This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Change EBCDIC macro definition
[perl5.git] / handy.h
CommitLineData
a0d0e21e 1/* handy.h
a687059c 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
da5d8dbb 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
a687059c 5 *
6e21c824
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8d063cd8 8 *
8d063cd8
LW
9 */
10
4650c663
KW
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
9d745869
KW
14#ifndef HANDY_H /* Guard against nested #inclusion */
15#define HANDY_H
16
85e6fe83 17#if !defined(__STDC__)
378cc40b
LW
18#ifdef NULL
19#undef NULL
20#endif
a687059c 21# define NULL 0
85e6fe83
LW
22#endif
23
24792b8d
NC
24#ifndef PERL_CORE
25# define Null(type) ((type)NULL)
954c1994
GS
26
27/*
ccfc67b7 28=head1 Handy Values
954c1994 29
3609ea0d 30=for apidoc AmU||Nullch
72d33970
FC
31Null character pointer. (No longer available when C<PERL_CORE> is
32defined.)
2307c6d0 33
954c1994 34=for apidoc AmU||Nullsv
72d33970 35Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
36
37=cut
38*/
39
24792b8d
NC
40# define Nullch Null(char*)
41# define Nullfp Null(PerlIO*)
42# define Nullsv Null(SV*)
43#endif
8d063cd8 44
641d3f0b
PP
45#ifdef TRUE
46#undef TRUE
47#endif
48#ifdef FALSE
49#undef FALSE
50#endif
51#define TRUE (1)
52#define FALSE (0)
53
cf3f0ffb
DM
54/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
55 * (compiler permitting) that casting away const-ness will give a warning;
56 * e.g.:
57 *
58 * const SV *sv = ...;
59 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
60 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
61 */
62
b1bc3f34 63#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
f065df04 64# define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
b1bc3f34
NC
65#else
66# define MUTABLE_PTR(p) ((void *) (p))
67#endif
68
a062e10d 69#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 70#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 71#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 72#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 73#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 74#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 75
f789f6a4 76#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
77# include <stdbool.h>
78# ifndef HAS_BOOL
79# define HAS_BOOL 1
80# endif
81#endif
82
8e84507e 83/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
84 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
85 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 86 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
87 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
88 g++ can be identified by __GNUG__.
89 Andy Dougherty February 2000
5d94fbed 90*/
3609ea0d 91#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 92# ifndef PERL_BOOL_AS_CHAR
5d94fbed 93# ifndef HAS_BOOL
c1d22f6b 94# define HAS_BOOL 1
5d94fbed 95# endif
f789f6a4 96# endif
5d94fbed
AD
97#endif
98
99#ifndef HAS_BOOL
f789f6a4
FC
100# ifdef bool
101# undef bool
102# endif
70d5cb32 103# define bool char
c1d22f6b 104# define HAS_BOOL 1
a687059c 105#endif
0d3e774c 106
236415e2
KW
107/* cast-to-bool. A simple (bool) cast may not do the right thing: if bool is
108 * defined as char for example, then the cast from int is
109 * implementation-defined (bool)!!(cbool) in a ternary triggers a bug in xlc on
110 * AIX */
18f5643b 111#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 112
46c6c7e2 113/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
114 * XXX Should really be a Configure probe, with HAS__FUNCTION__
115 * and FUNCTION__ as results.
116 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
46c6c7e2
JH
117#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
118# define FUNCTION__ __func__
119#else
9ea40801 120# if (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
be73ebb0 121 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
46c6c7e2
JH
122# define FUNCTION__ ""
123# else
124# define FUNCTION__ __FUNCTION__ /* Common extension. */
125# endif
126#endif
127
27d4fb96
PP
128/* XXX A note on the perl source internal type system. The
129 original intent was that I32 be *exactly* 32 bits.
130
131 Currently, we only guarantee that I32 is *at least* 32 bits.
132 Specifically, if int is 64 bits, then so is I32. (This is the case
133 for the Cray.) This has the advantage of meshing nicely with
134 standard library calls (where we pass an I32 and the library is
135 expecting an int), but the disadvantage that an I32 is not 32 bits.
136 Andy Dougherty August 1996
24fef2a7 137
dc45a647
MB
138 There is no guarantee that there is *any* integral type with
139 exactly 32 bits. It is perfectly legal for a system to have
140 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 141
dc45a647
MB
142 Similarly, there is no guarantee that I16 and U16 have exactly 16
143 bits.
