version of C<savepvs()> which allocates the duplicate string in memory
339which is shared between threads.
340
3fe05580
MHM
341=head1 GV Functions
342
343=for apidoc Am|HV*|gv_stashpvs|const char* name|I32 create
344Like C<gv_stashpvn>, but takes a literal string instead of a string/length pair.
345
346=head1 Hash Manipulation Functions
347
348=for apidoc Am|SV**|hv_fetchs|HV* tb|const char* key|I32 lval
349Like C<hv_fetch>, but takes a literal string instead of a string/length pair.
350
351=for apidoc Am|SV**|hv_stores|HV* tb|const char* key|NULLOK SV* val
352Like C<hv_store>, but takes a literal string instead of a string/length pair
353and omits the hash parameter.
354
510966aa
Z
355=head1 Lexer interface
356
357=for apidoc Amx|void|lex_stuff_pvs|const char *pv|U32 flags
358
359Like L</lex_stuff_pvn>, but takes a literal string instead of a
360string/length pair.
361
3fe05580
MHM
362=cut
363*/
364
926bb54c
KW
365/* concatenating with "" ensures that only literal strings are accepted as
366 * argument */
26bfd924 367#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
2efa8cc7 368
926bb54c
KW
369/* note that STR_WITH_LEN() can't be used as argument to macros or functions
370 * that under some configurations might be macros, which means that it requires
371 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
ba3a79e7
GA
372 */
373
374/* STR_WITH_LEN() shortcuts */
375#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
376#define newSVpvs_flags(str,flags) \
377 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 378#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
379#define sv_catpvs_flags(sv, str, flags) \
380 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
381#define sv_catpvs_nomg(sv, str) \
382 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
383#define sv_catpvs(sv, str) \
384 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
385#define sv_catpvs_mg(sv, str) \
386 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 387#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
388#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
389#define sv_setref_pvs(rv, classname, str) \
390 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 391#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
392#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
393#define gv_stashpvs(str, create) \
394 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
395#define gv_fetchpvs(namebeg, add, sv_type) \
396 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
397#define gv_fetchpvn(namebeg, len, add, sv_type) \
398 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
399#define sv_catxmlpvs(dsv, str, utf8) \
400 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
a038e571
NC
401#define hv_fetchs(hv,key,lval) \
402 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
403 (lval) ? (HV_FETCH_JUST_SV | HV_FETCH_LVALUE) \
404 : HV_FETCH_JUST_SV, NULL, 0))
405
406#define hv_stores(hv,key,val) \
407 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
408 (HV_FETCH_ISSTORE|HV_FETCH_JUST_SV), (val), 0))
ba3a79e7 409
510966aa
Z
410#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
411
b96d8cd9
NC
412#define get_cvs(str, flags) \
413 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 414
954c1994 415/*
ccfc67b7
JH
416=head1 Miscellaneous Functions
417
954c1994
GS
418=for apidoc Am|bool|strNE|char* s1|char* s2
419Test two strings to see if they are different. Returns true or
420false.
421
422=for apidoc Am|bool|strEQ|char* s1|char* s2
423Test two strings to see if they are equal. Returns true or false.
424
425=for apidoc Am|bool|strLT|char* s1|char* s2
426Test two strings to see if the first, C<s1>, is less than the second,
427C<s2>. Returns true or false.
428
429=for apidoc Am|bool|strLE|char* s1|char* s2
430Test two strings to see if the first, C<s1>, is less than or equal to the
431second, C<s2>. Returns true or false.
432
433=for apidoc Am|bool|strGT|char* s1|char* s2
434Test two strings to see if the first, C<s1>, is greater than the second,
435C<s2>. Returns true or false.
436
437=for apidoc Am|bool|strGE|char* s1|char* s2
438Test two strings to see if the first, C<s1>, is greater than or equal to
439the second, C<s2>. Returns true or false.
440
441=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
442Test two strings to see if they are different. The C<len> parameter
72d33970 443indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
444wrapper for C<strncmp>).
445
446=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
447Test two strings to see if they are equal. The C<len> parameter indicates
72d33970 448the number of bytes to compare. Returns true or false. (A wrapper for
954c1994
GS
449C<strncmp>).
450
451=cut
452*/
453
8d063cd8
LW
454#define strNE(s1,s2) (strcmp(s1,s2))
455#define strEQ(s1,s2) (!strcmp(s1,s2))
456#define strLT(s1,s2) (strcmp(s1,s2) < 0)
457#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
458#define strGT(s1,s2) (strcmp(s1,s2) > 0)
459#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
460#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
461#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
378cc40b 462
36477c24
PP
463#ifdef HAS_MEMCMP
464# define memNE(s1,s2,l) (memcmp(s1,s2,l))
465# define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
466#else
467# define memNE(s1,s2,l) (bcmp(s1,s2,l))
468# define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
469#endif
470
568a785a 471#define memEQs(s1, l, s2) \
8e6497b6 472 (sizeof(s2)-1 == l && memEQ(s1, ("" s2 ""), (sizeof(s2)-1)))
568a785a
NC
473#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
474
bbce6d69
PP
475/*
476 * Character classes.
477 *
478 * Unfortunately, the introduction of locales means that we
479 * can't trust isupper(), etc. to tell the truth. And when
480 * it comes to /\w+/ with tainting enabled, we *must* be able
481 * to trust our character classes.
482 *
483 * Therefore, the default tests in the text of Perl will be
484 * independent of locale. Any code that wants to depend on
485 * the current locale will use the tests that begin with "lc".
486 */
487
2304df62
AD
488#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
489# ifndef CTYPE256
490# define CTYPE256
491# endif
492#endif
493
954c1994 494/*
ccfc67b7 495
dcccc8ff 496=head1 Character classification
243effed
KW
497This section is about functions (really macros) that classify characters
498into types, such as punctuation versus alphabetic, etc. Most of these are
499analogous to regular expression character classes. (See
500L<perlrecharclass/POSIX Character Classes>.) There are several variants for
501each class. (Not all macros have all variants; each item below lists the
502ones valid for it.) None are affected by C<use bytes>, and only the ones
503with C<LC> in the name are affected by the current locale.
504
505The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
506C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
507represented by that octet is (or on non-ASCII platforms, corresponds to) an
508ASCII character in the named class based on platform, Unicode, and Perl rules.
509If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 510
6aff1f14
KW
511Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
512with no suffix C<"_A">.
4b9734bf
KW
513
514Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
515onto the platform. That is, the code points that are ASCII are unaffected,
516since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
517as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
518true when called with the code point 0xDF, which is a word character in both
4650c663 519ASCII and EBCDIC (though it represents different characters in each).
243effed
KW
520
521Variant C<isFOO_uni> is like the C<isFOO_L1> variant, but accepts any UV code
522point as input. If the code point is larger than 255, Unicode rules are used
523to determine if it is in the character class. For example,
6aff1f14
KW
524C<isWORDCHAR_uni(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
525WITH MACRON in Unicode, and is a word character.
243effed
KW
526
527Variant C<isFOO_utf8> is like C<isFOO_uni>, but the input is a pointer to a
528(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
6aff1f14
KW
529classification of just the first (possibly multi-byte) character in the string
530is tested.
243effed 531
31f05a37
KW
532Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
533result is based on the current locale, which is what C<LC> in the name stands
534for. If Perl can determine that the current locale is a UTF-8 locale, it uses
535the published Unicode rules; otherwise, it uses the C library function that
536gives the named classification. For example, C<isDIGIT_LC()> when not in a
537UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
538returned if the input won't fit into an octet. On some platforms where the C
539library function is known to be defective, Perl changes its result to follow
540the POSIX standard's rules.
243effed
KW
541
542Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
543returns the same as C<isFOO_LC> for input code points less than 256, and
544returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
545
546Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer to a
547(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
6aff1f14
KW
548classification of just the first (possibly multi-byte) character in the string
549is tested.
ccfc67b7 550
954c1994 551=for apidoc Am|bool|isALPHA|char ch
2787a470 552Returns a boolean indicating whether the specified character is an
6aff1f14 553alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
554See the L<top of this section|/Character classification> for an explanation of
555variants
e5ad6aba 556C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uni>, C<isALPHA_utf8>, C<isALPHA_LC>,
243effed 557C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
8a58bdcf 558
15861f94
KW
559=for apidoc Am|bool|isALPHANUMERIC|char ch
560Returns a boolean indicating whether the specified character is a either an
6aff1f14 561alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
562See the L<top of this section|/Character classification> for an explanation of
563variants
15861f94 564C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uni>,
e5ad6aba 565C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
15861f94
KW
566C<isALPHANUMERIC_LC_utf8>.