693762b4 144
8e84507e
NIS
145 For dealing with issues that may arise from various 32/64-bit
146 systems, we will ask Configure to check out
8175356b 147
3609ea0d
JH
148 SHORTSIZE == sizeof(short)
149 INTSIZE == sizeof(int)
150 LONGSIZE == sizeof(long)
dc45a647 151 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 152 PTRSIZE == sizeof(void *)
dc45a647
MB
153 DOUBLESIZE == sizeof(double)
154 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 155
27d4fb96
PP
156*/
157
69512466
JH
158#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
159# include <inttypes.h>
dd0eed91
JH
160# ifdef INT32_MIN_BROKEN
161# undef INT32_MIN
162# define INT32_MIN (-2147483647-1)
163# endif
164# ifdef INT64_MIN_BROKEN
165# undef INT64_MIN
166# define INT64_MIN (-9223372036854775807LL-1)
167# endif
69512466
JH
168#endif
169
8175356b
JH
170typedef I8TYPE I8;
171typedef U8TYPE U8;
172typedef I16TYPE I16;
173typedef U16TYPE U16;
174typedef I32TYPE I32;
175typedef U32TYPE U32;
16d89be8
JH
176
177#ifdef HAS_QUAD
8175356b
JH
178typedef I64TYPE I64;
179typedef U64TYPE U64;
16d89be8 180#endif
8175356b 181
9c39b21c
JH
182/* INT64_C/UINT64_C are C99 from <stdint.h> (so they will not be
183 * available in strict C89 mode), but they are nice, so let's define
184 * them if necessary. */
0d3bd0a9 185#if defined(HAS_QUAD)
9c39b21c
JH
186# undef PeRl_INT64_C
187# undef PeRl_UINT64_C
188/* Prefer the native integer types (int and long) over long long
189 * (which is not C89) and Win32-specific __int64. */
190# if QUADKIND == QUAD_IS_INT && INTSIZE == 8
191# define PeRl_INT64_C(c) (c)
192# define PeRl_UINT64_C(c) CAT2(c,U)
193# endif
194# if QUADKIND == QUAD_IS_LONG && LONGSIZE == 8
195# define PeRl_INT64_C(c) CAT2(c,L)
196# define PeRl_UINT64_C(c) CAT2(c,UL)
197# endif
198# if QUADKIND == QUAD_IS_LONG_LONG && defined(HAS_LONG_LONG)
199# define PeRl_INT64_C(c) CAT2(c,LL)
200# define PeRl_UINT64_C(c) CAT2(c,ULL)
201# endif
202# if QUADKIND == QUAD_IS___INT64
203# define PeRl_INT64_C(c) CAT2(c,I64)
204# define PeRl_UINT64_C(c) CAT2(c,UI64)
205# endif
206# ifndef PeRl_INT64_C
702c92eb
JH
207# define PeRl_INT64_C(c) ((I64)(c)) /* last resort */
208# define PeRl_UINT64_C(c) ((U64)(c))
9c39b21c
JH
209# endif
210/* In OS X the INT64_C/UINT64_C are defined with LL/ULL, which will
211 * not fly with C89-pedantic gcc, so let's undefine them first so that
212 * we can redefine them with our native integer preferring versions. */
8eafc6eb 213# if defined(PERL_DARWIN) && defined(PERL_GCC_PEDANTIC)
9c39b21c
JH
214# undef INT64_C
215# undef UINT64_C
216# endif
217# ifndef INT64_C
218# define INT64_C(c) PeRl_INT64_C(c)
219# endif
220# ifndef UINT64_C
221# define UINT64_C(c) PeRl_UINT64_C(c)
222# endif
e8c95190 223#endif
dc750f2e 224
d8668976 225#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 226
5ff3f7a4
GS
227/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
228 Please search CHAR_MAX in perl.h for further details. */
229#define U8_MAX UINT8_MAX
230#define U8_MIN UINT8_MIN
231
5ff3f7a4
GS
232#define I16_MAX INT16_MAX
233#define I16_MIN INT16_MIN
234#define U16_MAX UINT16_MAX
235#define U16_MIN UINT16_MIN
236
5ff3f7a4
GS
237#define I32_MAX INT32_MAX
238#define I32_MIN INT32_MIN
0e983133
GS
239#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
240# define U32_MAX UINT32_MAX
241#else
242# define U32_MAX 4294967295U
243#endif
5ff3f7a4
GS
244#define U32_MIN UINT32_MIN
245
246#else
247
5c9fa16e
KA
248/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
249 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
250#define U8_MAX PERL_UCHAR_MAX
251#define U8_MIN PERL_UCHAR_MIN
79072805 252
27d4fb96
PP
253#define I16_MAX PERL_SHORT_MAX
254#define I16_MIN PERL_SHORT_MIN
255#define U16_MAX PERL_USHORT_MAX
256#define U16_MIN PERL_USHORT_MIN
79072805 257
c4f23d77 258#if LONGSIZE > 4
27d4fb96
PP
259# define I32_MAX PERL_INT_MAX
260# define I32_MIN PERL_INT_MIN
261# define U32_MAX PERL_UINT_MAX
262# define U32_MIN PERL_UINT_MIN
79072805 263#else
27d4fb96
PP
264# define I32_MAX PERL_LONG_MAX
265# define I32_MIN PERL_LONG_MIN
266# define U32_MAX PERL_ULONG_MAX
267# define U32_MIN PERL_ULONG_MIN
79072805
LW
268#endif
269
5ff3f7a4
GS
270#endif
271
58a9a5d5 272/* log(2) is pretty close to 0.30103, just in case anyone is grepping for it */
fc36a67e
PP
273#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
274#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
275#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
276
88794300 277/* Unused by core; should be deprecated */
ff68c719 278#define Ctl(ch) ((ch) & 037)
8d063cd8 279
4650c663
KW
280/* This is a helper macro to avoid preprocessor issues, replaced by nothing
281 * unless under DEBUGGING, where it expands to an assert of its argument,
282 * followed by a comma (hence the comma operator). If we just used a straight
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
bd18bd40
KW
451=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
452Test two buffers (which may contain embedded C<NUL> characters, to see if they
453are equal. The C<len> parameter indicates the number of bytes to compare.
454Returns zero if equal, or non-zero if non-equal.
455
456=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
457Test two buffers (which may contain embedded C<NUL> characters, to see if they
458are not equal. The C<len> parameter indicates the number of bytes to compare.
459Returns zero if non-equal, or non-zero if equal.
460
954c1994
GS
461=cut
462*/
463
8d063cd8
LW
464#define strNE(s1,s2) (strcmp(s1,s2))
465#define strEQ(s1,s2) (!strcmp(s1,s2))
466#define strLT(s1,s2) (strcmp(s1,s2) < 0)
467#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
468#define strGT(s1,s2) (strcmp(s1,s2) > 0)
469#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
470#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
471#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
378cc40b 472
36477c24
PP
473#ifdef HAS_MEMCMP
474# define memNE(s1,s2,l) (memcmp(s1,s2,l))
475# define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
476#else
477# define memNE(s1,s2,l) (bcmp(s1,s2,l))
478# define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
479#endif
480
568a785a 481#define memEQs(s1, l, s2) \
8e6497b6 482 (sizeof(s2)-1 == l && memEQ(s1, ("" s2 ""), (sizeof(s2)-1)))
568a785a
NC
483#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
484
bbce6d69
PP
485/*
486 * Character classes.
487 *
488 * Unfortunately, the introduction of locales means that we
489 * can't trust isupper(), etc. to tell the truth. And when
490 * it comes to /\w+/ with tainting enabled, we *must* be able
491 * to trust our character classes.
492 *
493 * Therefore, the default tests in the text of Perl will be
494 * independent of locale. Any code that wants to depend on
495 * the current locale will use the tests that begin with "lc".
496 */
497
2304df62
AD
498#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
499# ifndef CTYPE256
500# define CTYPE256
501# endif
502#endif
503
954c1994 504/*
ccfc67b7 505
dcccc8ff 506=head1 Character classification
243effed
KW
507This section is about functions (really macros) that classify characters
508into types, such as punctuation versus alphabetic, etc. Most of these are
509analogous to regular expression character classes. (See
510L<perlrecharclass/POSIX Character Classes>.) There are several variants for
511each class. (Not all macros have all variants; each item below lists the
512ones valid for it.) None are affected by C<use bytes>, and only the ones
513with C<LC> in the name are affected by the current locale.
514
515The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
516C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
517represented by that octet is (or on non-ASCII platforms, corresponds to) an
518ASCII character in the named class based on platform, Unicode, and Perl rules.
519If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 520
6aff1f14
KW
521Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
522with no suffix C<"_A">.
4b9734bf
KW
523
524Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
525onto the platform. That is, the code points that are ASCII are unaffected,
526since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
527as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
528true when called with the code point 0xDF, which is a word character in both
4650c663 529ASCII and EBCDIC (though it represents different characters in each).
243effed
KW
530
531Variant C<isFOO_uni> is like the C<isFOO_L1> variant, but accepts any UV code
532point as input. If the code point is larger than 255, Unicode rules are used
533to determine if it is in the character class. For example,
6aff1f14
KW
534C<isWORDCHAR_uni(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
535WITH MACRON in Unicode, and is a word character.
243effed
KW
536
537Variant C<isFOO_utf8> is like C<isFOO_uni>, but the input is a pointer to a
538(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
6aff1f14
KW
539classification of just the first (possibly multi-byte) character in the string
540is tested.
243effed 541
31f05a37
KW
542Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
543result is based on the current locale, which is what C<LC> in the name stands
544for. If Perl can determine that the current locale is a UTF-8 locale, it uses
545the published Unicode rules; otherwise, it uses the C library function that
546gives the named classification. For example, C<isDIGIT_LC()> when not in a
547UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
548returned if the input won't fit into an octet. On some platforms where the C
549library function is known to be defective, Perl changes its result to follow
550the POSIX standard's rules.
243effed
KW
551
552Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
553returns the same as C<isFOO_LC> for input code points less than 256, and
554returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
555
b7d90381
KW
556Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer
557to a (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
6aff1f14
KW
558classification of just the first (possibly multi-byte) character in the string
559is tested.
ccfc67b7 560
954c1994 561=for apidoc Am|bool|isALPHA|char ch
2787a470 562Returns a boolean indicating whether the specified character is an
6aff1f14 563alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
564See the L<top of this section|/Character classification> for an explanation of
565variants
e5ad6aba 566C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uni>, C<isALPHA_utf8>, C<isALPHA_LC>,
243effed 567C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
8a58bdcf 568
15861f94
KW
569=for apidoc Am|bool|isALPHANUMERIC|char ch
570Returns a boolean indicating whether the specified character is a either an
6aff1f14 571alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
572See the L<top of this section|/Character classification> for an explanation of
573variants
15861f94 574C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uni>,
e5ad6aba 575C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
15861f94
KW
576C<isALPHANUMERIC_LC_utf8>.
577
8a58bdcf
KW
578=for apidoc Am|bool|isASCII|char ch
579Returns a boolean indicating whether the specified character is one of the 128
243effed 580characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 581On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
582character corresponds to an ASCII character. Variants C<isASCII_A()> and
583C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
584See the L<top of this section|/Character classification> for an explanation of
585variants
243effed
KW
586C<isASCII_uni>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
587C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
588library routine C<isascii()>. In these cases, the variants whose names contain
589C<LC> are the same as the corresponding ones without.
590
d98532ea
KW
591Also note, that because all ASCII characters are UTF-8 invariant (meaning they
592have the exact same representation (always a single byte) whether encoded in
593UTF-8 or not), C<isASCII> will give the correct results when called with any
594byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
595work properly on any string encoded or not in UTF-8.
596
243effed
KW
597=for apidoc Am|bool|isBLANK|char ch
598Returns a boolean indicating whether the specified character is a
6aff1f14 599character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
600See the L<top of this section|/Character classification> for an explanation of
601variants
e5ad6aba 602C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uni>, C<isBLANK_utf8>, C<isBLANK_LC>,
243effed
KW
603C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
604platforms do not have the C library routine C<isblank()>. In these cases, the
605variants whose names contain C<LC> are the same as the corresponding ones
606without.