567
8a58bdcf
KW
568=for apidoc Am|bool|isASCII|char ch
569Returns a boolean indicating whether the specified character is one of the 128
243effed 570characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 571On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
572character corresponds to an ASCII character. Variants C<isASCII_A()> and
573C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
574See the L<top of this section|/Character classification> for an explanation of
575variants
243effed
KW
576C<isASCII_uni>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
577C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
578library routine C<isascii()>. In these cases, the variants whose names contain
579C<LC> are the same as the corresponding ones without.
580
d98532ea
KW
581Also note, that because all ASCII characters are UTF-8 invariant (meaning they
582have the exact same representation (always a single byte) whether encoded in
583UTF-8 or not), C<isASCII> will give the correct results when called with any
584byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
585work properly on any string encoded or not in UTF-8.
586
243effed
KW
587=for apidoc Am|bool|isBLANK|char ch
588Returns a boolean indicating whether the specified character is a
6aff1f14 589character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
590See the L<top of this section|/Character classification> for an explanation of
591variants
e5ad6aba 592C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uni>, C<isBLANK_utf8>, C<isBLANK_LC>,
243effed
KW
593C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
594platforms do not have the C library routine C<isblank()>. In these cases, the
595variants whose names contain C<LC> are the same as the corresponding ones
596without.
597
598=for apidoc Am|bool|isCNTRL|char ch
599Returns a boolean indicating whether the specified character is a
6aff1f14 600control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
601See the L<top of this section|/Character classification> for an explanation of
602variants
e5ad6aba 603C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uni>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
6aff1f14
KW
604C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
605On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 606
954c1994 607=for apidoc Am|bool|isDIGIT|char ch
2787a470 608Returns a boolean indicating whether the specified character is a
6aff1f14 609digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 610Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
611See the L<top of this section|/Character classification> for an explanation of
612variants
e5ad6aba 613C<isDIGIT_uni>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
243effed
KW
614C<isDIGIT_LC_utf8>.
615
616=for apidoc Am|bool|isGRAPH|char ch
617Returns a boolean indicating whether the specified character is a
6aff1f14 618graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff
KW
619See the L<top of this section|/Character classification> for an explanation of
620variants
e5ad6aba 621C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uni>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
243effed 622C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
954c1994 623
0c82b6df 624=for apidoc Am|bool|isLOWER|char ch
2787a470 625Returns a boolean indicating whether the specified character is a
6aff1f14 626lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
627See the L<top of this section|/Character classification> for an explanation of
628variants
e5ad6aba 629C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uni>, C<isLOWER_utf8>, C<isLOWER_LC>,
243effed 630C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
0c82b6df 631
c99e91e9 632=for apidoc Am|bool|isOCTAL|char ch
2787a470 633Returns a boolean indicating whether the specified character is an
6aff1f14 634octal digit, [0-7].
243effed
KW
635The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
636C<isOCTAL>.
637
638=for apidoc Am|bool|isPUNCT|char ch
639Returns a boolean indicating whether the specified character is a
6aff1f14
KW
640punctuation character, analogous to C<m/[[:punct:]]/>.
641Note that the definition of what is punctuation isn't as
243effed
KW
642straightforward as one might desire. See L<perlrecharclass/POSIX Character
643Classes> for details.
dcccc8ff
KW
644See the L<top of this section|/Character classification> for an explanation of
645variants
e5ad6aba 646C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uni>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
243effed 647C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
c99e91e9 648
0c82b6df 649=for apidoc Am|bool|isSPACE|char ch
2787a470 650Returns a boolean indicating whether the specified character is a
6aff1f14 651whitespace character. This is analogous
398d098a
KW
652to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
653(experimentally), this also matches what C<m/[[:space:]]/> does.
a04e6aad 654("Experimentally" means that this change may be backed out in 5.22 if
398d098a
KW
655field experience indicates that it was unwise.) Prior to 5.18, only the
656locale forms of this macro (the ones with C<LC> in their names) matched
657precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
658in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
659(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
660See the L<top of this section|/Character classification> for an explanation of
661variants
e5ad6aba 662C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uni>, C<isSPACE_utf8>, C<isSPACE_LC>,
243effed 663C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
0c82b6df 664
398d098a
KW
665=for apidoc Am|bool|isPSXSPC|char ch
666(short for Posix Space)
667Starting in 5.18, this is identical (experimentally) in all its forms to the
668corresponding C<isSPACE()> macros. ("Experimentally" means that this change
a04e6aad 669may be backed out in 5.22 if field experience indicates that it
398d098a
KW
670was unwise.)
671The locale forms of this macro are identical to their corresponding
672C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
673non-locale forms differ from their C<isSPACE()> forms only in that the
674C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
675Otherwise they are identical. Thus this macro is analogous to what
676C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff
KW
677See the L<top of this section|/Character classification> for an explanation of
678variants
e5ad6aba 679C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uni>, C<isPSXSPC_utf8>, C<isPSXSPC_LC>,
398d098a
KW
680C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
681
954c1994 682=for apidoc Am|bool|isUPPER|char ch
2787a470 683Returns a boolean indicating whether the specified character is an
6aff1f14 684uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff
KW
685See the L<top of this section|/Character classification> for an explanation of
686variants
e5ad6aba 687C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uni>, C<isUPPER_utf8>, C<isUPPER_LC>,
243effed 688C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
954c1994 689
243effed 690=for apidoc Am|bool|isPRINT|char ch
8eea39dd 691Returns a boolean indicating whether the specified character is a
6aff1f14 692printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
693See the L<top of this section|/Character classification> for an explanation of
694variants
e5ad6aba 695C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uni>, C<isPRINT_utf8>, C<isPRINT_LC>,
243effed
KW
696C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
697
698=for apidoc Am|bool|isWORDCHAR|char ch
699Returns a boolean indicating whether the specified character is a character
700that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
701in a regular expression. A word character is an alphabetic character, a
702decimal digit, a connecting punctuation character (such as an underscore), or
703a "mark" character that attaches to one of those (like some sort of accent).
704C<isALNUM()> is a synonym provided for backward compatibility, even though a
705word character includes more than the standard C language meaning of
706alphanumeric.
dcccc8ff
KW
707See the L<top of this section|/Character classification> for an explanation of
708variants
1a83413c
KW
709C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uni>, and C<isWORDCHAR_utf8>.
710C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
711described there, but additionally include the platform's native underscore.
8a58bdcf
KW
712
713=for apidoc Am|bool|isXDIGIT|char ch
714Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
715digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
716and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
717See the L<top of this section|/Character classification> for an explanation of
718variants
243effed
KW
719C<isXDIGIT_uni>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
720C<isXDIGIT_LC_utf8>.
721
3c3ecf18
KW
722=for apidoc Am|bool|isIDFIRST|char ch
723Returns a boolean indicating whether the specified character can be the first
724character of an identifier. This is very close to, but not quite the same as
725the official Unicode property C<XID_Start>. The difference is that this
726returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
727See the L<top of this section|/Character classification> for an explanation of
728variants
3c3ecf18
KW
729C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uni>, C<isIDFIRST_utf8>,
730C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
731
732=for apidoc Am|bool|isIDCONT|char ch
733Returns a boolean indicating whether the specified character can be the
734second or succeeding character of an identifier. This is very close to, but
735not quite the same as the official Unicode property C<XID_Continue>. The
736difference is that this returns true only if the input character also matches
dcccc8ff
KW
737L</isWORDCHAR>. See the L<top of this section|/Character classification> for
738an
3c3ecf18
KW
739explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uni>,
740C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
741C<isIDCONT_LC_utf8>.
742
243effed 743=head1 Miscellaneous Functions
8eea39dd 744
95a59cab 745=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 746Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
747Behaviour is only well defined when isXDIGIT(*str) is true.
748
e7c1e6c1
KW
749=head1 Character case changing
750
1f607577
KW
751=for apidoc Am|U8|toUPPER|U8 ch
752Converts the specified character to uppercase. If the input is anything but an
753ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 754C<toUPPER_A> is equivalent.
954c1994 755
1f607577
KW
756=for apidoc Am|UV|toUPPER_uni|UV cp|U8* s|STRLEN* lenp
757Converts the Unicode code point C<cp> to its uppercase version, and
758stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
759that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
760bytes since the uppercase version may be longer than the original character.
761
762The first code point of the uppercased version is returned
763(but note, as explained just above, that there may be more.)
764
765=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
766Converts the UTF-8 encoded character at C<p> to its uppercase version, and
767stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
768that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
769bytes since the uppercase version may be longer than the original character.
770
771The first code point of the uppercased version is returned
772(but note, as explained just above, that there may be more.)
773
774The input character at C<p> is assumed to be well-formed.
775
25200305
KW
776=for apidoc Am|U8|toFOLD|U8 ch
777Converts the specified character to foldcase. If the input is anything but an
778ASCII uppercase character, that input character itself is returned. Variant
779C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
780Latin1 range, as the full generality of L</toFOLD_uni> is needed there.)
781
1f607577
KW
782=for apidoc Am|UV|toFOLD_uni|UV cp|U8* s|STRLEN* lenp
783Converts the Unicode code point C<cp> to its foldcase version, and
784stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
785that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
786bytes since the foldcase version may be longer than the original character.
787
788The first code point of the foldcased version is returned
789(but note, as explained just above, that there may be more.)
790
791=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
792Converts the UTF-8 encoded character at C<p> to its foldcase version, and
793stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
794that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
795bytes since the foldcase version may be longer than the original character.
796
797The first code point of the foldcased version is returned
798(but note, as explained just above, that there may be more.)
799
800The input character at C<p> is assumed to be well-formed.
801
802=for apidoc Am|U8|toLOWER|U8 ch
803Converts the specified character to lowercase. If the input is anything but an
804ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 805C<toLOWER_A> is equivalent.
954c1994 806
1f607577
KW
807=for apidoc Am|U8|toLOWER_L1|U8 ch
808Converts the specified Latin1 character to lowercase. The results are undefined if
809the input doesn't fit in a byte.
810
811=for apidoc Am|U8|toLOWER_LC|U8 ch
812Converts the specified character to lowercase using the current locale's rules,
813if possible; otherwise returns the input character itself.
814
815=for apidoc Am|UV|toLOWER_uni|UV cp|U8* s|STRLEN* lenp
816Converts the Unicode code point C<cp> to its lowercase version, and
817stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
818that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
819bytes since the lowercase version may be longer than the original character.
820
821The first code point of the lowercased version is returned
822(but note, as explained just above, that there may be more.)
823
824=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
825Converts the UTF-8 encoded character at C<p> to its lowercase version, and
826stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
827that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
828bytes since the lowercase version may be longer than the original character.
829
830The first code point of the lowercased version is returned
831(but note, as explained just above, that there may be more.)
832
833The input character at C<p> is assumed to be well-formed.
834
835=for apidoc Am|U8|toLOWER_LC|U8 ch
836Converts the specified character to lowercase using the current locale's rules,
837if possible; otherwise returns the input character itself.
838
25200305
KW
839=for apidoc Am|U8|toTITLE|U8 ch
840Converts the specified character to titlecase. If the input is anything but an
841ASCII lowercase character, that input character itself is returned. Variant
842C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1 range,
843as the full generality of L</toTITLE_uni> is needed there. Titlecase is not a
844concept used in locale handling, so there is no functionality for that.)
845
1f607577
KW
846=for apidoc Am|UV|toTITLE_uni|UV cp|U8* s|STRLEN* lenp
847Converts the Unicode code point C<cp> to its titlecase version, and
848stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
849that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
850bytes since the titlecase version may be longer than the original character.
851
852The first code point of the titlecased version is returned
853(but note, as explained just above, that there may be more.)
854
855=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
856Converts the UTF-8 encoded character at C<p> to its titlecase version, and
857stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
858that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
859bytes since the titlecase version may be longer than the original character.
860
861The first code point of the titlecased version is returned
862(but note, as explained just above, that there may be more.)
863
864The input character at C<p> is assumed to be well-formed.
865
954c1994 866=cut
353c9b6f 867
1f607577 868XXX Still undocumented isVERTWS_uni and _utf8; it's unclear what their names
25200305 869really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 870
8a58bdcf 871Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
872patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
873
954c1994
GS
874*/
875
7c062697
KW
876/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
877 * is known only in the perl core, and this macro can be called from outside
878 * that */
879#ifdef HAS_QUAD
880# define WIDEST_UTYPE U64TYPE
881#else
882# define WIDEST_UTYPE U32
883#endif
884
3912bc88
KW
885/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
886 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
887 * bits of information are lost even on a 64-bit machine, but to get the
888 * compiler to optimize it out if possible. This is because Configure makes
889 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
890 * sizeof() guarantees that this evaluates to a constant true at compile time.
220c71bf
KW
891 */
892#define FITS_IN_8_BITS(c) ((sizeof(c) == 1) || !(((WIDEST_UTYPE)(c)) & ~0xFF))
cf301eb7 893
41f43cc2 894#ifdef EBCDIC
b6340bd0 895# ifndef _ALL_SOURCE
0852beac
KW
896 /* The native libc isascii() et.al. functions return the wrong results
897 * on at least z/OS unless this is defined. */
b6340bd0
KW
898# error _ALL_SOURCE should probably be defined
899# endif
41f43cc2 900#else
0852beac
KW
901 /* There is a simple definition of ASCII for ASCII platforms. But the
902 * EBCDIC one isn't so simple, so is defined using table look-up like the
903 * other macros below */
41f43cc2
KW
904# define isASCII(c) ((WIDEST_UTYPE)(c) < 128)
905#endif
906
c2da0b36
KW
907/* The lower 3 bits in both the ASCII and EBCDIC representations of '0' are 0,
908 * and the 8 possible permutations of those bits exactly comprise the 8 octal
909 * digits */
3d33b3e3 910#define isOCTAL_A(c) cBOOL(FITS_IN_8_BITS(c) && (0xF8 & (c)) == '0')
c2da0b36 911
9fb1bf9d 912#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 913
a500dc72
KW
914/* Character class numbers. For internal core Perl use only. The ones less
915 * than 32 are used in PL_charclass[] and the ones up through the one that
916 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
917 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
918 * their actual definitions are here. If that file has a name not used here,
919 * it won't compile.