607
608=for apidoc Am|bool|isCNTRL|char ch
609Returns a boolean indicating whether the specified character is a
6aff1f14 610control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
611See the L<top of this section|/Character classification> for an explanation of
612variants
e5ad6aba 613C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uni>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
6aff1f14
KW
614C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
615On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 616
954c1994 617=for apidoc Am|bool|isDIGIT|char ch
2787a470 618Returns a boolean indicating whether the specified character is a
6aff1f14 619digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 620Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
621See the L<top of this section|/Character classification> for an explanation of
622variants
e5ad6aba 623C<isDIGIT_uni>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
243effed
KW
624C<isDIGIT_LC_utf8>.
625
626=for apidoc Am|bool|isGRAPH|char ch
627Returns a boolean indicating whether the specified character is a
6aff1f14 628graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff
KW
629See the L<top of this section|/Character classification> for an explanation of
630variants
e5ad6aba 631C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uni>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
243effed 632C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
954c1994 633
0c82b6df 634=for apidoc Am|bool|isLOWER|char ch
2787a470 635Returns a boolean indicating whether the specified character is a
6aff1f14 636lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
637See the L<top of this section|/Character classification> for an explanation of
638variants
e5ad6aba 639C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uni>, C<isLOWER_utf8>, C<isLOWER_LC>,
243effed 640C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
0c82b6df 641
c99e91e9 642=for apidoc Am|bool|isOCTAL|char ch
2787a470 643Returns a boolean indicating whether the specified character is an
6aff1f14 644octal digit, [0-7].
243effed
KW
645The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
646C<isOCTAL>.
647
648=for apidoc Am|bool|isPUNCT|char ch
649Returns a boolean indicating whether the specified character is a
6aff1f14
KW
650punctuation character, analogous to C<m/[[:punct:]]/>.
651Note that the definition of what is punctuation isn't as
243effed
KW
652straightforward as one might desire. See L<perlrecharclass/POSIX Character
653Classes> for details.
dcccc8ff
KW
654See the L<top of this section|/Character classification> for an explanation of
655variants
e5ad6aba 656C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uni>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
243effed 657C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
c99e91e9 658
0c82b6df 659=for apidoc Am|bool|isSPACE|char ch
2787a470 660Returns a boolean indicating whether the specified character is a
6aff1f14 661whitespace character. This is analogous
398d098a 662to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 663this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
664locale forms of this macro (the ones with C<LC> in their names) matched
665precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
666in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
667(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
668See the L<top of this section|/Character classification> for an explanation of
669variants
e5ad6aba 670C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uni>, C<isSPACE_utf8>, C<isSPACE_LC>,
243effed 671C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
0c82b6df 672
398d098a
KW
673=for apidoc Am|bool|isPSXSPC|char ch
674(short for Posix Space)
779cf272
KW
675Starting in 5.18, this is identical in all its forms to the
676corresponding C<isSPACE()> macros.
398d098a
KW
677The locale forms of this macro are identical to their corresponding
678C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
679non-locale forms differ from their C<isSPACE()> forms only in that the
680C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
681Otherwise they are identical. Thus this macro is analogous to what
682C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 683See the L<top of this section|/Character classification> for an explanation of
b7d90381
KW
684variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uni>, C<isPSXSPC_utf8>,
685C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
398d098a 686
954c1994 687=for apidoc Am|bool|isUPPER|char ch
2787a470 688Returns a boolean indicating whether the specified character is an
6aff1f14 689uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff
KW
690See the L<top of this section|/Character classification> for an explanation of
691variants
e5ad6aba 692C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uni>, C<isUPPER_utf8>, C<isUPPER_LC>,
243effed 693C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
954c1994 694
243effed 695=for apidoc Am|bool|isPRINT|char ch
8eea39dd 696Returns a boolean indicating whether the specified character is a
6aff1f14 697printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
698See the L<top of this section|/Character classification> for an explanation of
699variants
e5ad6aba 700C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uni>, C<isPRINT_utf8>, C<isPRINT_LC>,
243effed
KW
701C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
702
703=for apidoc Am|bool|isWORDCHAR|char ch
704Returns a boolean indicating whether the specified character is a character
705that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
706in a regular expression. A word character is an alphabetic character, a
707decimal digit, a connecting punctuation character (such as an underscore), or
708a "mark" character that attaches to one of those (like some sort of accent).
709C<isALNUM()> is a synonym provided for backward compatibility, even though a
710word character includes more than the standard C language meaning of
711alphanumeric.
dcccc8ff
KW
712See the L<top of this section|/Character classification> for an explanation of
713variants
1a83413c
KW
714C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uni>, and C<isWORDCHAR_utf8>.
715C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
716described there, but additionally include the platform's native underscore.
8a58bdcf
KW
717
718=for apidoc Am|bool|isXDIGIT|char ch
719Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
720digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
721and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
722See the L<top of this section|/Character classification> for an explanation of
723variants
243effed
KW
724C<isXDIGIT_uni>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
725C<isXDIGIT_LC_utf8>.
726
3c3ecf18
KW
727=for apidoc Am|bool|isIDFIRST|char ch
728Returns a boolean indicating whether the specified character can be the first
729character of an identifier. This is very close to, but not quite the same as
730the official Unicode property C<XID_Start>. The difference is that this
731returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
732See the L<top of this section|/Character classification> for an explanation of
733variants
3c3ecf18
KW
734C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uni>, C<isIDFIRST_utf8>,
735C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
736
737=for apidoc Am|bool|isIDCONT|char ch
738Returns a boolean indicating whether the specified character can be the
739second or succeeding character of an identifier. This is very close to, but
740not quite the same as the official Unicode property C<XID_Continue>. The
741difference is that this returns true only if the input character also matches
dcccc8ff
KW
742L</isWORDCHAR>. See the L<top of this section|/Character classification> for
743an
3c3ecf18
KW
744explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uni>,
745C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
746C<isIDCONT_LC_utf8>.