1709d539
KW
920 *
921 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
922 * frequency of their use. This gives a slight edge to exiting a loop earlier
923 * (in reginclass() in regexec.c) */
1709d539
KW
924# define _CC_WORDCHAR 0 /* \w and [:word:] */
925# define _CC_DIGIT 1 /* \d and [:digit:] */
926# define _CC_ALPHA 2 /* [:alpha:] */
927# define _CC_LOWER 3 /* [:lower:] */
928# define _CC_UPPER 4 /* [:upper:] */
929# define _CC_PUNCT 5 /* [:punct:] */
930# define _CC_PRINT 6 /* [:print:] */
15861f94 931# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 932# define _CC_GRAPH 8 /* [:graph:] */
b0d691b2 933# define _CC_CASED 9 /* [:lower:] and [:upper:] under /i */
1709d539 934
b0d691b2 935#define _FIRST_NON_SWASH_CC 10
1709d539
KW
936/* The character classes above are implemented with swashes. The second group
937 * (just below) contains the ones implemented without. These are also sorted
938 * in rough order of the frequency of their use, except that \v should be last,
939 * as it isn't a real Posix character class, and some (small) inefficiencies in
940 * regular expression handling would be introduced by putting it in the middle
941 * of those that are. Also, cntrl and ascii come after the others as it may be
942 * useful to group these which have no members that match above Latin1, (or
943 * above ASCII in the latter case) */
944
b0d691b2
KW
945# define _CC_SPACE 10 /* \s */
946# define _CC_BLANK 11 /* [:blank:] */
947# define _CC_XDIGIT 12 /* [:xdigit:] */
948# define _CC_PSXSPC 13 /* [:space:] */
949# define _CC_CNTRL 14 /* [:cntrl:] */
950# define _CC_ASCII 15 /* [:ascii:] */
951# define _CC_VERTSPACE 16 /* \v */
1709d539 952
a0947d7b
KW
953# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
954
1709d539 955/* The members of the third group below do not need to be coordinated with data
3ffc8c70 956 * structures in regcomp.[ch] and regexec.c. */
f12c0118
KW
957# define _CC_IDFIRST 17
958# define _CC_CHARNAME_CONT 18
959# define _CC_NONLATIN1_FOLD 19
960# define _CC_NONLATIN1_SIMPLE_FOLD 20
961# define _CC_QUOTEMETA 21
962# define _CC_NON_FINAL_FOLD 22
963# define _CC_IS_IN_SOME_FOLD 23
8e35b056
KW
964# define _CC_MNEMONIC_CNTRL 24
965/* Unused: 25-31
f4cdb42c
KW
966 * If more bits are needed, one could add a second word for non-64bit
967 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
968 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
969 * is used only for optimization (as of this writing), and differs in the
970 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72
KW
971 * characters: the masculine and feminine ordinal indicators, so removing it
972 * would just cause /i regexes which match them to run less efficiently */
96ac0975 973
3a371f2f
KW
974#if defined(PERL_CORE) || defined(PERL_EXT)
975/* An enum version of the character class numbers, to help compilers
976 * optimize */
977typedef enum {
3a371f2f 978 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
979 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
980 _CC_ENUM_ASCII = _CC_ASCII,
981 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 982 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 983 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
984 _CC_ENUM_DIGIT = _CC_DIGIT,
985 _CC_ENUM_GRAPH = _CC_GRAPH,
986 _CC_ENUM_LOWER = _CC_LOWER,
987 _CC_ENUM_PRINT = _CC_PRINT,
e8d596e0 988 _CC_ENUM_PSXSPC = _CC_PSXSPC,
3a371f2f 989 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 990 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 991 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 992 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 993 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 994 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
995} _char_class_number;
996#endif
997
63c61c3f 998#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 999#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1000
3018b823 1001#if defined(PERL_IN_UTF8_C) || defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_REGEXEC_C)
922e8cb4 1002# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1003 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1004 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1005 #error Need to adjust order of swash_property_names[]
1006# endif
1007
b44daeeb
KW
1008/* This is declared static in each of the few files that this is #defined for
1009 * to keep them from being publicly accessible. Hence there is a small amount
1010 * of wasted space */
1011
922e8cb4
KW
1012static const char* const swash_property_names[] = {
1013 "XPosixWord",
1014 "XPosixDigit",
1015 "XPosixAlpha",
1016 "XPosixLower",
1017 "XPosixUpper",
1018 "XPosixPunct",
1019 "XPosixPrint",
1020 "XPosixAlnum",
b0d691b2
KW
1021 "XPosixGraph",
1022 "Cased"
922e8cb4
KW
1023};
1024#endif
1025
96ac0975
NC
1026# ifdef DOINIT
1027EXTCONST U32 PL_charclass[] = {
1028# include "l1_char_class_tab.h"
1029};
1030
1031# else /* ! DOINIT */
1032EXTCONST U32 PL_charclass[];
1033# endif
1034
265c1f46 1035 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1036# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1037
1038 /* For internal core Perl use only: the base macro for defining macros like
1039 * isALPHA */
ff7ecfc3 1040# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1041 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1042
f4cdb42c
KW
1043 /* The mask for the _A versions of the macros; it just adds in the bit for
1044 * ASCII. */
1045# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1046
4650c663
KW
1047 /* For internal core Perl use only: the base macro for defining macros like
1048 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1049 * the ASCII bit are present */
3ded5eb0 1050# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
f4cd282c 1051 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
f4cdb42c
KW
1052 == _CC_mask_A(classnum)))
1053
f4cdb42c 1054# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1055# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1056# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1057# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1058# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1059# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1060# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1061# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1062# define isPSXSPC_A(c) _generic_isCC_A(c, _CC_PSXSPC)
1063# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1064# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1065# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1066# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
96ca48da 1067# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits */
d95f8b6a 1068# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1069# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1070# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1071# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1072
1073 /* continuation character for legal NAME in \N{NAME} */
1074# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1075
1076# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1077# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1078# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1079# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1080# define isPSXSPC_L1(c) _generic_isCC(c, _CC_PSXSPC)
1081# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1082# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1083# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1084# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1085# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1086
0852beac
KW
1087# ifdef EBCDIC
1088# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1089# endif
1090
f12c0118
KW
1091 /* Participates in a single-character fold with a character above 255 */
1092# 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)))
1093
1094 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1095# 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 1096
4eeeb416 1097# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd
KW
1098# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1099 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926
KW
1100# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1101 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1102# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1103 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1104#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1105
1106 /* If we don't have perl.h, we are compiling a utility program. Below we
1107 * hard-code various macro definitions that wouldn't otherwise be available
0852beac
KW
1108 * to it. Most are coded based on first principals. First some ones common
1109 * to both ASCII and EBCDIC */
1110# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1111# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1112# define isSPACE_A(c) (isBLANK_A(c) \
1113 || (c) == '\n' \
1114 || (c) == '\r' \
1115 || (c) == '\v' \
1116 || (c) == '\f')
1117# ifdef EBCDIC /* There are gaps between 'i' and 'j'; 'r' and 's'. Same
1118 for uppercase. This is ordered to exclude most things
1119 early */
1120# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1121 && ((c) <= 'i' \
1122 || ((c) >= 'j' && (c) <= 'r') \
1123 || (c) >= 's'))
1124# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1125 && ((c) <= 'I' \
1126 || ((c) >= 'J' && (c) <= 'R') \
1127 || (c) >= 'S'))
1128# else /* ASCII platform. */
9fb1bf9d 1129# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z')
2181da29 1130# define isUPPER_A(c) ((c) <= 'Z' && (c) >= 'A')
0852beac
KW
1131# endif
1132
1133 /* Some more ASCII, non-ASCII common definitions */
a4d7a999
KW
1134# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1135# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1136# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac
KW
1137# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1138# define isXDIGIT_A(c) (isDIGIT_A(c) \
1139 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1140 || ((c) <= 'F' && (c) >= 'A'))
3ded5eb0 1141
0852beac
KW
1142# ifdef EBCDIC
1143# define isPUNCT_A(c) ((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) == '{' || (c) == '|' \
1153 || (c) == '}' || (c) == '~')
1154# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1155# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1156
1157# ifdef QUESTION_MARK_CTRL
1158# define _isQMC(c) ((c) == QUESTION_MARK_CTRL)
1159# else
1160# define _isQMC(c) 0
1161# endif
1162
1163 /* I (khw) can't think of a way to define all the ASCII controls
1164 * without resorting to a libc (locale-sensitive) call. But we know
1165 * that all controls but the question-mark one are in the range 0-0x3f.