747
243effed 748=head1 Miscellaneous Functions
8eea39dd 749
95a59cab 750=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 751Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
752Behaviour is only well defined when isXDIGIT(*str) is true.
753
e7c1e6c1
KW
754=head1 Character case changing
755
1f607577
KW
756=for apidoc Am|U8|toUPPER|U8 ch
757Converts the specified character to uppercase. If the input is anything but an
758ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 759C<toUPPER_A> is equivalent.
954c1994 760
1f607577
KW
761=for apidoc Am|UV|toUPPER_uni|UV cp|U8* s|STRLEN* lenp
762Converts the Unicode code point C<cp> to its uppercase version, and
763stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
764that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
765bytes since the uppercase version may be longer than the original character.
766
767The first code point of the uppercased version is returned
768(but note, as explained just above, that there may be more.)
769
770=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
771Converts the UTF-8 encoded character at C<p> to its uppercase version, and
772stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
773that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
774bytes since the uppercase version may be longer than the original character.
775
776The first code point of the uppercased version is returned
777(but note, as explained just above, that there may be more.)
778
779The input character at C<p> is assumed to be well-formed.
780
25200305
KW
781=for apidoc Am|U8|toFOLD|U8 ch
782Converts the specified character to foldcase. If the input is anything but an
783ASCII uppercase character, that input character itself is returned. Variant
784C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
785Latin1 range, as the full generality of L</toFOLD_uni> is needed there.)
786
1f607577
KW
787=for apidoc Am|UV|toFOLD_uni|UV cp|U8* s|STRLEN* lenp
788Converts the Unicode code point C<cp> to its foldcase version, and
789stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
790that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
791bytes since the foldcase version may be longer than the original character.
792
793The first code point of the foldcased version is returned
794(but note, as explained just above, that there may be more.)
795
796=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
797Converts the UTF-8 encoded character at C<p> to its foldcase version, and
798stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
799that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
800bytes since the foldcase version may be longer than the original character.
801
802The first code point of the foldcased version is returned
803(but note, as explained just above, that there may be more.)
804
805The input character at C<p> is assumed to be well-formed.
806
807=for apidoc Am|U8|toLOWER|U8 ch
808Converts the specified character to lowercase. If the input is anything but an
809ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 810C<toLOWER_A> is equivalent.
954c1994 811
1f607577 812=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
813Converts the specified Latin1 character to lowercase. The results are
814undefined if the input doesn't fit in a byte.
1f607577
KW
815
816=for apidoc Am|U8|toLOWER_LC|U8 ch
817Converts the specified character to lowercase using the current locale's rules,
818if possible; otherwise returns the input character itself.
819
820=for apidoc Am|UV|toLOWER_uni|UV cp|U8* s|STRLEN* lenp
821Converts the Unicode code point C<cp> to its lowercase version, and
822stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
823that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
824bytes since the lowercase version may be longer than the original character.
825
826The first code point of the lowercased version is returned
827(but note, as explained just above, that there may be more.)
828
829=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
830Converts the UTF-8 encoded character at C<p> to its lowercase version, and
831stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
832that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
833bytes since the lowercase version may be longer than the original character.
834
835The first code point of the lowercased version is returned
836(but note, as explained just above, that there may be more.)
837
838The input character at C<p> is assumed to be well-formed.
839
25200305
KW
840=for apidoc Am|U8|toTITLE|U8 ch
841Converts the specified character to titlecase. If the input is anything but an
842ASCII lowercase character, that input character itself is returned. Variant
b7d90381
KW
843C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
844range, as the full generality of L</toTITLE_uni> is needed there. Titlecase is
845not a concept used in locale handling, so there is no functionality for that.)
25200305 846
1f607577
KW
847=for apidoc Am|UV|toTITLE_uni|UV cp|U8* s|STRLEN* lenp
848Converts the Unicode code point C<cp> to its titlecase version, and
849stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
850that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
851bytes since the titlecase version may be longer than the original character.
852
853The first code point of the titlecased version is returned
854(but note, as explained just above, that there may be more.)
855
856=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
857Converts the UTF-8 encoded character at C<p> to its titlecase version, and
858stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
859that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
860bytes since the titlecase version may be longer than the original character.
861
862The first code point of the titlecased version is returned
863(but note, as explained just above, that there may be more.)
864
865The input character at C<p> is assumed to be well-formed.
866
954c1994 867=cut
353c9b6f 868
1f607577 869XXX Still undocumented isVERTWS_uni and _utf8; it's unclear what their names
25200305 870really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 871
8a58bdcf 872Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
873patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
874
954c1994
GS
875*/
876
7c062697
KW
877/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
878 * is known only in the perl core, and this macro can be called from outside
879 * that */
880#ifdef HAS_QUAD
881# define WIDEST_UTYPE U64TYPE
882#else
883# define WIDEST_UTYPE U32
884#endif
885
3912bc88
KW
886/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
887 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
888 * bits of information are lost even on a 64-bit machine, but to get the
889 * compiler to optimize it out if possible. This is because Configure makes
890 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
891 * sizeof() guarantees that this evaluates to a constant true at compile time.
220c71bf
KW
892 */
893#define FITS_IN_8_BITS(c) ((sizeof(c) == 1) || !(((WIDEST_UTYPE)(c)) & ~0xFF))
cf301eb7 894
41f43cc2 895#ifdef EBCDIC
b6340bd0 896# ifndef _ALL_SOURCE
0852beac
KW
897 /* The native libc isascii() et.al. functions return the wrong results
898 * on at least z/OS unless this is defined. */
b6340bd0
KW
899# error _ALL_SOURCE should probably be defined
900# endif
41f43cc2 901#else
0852beac
KW
902 /* There is a simple definition of ASCII for ASCII platforms. But the
903 * EBCDIC one isn't so simple, so is defined using table look-up like the
904 * other macros below */
41f43cc2
KW
905# define isASCII(c) ((WIDEST_UTYPE)(c) < 128)
906#endif
907
c2da0b36
KW
908/* The lower 3 bits in both the ASCII and EBCDIC representations of '0' are 0,
909 * and the 8 possible permutations of those bits exactly comprise the 8 octal
910 * digits */
3d33b3e3 911#define isOCTAL_A(c) cBOOL(FITS_IN_8_BITS(c) && (0xF8 & (c)) == '0')
c2da0b36 912
9fb1bf9d 913#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 914
a500dc72
KW
915/* Character class numbers. For internal core Perl use only. The ones less
916 * than 32 are used in PL_charclass[] and the ones up through the one that
917 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
918 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
919 * their actual definitions are here. If that file has a name not used here,
920 * it won't compile.
1709d539
KW
921 *
922 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
923 * frequency of their use. This gives a slight edge to exiting a loop earlier
924 * (in reginclass() in regexec.c) */
1709d539
KW
925# define _CC_WORDCHAR 0 /* \w and [:word:] */
926# define _CC_DIGIT 1 /* \d and [:digit:] */
927# define _CC_ALPHA 2 /* [:alpha:] */
928# define _CC_LOWER 3 /* [:lower:] */
929# define _CC_UPPER 4 /* [:upper:] */
930# define _CC_PUNCT 5 /* [:punct:] */
931# define _CC_PRINT 6 /* [:print:] */
15861f94 932# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 933# define _CC_GRAPH 8 /* [:graph:] */
b0d691b2 934# define _CC_CASED 9 /* [:lower:] and [:upper:] under /i */
1709d539 935
b0d691b2 936#define _FIRST_NON_SWASH_CC 10
1709d539
KW
937/* The character classes above are implemented with swashes. The second group
938 * (just below) contains the ones implemented without. These are also sorted
939 * in rough order of the frequency of their use, except that \v should be last,
940 * as it isn't a real Posix character class, and some (small) inefficiencies in
941 * regular expression handling would be introduced by putting it in the middle
942 * of those that are. Also, cntrl and ascii come after the others as it may be
943 * useful to group these which have no members that match above Latin1, (or
944 * above ASCII in the latter case) */
945
779cf272 946# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
947# define _CC_BLANK 11 /* [:blank:] */
948# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
949# define _CC_CNTRL 13 /* [:cntrl:] */
950# define _CC_ASCII 14 /* [:ascii:] */
951# define _CC_VERTSPACE 15 /* \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. */
779cf272
KW
957# define _CC_IDFIRST 16
958# define _CC_CHARNAME_CONT 17
959# define _CC_NONLATIN1_FOLD 18
960# define _CC_NONLATIN1_SIMPLE_FOLD 19
961# define _CC_QUOTEMETA 20
962# define _CC_NON_FINAL_FOLD 21
963# define _CC_IS_IN_SOME_FOLD 22
964# define _CC_MNEMONIC_CNTRL 23
073c22b3
KW
965
966/* This next group is only used on EBCDIC platforms, so theoretically could be
967 * shared with something entirely different that's only on ASCII platforms */
968# define _CC_UTF8_IS_START 29
969# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
970# define _CC_UTF8_IS_CONTINUATION 31
971/* Unused: 24-28
f4cdb42c
KW
972 * If more bits are needed, one could add a second word for non-64bit
973 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
974 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
975 * is used only for optimization (as of this writing), and differs in the
976 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 977 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
978 * would just cause /i regexes which match them to run less efficiently.
979 * Similarly the EBCDIC-only bits are used just for speed, and could be
980 * replaced by other means */
96ac0975 981
3a371f2f
KW
982#if defined(PERL_CORE) || defined(PERL_EXT)
983/* An enum version of the character class numbers, to help compilers
984 * optimize */
985typedef enum {
3a371f2f 986 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
987 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
988 _CC_ENUM_ASCII = _CC_ASCII,
989 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 990 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 991 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
992 _CC_ENUM_DIGIT = _CC_DIGIT,
993 _CC_ENUM_GRAPH = _CC_GRAPH,
994 _CC_ENUM_LOWER = _CC_LOWER,
995 _CC_ENUM_PRINT = _CC_PRINT,
996 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 997 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 998 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 999 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1000 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1001 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1002} _char_class_number;
1003#endif
1004
63c61c3f 1005#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1006#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1007
b7d90381
KW
1008#if defined(PERL_IN_UTF8_C) \
1009 || defined(PERL_IN_REGCOMP_C) \
1010 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1011# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1012 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1013 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1014 #error Need to adjust order of swash_property_names[]
1015# endif
1016
b44daeeb
KW
1017/* This is declared static in each of the few files that this is #defined for
1018 * to keep them from being publicly accessible. Hence there is a small amount
1019 * of wasted space */
1020
922e8cb4
KW
1021static const char* const swash_property_names[] = {
1022 "XPosixWord",
1023 "XPosixDigit",
1024 "XPosixAlpha",
1025 "XPosixLower",
1026 "XPosixUpper",
1027 "XPosixPunct",
1028 "XPosixPrint",
1029 "XPosixAlnum",
b0d691b2
KW
1030 "XPosixGraph",
1031 "Cased"
922e8cb4
KW
1032};
1033#endif
1034