1166 * This makes sure that all the controls that have names are included,
1167 * and all controls that are also considered ASCII in the locale. This
1168 * may include more or fewer than what it actually should, but the
1169 * wrong ones are less-important controls, so likely won't impact
1170 * things (keep in mind that this is compiled only if perl.h isn't
1171 * available). The question mark control is included if available */
1172# define isCNTRL_A(c) (((c) < 0x40 && isascii(c)) \
1173 || (c) == '\0' || (c) == '\a' || (c) == '\b' \
1174 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1175 || (c) == '\t' || (c) == '\v' || _isQMC(c))
1176
1177# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1178# else /* ASCII platform; things are simpler, and isASCII has already
1179 been defined */
1180# define isGRAPH_A(c) (((c) > ' ' && (c) < 127))
1181# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1182# define isPUNCT_A(c) (isGRAPH_A(c) && (! isALPHANUMERIC_A(c)))
1183# define isCNTRL_A(c) (isASCII(c) && (! isPRINT_A(c)))
1184# endif
1185
3ffc8c70 1186 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1187 * during debugging, and it seems best to keep them. We may be called
1188 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1189 * do anything anyway, so make it not a problem */
1190# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1191# define NATIVE_TO_LATIN1(ch) (ch)
1192# endif
3ded5eb0
KW
1193# define isPSXSPC_A(c) isSPACE_A(c) /* XXX Assumes SPACE matches '\v' */
1194# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1195# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1196# define isBLANK_L1(c) (isBLANK_A(c) \
1197 || (FITS_IN_8_BITS(c) \
1198 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1199# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1200# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1201# define isLOWER_L1(c) (isLOWER_A(c) \
1202 || (FITS_IN_8_BITS(c) \
1203 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1204 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1205 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1206 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1207 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1208# define isPRINT_L1(c) (isPRINT_A(c) \
1209 || (FITS_IN_8_BITS(c) \
1210 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1211# define isPSXSPC_L1(c) isSPACE_L1(c)
1212# define isPUNCT_L1(c) (isPUNCT_A(c) \
1213 || (FITS_IN_8_BITS(c) \
1214 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1215 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1216 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1217 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1218 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1219 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1220 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1221# define isSPACE_L1(c) (isSPACE_A(c) \
1222 || (FITS_IN_8_BITS(c) \
1223 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1224 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1225# define isUPPER_L1(c) (isUPPER_A(c) \
1226 || (FITS_IN_8_BITS(c) \
1227 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1228 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1229 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1230# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1231# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1232# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1233 || isBLANK_L1(c) \
1234 || (c) == '-' \
1235 || (c) == '(' \
1236 || (c) == ')')
1237 /* The following are not fully accurate in the above-ASCII range. I (khw)
1238 * don't think it's necessary to be so for the purposes where this gets
1239 * compiled */
1240# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1241# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1242
1243 /* And these aren't accurate at all. They are useful only for above
1244 * Latin1, which utilities and bootstrapping don't deal with */
1245# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1246# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1247# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1248
1249 /* Many of the macros later in this file are defined in terms of these. By
1250 * implementing them with a function, which converts the class number into
1251 * a call to the desired macro, all of the later ones work. However, that
1252 * function won't be actually defined when building a utility program (no
1253 * perl.h), and so a compiler error will be generated if one is attempted
1254 * to be used. And the above-Latin1 code points require Unicode tables to
1255 * be present, something unlikely to be the case when bootstrapping */
1256# define _generic_isCC(c, classnum) \
1257 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1258# define _generic_isCC_A(c, classnum) \
1259 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1260#endif /* End of no perl.h H_PERL */
8a58bdcf 1261
e66b99e9
KW
1262#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1263#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1264#define isASCII_A(c) isASCII(c)
1265#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1266#define isBLANK(c) isBLANK_A(c)
1267#define isCNTRL(c) isCNTRL_A(c)
1268#define isDIGIT(c) isDIGIT_A(c)
1269#define isGRAPH(c) isGRAPH_A(c)
1270#define isIDFIRST(c) isIDFIRST_A(c)
1271#define isLOWER(c) isLOWER_A(c)
1272#define isPRINT(c) isPRINT_A(c)
1273#define isPSXSPC(c) isPSXSPC_A(c)
1274#define isPUNCT(c) isPUNCT_A(c)
1275#define isSPACE(c) isSPACE_A(c)
1276#define isUPPER(c) isUPPER_A(c)
1277#define isWORDCHAR(c) isWORDCHAR_A(c)
1278#define isXDIGIT(c) isXDIGIT_A(c)
1279
1280/* ASCII casing. These could also be written as
1281 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1282 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1283 which uses table lookup and mask instead of subtraction. (This would
1284 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1285#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1286#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1287
25200305
KW
1288/* In the ASCII range, these are equivalent to what they're here defined to be.
1289 * But by creating these definitions, other code doesn't have to be aware of
1290 * this detail */
1291#define toFOLD(c) toLOWER(c)
25200305
KW
1292#define toTITLE(c) toUPPER(c)
1293
c753c8d3
KW
1294#define toLOWER_A(c) toLOWER(c)
1295#define toUPPER_A(c) toUPPER(c)
25200305
KW
1296#define toFOLD_A(c) toFOLD(c)
1297#define toTITLE_A(c) toTITLE(c)
1a0901db 1298
4650c663 1299/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1300#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1301 ? (c) \
f4cd282c 1302 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1303#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1304
1a0901db 1305/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1306 * all mapped to one of them, and these need special handling; returns the
1307 * input itself if is out-of-range */
b2bf251f 1308#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1309 ? (c) \
f4cd282c 1310 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1311#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1312
beab9ebe
KW
1313/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1314
1315/* For internal core Perl use only: the base macro for defining macros like
1316 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1317 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1318 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1319 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1320 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1321 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1322 * code points under locale rules, use the macros based on _generic_LC_uvchr
1323 * instead of this. */
beab9ebe
KW
1324#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1325 (! FITS_IN_8_BITS(c) \
1326 ? 0 \
31f05a37
KW
1327 : IN_UTF8_CTYPE_LOCALE \
1328 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1329 : cBOOL(non_utf8))
1330
1331/* For internal core Perl use only: a helper macro for defining macros like
1332 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1333 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1334 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1335 * _generic_LC_base for more info */
1336#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1337 _generic_LC_base(c,utf8_locale_classnum, \
1338 non_utf8_func( (_LC_CAST) (c)))
1339
1340/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1341 * 'c' is the platform's native underscore character */
1342#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1343 _generic_LC_base(c, utf8_locale_classnum, \
1344 (non_utf8_func( (_LC_CAST) (c)) \
1345 || (char)(c) == '_'))
1346
1347/* These next three are also for internal core Perl use only: case-change
1348 * helper macros */
1349#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1350 ? (c) \
31f05a37
KW
1351 : (IN_UTF8_CTYPE_LOCALE) \
1352 ? PL_latin1_lc[ (U8) (c) ] \
beab9ebe
KW
1353 : function((cast)(c)))
1354
31f05a37
KW
1355/* Note that the result can be larger than a byte in a UTF-8 locale. It
1356 * returns a single value, so can't adequately return the upper case of LATIN
1357 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1358 * values "SS"); instead it asserts against that under DEBUGGING, and
1359 * otherwise returns its input */
beab9ebe
KW
1360#define _generic_toUPPER_LC(c, function, cast) \
1361 (! FITS_IN_8_BITS(c) \
1362 ? (c) \
31f05a37
KW
1363 : ((! IN_UTF8_CTYPE_LOCALE) \
1364 ? function((cast)(c)) \
1365 : ((((U8)(c)) == MICRO_SIGN) \
1366 ? GREEK_CAPITAL_LETTER_MU \
1367 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1368 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1369 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1370 ? (__ASSERT_(0) (c)) \
1371 : PL_mod_latin1_uc[ (U8) (c) ])))))
1372
1373/* Note that the result can be larger than a byte in a UTF-8 locale. It
1374 * returns a single value, so can't adequately return the fold case of LATIN
1375 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1376 * values "ss"); instead it asserts against that under DEBUGGING, and
1377 * otherwise returns its input */
beab9ebe 1378#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1379 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1380 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1381 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1382 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1383 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1384
84061b6a 1385/* Use the libc versions for these if available. */
f05550c0 1386#if defined(HAS_ISASCII)
84061b6a
KW
1387# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1388#else
1389# define isASCII_LC(c) isASCII(c)
1390#endif
1391
f05550c0 1392#if defined(HAS_ISBLANK)
84061b6a 1393# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f
KW
1394#else /* Unlike isASCII, varies if in a UTF-8 locale */
1395# define isBLANK_LC(c) (IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c)
84061b6a
KW
1396#endif
1397
f05550c0 1398#define _LC_CAST U8
bbce6d69 1399
f05550c0 1400#ifdef WIN32
375f5f06
KW
1401 /* The Windows functions don't bother to follow the POSIX standard, which
1402 * for example says that something can't both be a printable and a control.
1403 * But Windows treats the \t control as a printable, and does such things
1404 * as making superscripts into both digits and punctuation. This tames
1405 * these flaws by assuming that the definitions of both controls and space
1406 * are correct, and then making sure that other definitions don't have
1407 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1408 * Not all possible weirdnesses are checked for, just the ones that were
1409 * detected on actual Microsoft code pages */
1410
f05550c0
BF
1411# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1412# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1413
1414# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) && isALPHANUMERIC_LC(c))
1415# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && ! isPUNCT_LC(c))
1416# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && isALPHANUMERIC_LC(c))
1417# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1418# define isIDFIRST_LC(c) (((c) == '_') || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1419# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1420# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1421# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1422# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1423# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1424# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) && isALPHANUMERIC_LC(c))
1425
1426# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1427# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1428# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1429
1430#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1431 /* For most other platforms */
beab9ebe 1432
f05550c0
BF
1433# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1434# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1435# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1436# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1437# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1438# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1439# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1440# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1441# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1442# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1443# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1444# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1445# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1446
1447
1448# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1449# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1450# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1451
1452#else /* The final fallback position */
1453
1454# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1455# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1456# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1457# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1458# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1459# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1460# define isLOWER_LC(c) (isascii(c) && islower(c))
1461# define isPRINT_LC(c) (isascii(c) && isprint(c))
1462# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1463# define isSPACE_LC(c) (isascii(c) && isspace(c))
1464# define isUPPER_LC(c) (isascii(c) && isupper(c))
1465# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1466# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1467
1468# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1469# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1470# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1471
f05550c0 1472#endif
55204971 1473
eba68aa0
KW
1474#define isIDCONT(c) isWORDCHAR(c)
1475#define isIDCONT_A(c) isWORDCHAR_A(c)
1476#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1477#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1478#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1479
4650c663
KW
1480/* For internal core Perl use only: the base macros for defining macros like
1481 * isALPHA_uni. 'c' is the code point to check. 'classnum' is the POSIX class
1482 * number defined earlier in this file. _generic_uni() is used for POSIX
1483 * classes where there is a macro or function 'above_latin1' that takes the
1484 * single argument 'c' and returns the desired value. These exist for those
1485 * classes which have simple definitions, avoiding the overhead of a hash
1486 * lookup or inversion list binary search. _generic_swash_uni() can be used
1487 * for classes where that overhead is faster than a direct lookup.