6635f04f 1035START_EXTERN_C
96ac0975
NC
1036# ifdef DOINIT
1037EXTCONST U32 PL_charclass[] = {
1038# include "l1_char_class_tab.h"
1039};
1040
1041# else /* ! DOINIT */
1042EXTCONST U32 PL_charclass[];
1043# endif
6635f04f 1044END_EXTERN_C
96ac0975 1045
265c1f46 1046 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1047# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1048
1049 /* For internal core Perl use only: the base macro for defining macros like
1050 * isALPHA */
ff7ecfc3 1051# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1052 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1053
f4cdb42c
KW
1054 /* The mask for the _A versions of the macros; it just adds in the bit for
1055 * ASCII. */
1056# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1057
4650c663
KW
1058 /* For internal core Perl use only: the base macro for defining macros like
1059 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1060 * the ASCII bit are present */
b7d90381
KW
1061# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1062 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1063 == _CC_mask_A(classnum)))
f4cdb42c 1064
f4cdb42c 1065# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1066# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1067# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1068# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1069# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1070# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1071# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1072# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1073# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1074# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1075# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1076# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1077# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1078 */
d95f8b6a 1079# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1080# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1081# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1082# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1083
1084 /* continuation character for legal NAME in \N{NAME} */
1085# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1086
1087# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1088# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1089# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1090# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1091# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1092# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1093# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1094# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1095# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1096# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1097
0852beac
KW
1098# ifdef EBCDIC
1099# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1100# endif
1101
f12c0118
KW
1102 /* Participates in a single-character fold with a character above 255 */
1103# 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)))
1104
1105 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1106# 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 1107
4eeeb416 1108# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1109# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1110 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1111# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1112 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1113# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1114 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1115#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1116
1117 /* If we don't have perl.h, we are compiling a utility program. Below we
1118 * hard-code various macro definitions that wouldn't otherwise be available
0852beac
KW
1119 * to it. Most are coded based on first principals. First some ones common
1120 * to both ASCII and EBCDIC */
1121# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1122# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1123# define isSPACE_A(c) (isBLANK_A(c) \
1124 || (c) == '\n' \
1125 || (c) == '\r' \
1126 || (c) == '\v' \
1127 || (c) == '\f')
1128# ifdef EBCDIC /* There are gaps between 'i' and 'j'; 'r' and 's'. Same
1129 for uppercase. This is ordered to exclude most things
1130 early */
1131# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1132 && ((c) <= 'i' \
1133 || ((c) >= 'j' && (c) <= 'r') \
1134 || (c) >= 's'))
1135# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1136 && ((c) <= 'I' \
1137 || ((c) >= 'J' && (c) <= 'R') \
1138 || (c) >= 'S'))
1139# else /* ASCII platform. */
9fb1bf9d 1140# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z')
2181da29 1141# define isUPPER_A(c) ((c) <= 'Z' && (c) >= 'A')
0852beac
KW
1142# endif
1143
1144 /* Some more ASCII, non-ASCII common definitions */
a4d7a999
KW
1145# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1146# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1147# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac
KW
1148# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1149# define isXDIGIT_A(c) (isDIGIT_A(c) \
1150 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1151 || ((c) <= 'F' && (c) >= 'A'))
3ded5eb0 1152
0852beac
KW
1153# ifdef EBCDIC
1154# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1155 || (c) == '#' || (c) == '$' || (c) == '%' \
1156 || (c) == '&' || (c) == '\'' || (c) == '(' \
1157 || (c) == ')' || (c) == '*' || (c) == '+' \
1158 || (c) == ',' || (c) == '.' || (c) == '/' \
1159 || (c) == ':' || (c) == ';' || (c) == '<' \
1160 || (c) == '=' || (c) == '>' || (c) == '?' \
1161 || (c) == '@' || (c) == '[' || (c) == '\\' \
1162 || (c) == ']' || (c) == '^' || (c) == '_' \
1163 || (c) == '`' || (c) == '{' || (c) == '|' \
1164 || (c) == '}' || (c) == '~')
1165# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1166# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1167
1168# ifdef QUESTION_MARK_CTRL
1169# define _isQMC(c) ((c) == QUESTION_MARK_CTRL)
1170# else
1171# define _isQMC(c) 0
1172# endif
1173
1174 /* I (khw) can't think of a way to define all the ASCII controls
1175 * without resorting to a libc (locale-sensitive) call. But we know
1176 * that all controls but the question-mark one are in the range 0-0x3f.