1488 * _generic_uni() won't compile if 'c' isn't unsigned, as it won't match the
1489 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1490 * have duplicate checks here, so could create versions of the macros that
1491 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1492
1493/* Note that all ignore 'use bytes' */
cd500f2f
KW
1494#define _generic_uni(classnum, above_latin1, c) ((c) < 256 \
1495 ? _generic_isCC(c, classnum) \
1496 : above_latin1(c))
1497#define _generic_swash_uni(classnum, c) ((c) < 256 \
1498 ? _generic_isCC(c, classnum) \
922e8cb4 1499 : _is_uni_FOO(classnum, c))
cd500f2f
KW
1500#define isALPHA_uni(c) _generic_swash_uni(_CC_ALPHA, c)
1501#define isALPHANUMERIC_uni(c) _generic_swash_uni(_CC_ALPHANUMERIC, c)
509fb054
KW
1502#define isASCII_uni(c) isASCII(c)
1503#define isBLANK_uni(c) _generic_uni(_CC_BLANK, is_HORIZWS_cp_high, c)
1504#define isCNTRL_uni(c) isCNTRL_L1(c) /* All controls are in Latin1 */
cd500f2f
KW
1505#define isDIGIT_uni(c) _generic_swash_uni(_CC_DIGIT, c)
1506#define isGRAPH_uni(c) _generic_swash_uni(_CC_GRAPH, c)
eba68aa0 1507#define isIDCONT_uni(c) _generic_uni(_CC_WORDCHAR, _is_uni_perl_idcont, c)
509fb054 1508#define isIDFIRST_uni(c) _generic_uni(_CC_IDFIRST, _is_uni_perl_idstart, c)
cd500f2f
KW
1509#define isLOWER_uni(c) _generic_swash_uni(_CC_LOWER, c)
1510#define isPRINT_uni(c) _generic_swash_uni(_CC_PRINT, c)
aea0c08c 1511
509fb054
KW
1512/* Posix and regular space are identical above Latin1 */
1513#define isPSXSPC_uni(c) _generic_uni(_CC_PSXSPC, is_XPERLSPACE_cp_high, c)
b18192e9 1514
cd500f2f 1515#define isPUNCT_uni(c) _generic_swash_uni(_CC_PUNCT, c)
509fb054 1516#define isSPACE_uni(c) _generic_uni(_CC_SPACE, is_XPERLSPACE_cp_high, c)
cd500f2f 1517#define isUPPER_uni(c) _generic_swash_uni(_CC_UPPER, c)
509fb054 1518#define isVERTWS_uni(c) _generic_uni(_CC_VERTSPACE, is_VERTWS_cp_high, c)
cd500f2f 1519#define isWORDCHAR_uni(c) _generic_swash_uni(_CC_WORDCHAR, c)
509fb054 1520#define isXDIGIT_uni(c) _generic_uni(_CC_XDIGIT, is_XDIGIT_cp_high, c)
b18192e9 1521
b8d68ded 1522#define toFOLD_uni(c,s,l) to_uni_fold(c,s,l)
509fb054
KW
1523#define toLOWER_uni(c,s,l) to_uni_lower(c,s,l)
1524#define toTITLE_uni(c,s,l) to_uni_title(c,s,l)
1525#define toUPPER_uni(c,s,l) to_uni_upper(c,s,l)
a0ed51b3 1526
4650c663
KW
1527/* For internal core Perl use only: the base macros for defining macros like
1528 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1529 * point, not just 0-255. Like _generic_uni, there are two versions, one for
1530 * simple class definitions; the other for more complex. These are like
1531 * _generic_uni, so see it for more info. */
cd500f2f
KW
1532#define _generic_LC_uvchr(latin1, above_latin1, c) \
1533 (c < 256 ? latin1(c) : above_latin1(c))
1534#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1535 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1536
1537#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1538#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1539 _CC_ALPHANUMERIC, c)
509fb054 1540#define isASCII_LC_uvchr(c) isASCII_LC(c)
cd500f2f 1541#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, is_HORIZWS_cp_high, c)
feeab5a9 1542#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1543#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1544#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
eba68aa0
KW
1545#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
1546 _is_uni_perl_idcont, c)
cd500f2f
KW
1547#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
1548 _is_uni_perl_idstart, c)
1549#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1550#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
509fb054
KW
1551#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c) /* space is identical to posix
1552 space under locale */
cd500f2f
KW
1553#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
1554#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1555 is_XPERLSPACE_cp_high, c)
cd500f2f
KW
1556#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
1557#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
1558 _CC_WORDCHAR, c)
1559#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, is_XDIGIT_cp_high, c)
e712593e 1560
36b00f93 1561#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1562
4650c663
KW
1563/* For internal core Perl use only: the base macros for defining macros like
1564 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1565 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1566 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1567 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1568 * same representation whether utf8 or not. Note that it assumes that the utf8
1569 * has been validated, and ignores 'use bytes' */
cd500f2f 1570#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1571 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1572 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1573 ? _generic_isCC( \
94bb8c36 1574 TWO_BYTE_UTF8_TO_NATIVE(*(p), \
aea0c08c 1575 *((p)+1 )), \
94bb8c36 1576 classnum) \
252810d0 1577 : utf8)
cd500f2f
KW
1578/* Like the above, but calls 'above_latin1(p)' to get the utf8 value. 'above_latin1'
1579 * can be a macro */
1580#define _generic_func_utf8(classnum, above_latin1, p) \
1581 _generic_utf8(classnum, p, above_latin1(p))
4650c663 1582/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1583 * 'above_latin1' parameter */
922e8cb4 1584#define _generic_swash_utf8(classnum, p) \
cd500f2f 1585 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
922e8cb4 1586
cc8ab7c0 1587/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1588 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1589 * class is TRUE for. Hence it can skip the tests for this range.
1590 * 'above_latin1' should include its arguments */
3ded5eb0 1591#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1592 (UTF8_IS_INVARIANT(*(p)) \
1593 ? _generic_isCC(*(p), classnum) \
1594 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1595 ? above_latin1 \
cc8ab7c0
KW
1596 : 0)
1597
84238efa
KW
1598/* NOTE that some of these macros have very similar ones in regcharclass.h.
1599 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1600 * there, differing in name only by an underscore from the one here
1601 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1602 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1603 * points; the regcharclass.h ones are implemented as a series of
1604 * "if-else-if-else ..." */
1605
922e8cb4
KW
1606#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1607#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
509fb054
KW
1608#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
1609 utf8, the non-utf8 macro works
1610 */
cd500f2f 1611#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2
KW
1612
1613#ifdef EBCDIC
1614 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1615 * more efficient macro instead of the more general one */
1616# define isCNTRL_utf8(p) isCNTRL_L1(p)
1617#else
1618# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1619#endif
1620
cd500f2f
KW
1621#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1622 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1623#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1624#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1625 _is_utf8_perl_idcont, p)
e5dcd934 1626
c11ff943
KW
1627/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1628 * IDFIRST is an alnum. See
c8362b00 1629 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1630 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1631 * This used to be not the XID version, but we decided to go with the more
1632 * modern Unicode definition */
cd500f2f 1633#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1634 _is_utf8_perl_idstart, p)
e5dcd934 1635
922e8cb4
KW
1636#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1637#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
aea0c08c 1638
509fb054 1639/* Posix and regular space are identical above Latin1 */
cd500f2f 1640#define isPSXSPC_utf8(p) _generic_func_utf8(_CC_PSXSPC, is_XPERLSPACE_high, p)
aea0c08c 1641
922e8cb4 1642#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
cd500f2f 1643#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
922e8cb4 1644#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
cd500f2f 1645#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
922e8cb4 1646#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
cd500f2f
KW
1647#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
1648 is_XDIGIT_high(p))
a0ed51b3 1649
7d2b61dd 1650#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1651#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1652#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1653#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1654
4650c663
KW
1655/* For internal core Perl use only: the base macros for defining macros like
1656 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1657 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1658 * 'macro' parameter */
94bb8c36
KW
1659#define _generic_LC_utf8(macro, p, utf8) \
1660 (UTF8_IS_INVARIANT(*(p)) \
1661 ? macro(*(p)) \
1662 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1663 ? macro(TWO_BYTE_UTF8_TO_NATIVE(*(p), *((p)+1))) \
9201a572
KW
1664 : utf8)
1665
cd500f2f
KW
1666#define _generic_LC_swash_utf8(macro, classnum, p) \
1667 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1668#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1669 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1670
cd500f2f
KW
1671#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
1672 _CC_ALPHANUMERIC, p)
1673#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
509fb054 1674#define isASCII_LC_utf8(p) isASCII_LC(*p)
cd500f2f
KW
1675#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, is_HORIZWS_high, p)
1676#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1677#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1678#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
eba68aa0 1679#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, _is_utf8_perl_idcont, p)
cd500f2f
KW
1680#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, _is_utf8_perl_idstart, p)
1681#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1682#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
509fb054
KW
1683#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p) /* space is identical to posix
1684 space under locale */
cd500f2f
KW
1685#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1686#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, is_XPERLSPACE_high, p)
1687#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
1688#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1689 _CC_WORDCHAR, p)
1690#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, is_XDIGIT_high, p)
aaa51d5e 1691
fbc19f27
KW
1692/* Macros for backwards compatibility and for completeness when the ASCII and
1693 * Latin1 values are identical */
fbc19f27
KW
1694#define isALPHAU(c) isALPHA_L1(c)
1695#define isDIGIT_L1(c) isDIGIT_A(c)
1696#define isOCTAL(c) isOCTAL_A(c)
1697#define isOCTAL_L1(c) isOCTAL_A(c)
1698#define isXDIGIT_L1(c) isXDIGIT_A(c)
2e28f0b9
KW
1699#define isALNUM(c) isWORDCHAR(c)
1700#define isALNUMU(c) isWORDCHAR_L1(c)
1701#define isALNUM_LC(c) isWORDCHAR_LC(c)
1702#define isALNUM_uni(c) isWORDCHAR_uni(c)
1703#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
1704#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1705#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
15861f94
KW
1706#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1707#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1708#define isALNUMC(c) isALPHANUMERIC(c)
1709#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1710#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
1711#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
1712#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
1713#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1714
2bd1cbf6
KW
1715/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1716 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1717 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1718 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1719 * the outlier from the block that contains the other controls, just like
1720 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1721 * block. If it weren't special cased, it would yield a non-control.
1722 * The conversion works both ways, so CTRL('D') is 4, and CTRL(4) is D, etc. */
1723#ifndef EBCDIC
1724# define toCTRL(c) (toUPPER(c) ^ 64)
1725#else
0ed2b00b
KW
1726# define toCTRL(c) ((isPRINT_A(c)) \
1727 ? UNLIKELY((c) == '?') \
1728 ? QUESTION_MARK_CTRL \
1729 : (NATIVE_TO_LATIN1(toUPPER(c)) ^ 64) \
1730 : UNLIKELY((c) == QUESTION_MARK_CTRL) \
1731 ? ((c) == '?') \
1732 : (LATIN1_TO_NATIVE((c) ^ 64)))
2bd1cbf6 1733#endif
bbce6d69 1734
dea28490
JJ
1735/* Line numbers are unsigned, 32 bits. */
1736typedef U32 line_t;
e5dcd934 1737#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 1738
91152fc1
DG
1739/* Helpful alias for version prescan */
1740#define is_LAX_VERSION(a,b) \
1741 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1742
1743#define is_STRICT_VERSION(a,b) \
1744 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1745
1746#define BADVERSION(a,b,c) \
1747 if (b) { \
1748 *b = c; \
1749 } \
1750 return a;
8c52afec 1751
cb27eebd
KW
1752/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1753 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1754 * which is advanced. The input is validated only by an assert() in DEBUGGING
1755 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1756 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1757#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1758 ? (c) \
1759 : ((c) + 9))))
902008b8
KW
1760#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1761 ? (*(s)++) \
1762 : (*(s)++ + 9))))
95a59cab 1763
cb27eebd
KW
1764/* Converts a character known to represent an octal digit (0-7) to its numeric
1765 * value. The input is validated only by an assert() in DEBUGGING builds. In
1766 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1767#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1768
305b8651
KW
1769/* Efficiently returns a boolean as to if two native characters are equivalent
1770 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1771 * the ALPHA in the name is to remind you of that. This is asserted() in
1772 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1773 * works (on valid input) for both non- and UTF-8-encoded bytes.
1774 *
1775 * When one of the inputs is a compile-time constant and gets folded by the
1776 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1777 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1778 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
1779 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1780 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1781 * just a single 0, in the bit position where the upper- and lowercase differ.
1782 * */
305b8651
KW
1783#define isALPHA_FOLD_EQ(c1, c2) \
1784 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1785 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1786#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1787
8e84507e 1788/*
ccfc67b7
JH
1789=head1 Memory Management
1790
a02a5408 1791=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
1792The XSUB-writer's interface to the C C<malloc> function.
1793
0d7b2759
MH
1794Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1795
c5008215
JC
1796In 5.9.3, Newx() and friends replace the older New() API, and drops
1797the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 1798themselves. This aid has been superseded by a new build option,
d10b4965 1799PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
1800there for use in XS modules supporting older perls.
1801
a02a5408 1802=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 1803The XSUB-writer's interface to the C C<malloc> function, with
c5008215 1804cast. See also C<Newx>.
954c1994 1805
0d7b2759
MH
1806Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1807
a02a5408 1808=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 1809The XSUB-writer's interface to the C C<malloc> function. The allocated
c5008215 1810memory is zeroed with C<memzero>. See also C<Newx>.
a02a5408 1811
0d7b2759
MH
1812Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1813
954c1994
GS
1814=for apidoc Am|void|Renew|void* ptr|int nitems|type
1815The XSUB-writer's interface to the C C<realloc> function.
1816
0d7b2759
MH
1817Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1818
954c1994
GS
1819=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1820The XSUB-writer's interface to the C C<realloc> function, with
1821cast.
1822
0d7b2759
MH
1823Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1824
49b8b560 1825=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
1826The XSUB-writer's interface to the C C<free> function.
1827
0d7b2759
MH
1828This should B<ONLY> be used on memory obtained using L<"Newx"> and friends.
1829
954c1994
GS
1830=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1831The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c
KW
1832source, C<dest> is the destination, C<nitems> is the number of items, and
1833C<type> is the type. Can do overlapping moves. See also C<Copy>.
954c1994 1834
e90e2364 1835=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
72d33970
FC
1836Like C<Move> but returns dest. Useful
1837for encouraging compilers to tail-call
e90e2364
NC
1838optimise.
1839
954c1994
GS
1840=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1841The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c
KW
1842source, C<dest> is the destination, C<nitems> is the number of items, and
1843C<type> is the type. May fail on overlapping copies. See also C<Move>.
954c1994 1844
e90e2364
NC
1845=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1846
72d33970
FC
1847Like C<Copy> but returns dest. Useful
1848for encouraging compilers to tail-call
e90e2364
NC
1849optimise.
1850
954c1994
GS
1851=for apidoc Am|void|Zero|void* dest|int nitems|type
1852
1853The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1854destination, C<nitems> is the number of items, and C<type> is the type.
1855
e90e2364
NC
1856=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1857
72d33970
FC
1858Like C<Zero> but returns dest. Useful
1859for encouraging compilers to tail-call
e90e2364
NC
1860optimise.
1861
da5d8dbb 1862=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 1863This is an architecture-independent macro to copy one structure to another.
954c1994 1864
7e337ee0
JH
1865=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1866
1867Fill up memory with a byte pattern (a byte repeated over and over
1868again) that hopefully catches attempts to access uninitialized memory.
1869
1870=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1871
1872PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1873
1c12ffb4 1874=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
1875
1876PoisonWith(0xEF) for catching access to freed memory.
1877
9965345d
JH
1878=for apidoc Am|void|Poison|void* dest|int nitems|type
1879
7e337ee0 1880PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
1881
1882=cut */
954c1994 1883
561b68a9
SH
1884/* Maintained for backwards-compatibility only. Use newSV() instead. */
1885#ifndef PERL_CORE
ff06c60c 1886#define NEWSV(x,len) newSV(len)
561b68a9 1887#endif
ff06c60c 1888
19a94d75
JH
1889#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1890
2720f793
JH
1891/* The +0.0 in MEM_WRAP_CHECK_ is an attempt to foil
1892 * overly eager compilers that will bleat about e.g.
1893 * (U16)n > (size_t)~0/sizeof(U16) always being false. */
27d5b266 1894#ifdef PERL_MALLOC_WRAP
f019c49e 1895#define MEM_WRAP_CHECK(n,t) \
5d9574c1 1896 (void)(UNLIKELY(sizeof(t) > 1 && ((MEM_SIZE)(n)+0.0) > MEM_SIZE_MAX/sizeof(t)) && (croak_memory_wrap(),0))
27d5b266 1897#define MEM_WRAP_CHECK_1(n,t,a) \
5d9574c1 1898 (void)(UNLIKELY(sizeof(t) > 1 && ((MEM_SIZE)(n)+0.0) > MEM_SIZE_MAX/sizeof(t)) && (Perl_croak_nocontext("%s",(a)),0))
8b44ba4c 1899#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 1900
d1decf2b 1901#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
1902#else
1903
410319be
NC
1904#define MEM_WRAP_CHECK(n,t)
1905#define MEM_WRAP_CHECK_1(n,t,a)
1906#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
1907#define MEM_WRAP_CHECK_(n,t)
1908
6fff79ce 1909#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 1910
1936d2a7 1911#endif
8b44ba4c 1912
fe4f188c 1913#ifdef PERL_MEM_LOG
46c6c7e2 1914/*
9f653bb5 1915 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
1916 * go through functions, which are handy for debugging breakpoints, but
1917 * which more importantly get the immediate calling environment (file and
e352bcff
JH
1918 * line number, and C function name if available) passed in. This info can
1919 * then be used for logging the calls, for which one gets a sample
73d1d973 1920 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 1921 *
46c6c7e2 1922 * Known problems:
94e892a6 1923 * - not all memory allocs get logged, only those
46c6c7e2 1924 * that go through Newx() and derivatives (while all
94e892a6 1925 * Safefrees do get logged)
46c6c7e2
JH
1926 * - __FILE__ and __LINE__ do not work everywhere
1927 * - __func__ or __FUNCTION__ even less so
1928 * - I think more goes on after the perlio frees but
1929 * the thing is that STDERR gets closed (as do all
1930 * the file descriptors)
1931 * - no deeper calling stack than the caller of the Newx()
1932 * or the kind, but do I look like a C reflection/introspection
1933 * utility to you?
1934 * - the function prototypes for the logging functions
1935 * probably should maybe be somewhere else than handy.h
1936 * - one could consider inlining (macrofying) the logging
1937 * for speed, but I am too lazy
1938 * - one could imagine recording the allocations in a hash,
1939 * (keyed by the allocation address?), and maintain that
1940 * through reallocs and frees, but how to do that without
1941 * any News() happening...?
73d1d973 1942 * - lots of -Ddefines to get useful/controllable output
b953482e 1943 * - lots of ENV reads
46c6c7e2
JH
1944 */
1945
12754f92 1946PERL_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 1947
12754f92 1948PERL_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 1949
12754f92 1950PERL_EXPORT_C Malloc_t Perl_mem_log_free(Malloc_t oldalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 1951
0b0ab801 1952# ifdef PERL_CORE
73d1d973 1953# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
1954enum mem_log_type {
1955 MLT_ALLOC,
1956 MLT_REALLOC,
d7a2c63c
MHM
1957 MLT_FREE,
1958 MLT_NEW_SV,
1959 MLT_DEL_SV
0b0ab801
MHM
1960};
1961# endif
12754f92 1962# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
1963void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
1964void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 1965# endif
0b0ab801
MHM
1966# endif
1967
fe4f188c
JH
1968#endif
1969
1970#ifdef PERL_MEM_LOG
d1401ee9
MHM
1971#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
1972#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 1973#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
1974#endif
1975
1976#ifndef MEM_LOG_ALLOC
1977#define MEM_LOG_ALLOC(n,t,a) (a)
1978#endif
1979#ifndef MEM_LOG_REALLOC
1980#define MEM_LOG_REALLOC(n,t,v,a) (a)
1981#endif
1982#ifndef MEM_LOG_FREE
1983#define MEM_LOG_FREE(a) (a)
1984#endif
1985
d1401ee9
MHM
1986#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
1987#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
1988#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
1989
1990#ifndef PERL_CORE
a02a5408
JC
1991/* pre 5.9.x compatibility */
1992#define New(x,v,n,t) Newx(v,n,t)
1993#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 1994#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 1995#endif
a02a5408 1996
ff68c719 1997#define Renew(v,n,t) \
d1401ee9 1998 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 1999#define Renewc(v,n,t,c) \
d1401ee9 2000 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2001
2002#ifdef PERL_POISON
2003#define Safefree(d) \
06c0cc96 2004 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2005#else
fe4f188c 2006#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2007#endif
55497cff 2008
8b44ba4c
NC
2009#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2010#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2011#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2012
8b44ba4c
NC
2013#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2014#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2015#ifdef HAS_MEMSET
8b44ba4c 2016#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2017#else
8b44ba4c
NC
2018/* Using bzero(), which returns void. */
2019#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2020#endif
2021
7e337ee0
JH
2022#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2023#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2024#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2025#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2026
caa674f3
DD
2027#ifdef PERL_POISON
2028# define PERL_POISON_EXPR(x) x
2029#else
2030# define PERL_POISON_EXPR(x)
2031#endif
2032
2304df62 2033#ifdef USE_STRUCT_COPY
ff68c719 2034#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2035#else
2036#define StructCopy(s,d,t) Copy(s,d,1,t)
2037#endif
2cc61e15 2038
c3caa5c3
JH
2039/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2040 * want your zero-based indices to be less than but not equal to).
2041 *
2042 * C_ARRAY_END is one past the last: half-open/half-closed range,
2043 * not last-inclusive range. */
622913ab 2044#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2045#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2046
2cc61e15
DD
2047#ifdef NEED_VA_COPY
2048# ifdef va_copy
2049# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2050# else
a1866d1b
JH
2051# if defined(__va_copy)
2052# define Perl_va_copy(s, d) __va_copy(d, s)
2053# else
2054# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2055# endif
2cc61e15
DD
2056# endif
2057#endif
2058
472d47bc
SB
2059/* convenience debug macros */
2060#ifdef USE_ITHREADS
2061#define pTHX_FORMAT "Perl interpreter: 0x%p"
2062#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2063#define pTHX_VALUE_ (void *)my_perl,
2064#define pTHX_VALUE (void *)my_perl
2065#define pTHX__VALUE_ ,(void *)my_perl,
2066#define pTHX__VALUE ,(void *)my_perl
472d47bc 2067#else
3609ea0d 2068#define pTHX_FORMAT
472d47bc 2069#define pTHX__FORMAT
3609ea0d 2070#define pTHX_VALUE_
472d47bc 2071#define pTHX_VALUE
3609ea0d 2072#define pTHX__VALUE_
472d47bc
SB
2073#define pTHX__VALUE
2074#endif /* USE_ITHREADS */
3609ea0d 2075
2acdbac1
NC
2076/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2077 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2078 CPAN::Unpack show any users outside the core. */
2079#ifdef PERL_CORE
d1d15184 2080# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), "Use of " s " is deprecated")
2acdbac1
NC
2081#endif
2082
dfff4baf
BF
2083/* Internal macros to deal with gids and uids */
2084#ifdef PERL_CORE
2085
2086# if Uid_t_size > IVSIZE
2087# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2088# define SvUID(sv) SvNV(sv)
2089# else
2090# if Uid_t_sign <= 0
2091# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2092# define SvUID(sv) SvIV(sv)
2093# else
2094# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2095# define SvUID(sv) SvUV(sv)
2096# endif
2097# endif /* Uid_t_size */
2098
2099# if Gid_t_size > IVSIZE
2100# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2101# define SvGID(sv) SvNV(sv)
2102# else
2103# if Gid_t_sign <= 0
2104# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2105# define SvGID(sv) SvIV(sv)
2106# else
2107# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2108# define SvGID(sv) SvUV(sv)
2109# endif
2110# endif /* Gid_t_size */
2111
2112#endif
2113
9d745869
KW
2114#endif /* HANDY_H */
2115
e9a8c099
MHM
2116/*
2117 * Local variables:
2118 * c-indentation-style: bsd
2119 * c-basic-offset: 4
14d04a33 2120 * indent-tabs-mode: nil
e9a8c099
MHM
2121 * End:
2122 *
14d04a33 2123 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2124 */