1177 * This makes sure that all the controls that have names are included,
1178 * and all controls that are also considered ASCII in the locale. This
1179 * may include more or fewer than what it actually should, but the
1180 * wrong ones are less-important controls, so likely won't impact
1181 * things (keep in mind that this is compiled only if perl.h isn't
1182 * available). The question mark control is included if available */
1183# define isCNTRL_A(c) (((c) < 0x40 && isascii(c)) \
1184 || (c) == '\0' || (c) == '\a' || (c) == '\b' \
1185 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1186 || (c) == '\t' || (c) == '\v' || _isQMC(c))
1187
1188# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1189# else /* ASCII platform; things are simpler, and isASCII has already
1190 been defined */
1191# define isGRAPH_A(c) (((c) > ' ' && (c) < 127))
1192# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1193# define isPUNCT_A(c) (isGRAPH_A(c) && (! isALPHANUMERIC_A(c)))
1194# define isCNTRL_A(c) (isASCII(c) && (! isPRINT_A(c)))
1195# endif
1196
3ffc8c70 1197 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1198 * during debugging, and it seems best to keep them. We may be called
1199 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1200 * do anything anyway, so make it not a problem */
1201# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1202# define NATIVE_TO_LATIN1(ch) (ch)
1203# endif
3ded5eb0
KW
1204# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1205# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1206# define isBLANK_L1(c) (isBLANK_A(c) \
1207 || (FITS_IN_8_BITS(c) \
1208 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1209# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1210# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1211# define isLOWER_L1(c) (isLOWER_A(c) \
1212 || (FITS_IN_8_BITS(c) \
1213 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1214 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1215 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1216 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1217 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1218# define isPRINT_L1(c) (isPRINT_A(c) \
1219 || (FITS_IN_8_BITS(c) \
1220 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1221# define isPUNCT_L1(c) (isPUNCT_A(c) \
1222 || (FITS_IN_8_BITS(c) \
1223 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1224 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1225 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1226 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1227 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1228 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1229 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1230# define isSPACE_L1(c) (isSPACE_A(c) \
1231 || (FITS_IN_8_BITS(c) \
1232 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1233 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1234# define isUPPER_L1(c) (isUPPER_A(c) \
1235 || (FITS_IN_8_BITS(c) \
1236 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1237 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1238 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1239# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1240# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1241# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1242 || isBLANK_L1(c) \
1243 || (c) == '-' \
1244 || (c) == '(' \
1245 || (c) == ')')
1246 /* The following are not fully accurate in the above-ASCII range. I (khw)
1247 * don't think it's necessary to be so for the purposes where this gets
1248 * compiled */
1249# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1250# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1251
1252 /* And these aren't accurate at all. They are useful only for above
1253 * Latin1, which utilities and bootstrapping don't deal with */
1254# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1255# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1256# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1257
1258 /* Many of the macros later in this file are defined in terms of these. By
1259 * implementing them with a function, which converts the class number into
1260 * a call to the desired macro, all of the later ones work. However, that
1261 * function won't be actually defined when building a utility program (no
1262 * perl.h), and so a compiler error will be generated if one is attempted
1263 * to be used. And the above-Latin1 code points require Unicode tables to
1264 * be present, something unlikely to be the case when bootstrapping */
1265# define _generic_isCC(c, classnum) \
1266 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1267# define _generic_isCC_A(c, classnum) \
1268 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1269#endif /* End of no perl.h H_PERL */
8a58bdcf 1270
e66b99e9
KW
1271#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1272#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1273#define isASCII_A(c) isASCII(c)
1274#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1275#define isBLANK(c) isBLANK_A(c)
1276#define isCNTRL(c) isCNTRL_A(c)
1277#define isDIGIT(c) isDIGIT_A(c)
1278#define isGRAPH(c) isGRAPH_A(c)
1279#define isIDFIRST(c) isIDFIRST_A(c)
1280#define isLOWER(c) isLOWER_A(c)
1281#define isPRINT(c) isPRINT_A(c)
779cf272 1282#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1283#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1284#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1285#define isPUNCT(c) isPUNCT_A(c)
1286#define isSPACE(c) isSPACE_A(c)
1287#define isUPPER(c) isUPPER_A(c)
1288#define isWORDCHAR(c) isWORDCHAR_A(c)
1289#define isXDIGIT(c) isXDIGIT_A(c)
1290
1291/* ASCII casing. These could also be written as
1292 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1293 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1294 which uses table lookup and mask instead of subtraction. (This would
1295 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1296#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1297#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1298
25200305
KW
1299/* In the ASCII range, these are equivalent to what they're here defined to be.
1300 * But by creating these definitions, other code doesn't have to be aware of
1301 * this detail */
1302#define toFOLD(c) toLOWER(c)
25200305
KW
1303#define toTITLE(c) toUPPER(c)
1304
c753c8d3
KW
1305#define toLOWER_A(c) toLOWER(c)
1306#define toUPPER_A(c) toUPPER(c)
25200305
KW
1307#define toFOLD_A(c) toFOLD(c)
1308#define toTITLE_A(c) toTITLE(c)
1a0901db 1309
4650c663 1310/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1311#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1312 ? (c) \
f4cd282c 1313 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1314#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1315
1a0901db 1316/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1317 * all mapped to one of them, and these need special handling; returns the
1318 * input itself if is out-of-range */
b2bf251f 1319#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1320 ? (c) \
f4cd282c 1321 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1322#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1323
beab9ebe
KW
1324/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1325
1326/* For internal core Perl use only: the base macro for defining macros like
1327 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1328 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1329 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1330 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1331 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1332 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1333 * code points under locale rules, use the macros based on _generic_LC_uvchr
1334 * instead of this. */
beab9ebe
KW
1335#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1336 (! FITS_IN_8_BITS(c) \
1337 ? 0 \
31f05a37
KW
1338 : IN_UTF8_CTYPE_LOCALE \
1339 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1340 : cBOOL(non_utf8))
1341
1342/* For internal core Perl use only: a helper macro for defining macros like
1343 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1344 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1345 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1346 * _generic_LC_base for more info */
1347#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1348 _generic_LC_base(c,utf8_locale_classnum, \
1349 non_utf8_func( (_LC_CAST) (c)))
1350
1351/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1352 * 'c' is the platform's native underscore character */
1353#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1354 _generic_LC_base(c, utf8_locale_classnum, \
1355 (non_utf8_func( (_LC_CAST) (c)) \
1356 || (char)(c) == '_'))
1357
1358/* These next three are also for internal core Perl use only: case-change
1359 * helper macros */
1360#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1361 ? (c) \
31f05a37
KW
1362 : (IN_UTF8_CTYPE_LOCALE) \
1363 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1364 : (cast)function((cast)(c)))
beab9ebe 1365
31f05a37
KW
1366/* Note that the result can be larger than a byte in a UTF-8 locale. It
1367 * returns a single value, so can't adequately return the upper case of LATIN
1368 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1369 * values "SS"); instead it asserts against that under DEBUGGING, and
1370 * otherwise returns its input */
beab9ebe
KW
1371#define _generic_toUPPER_LC(c, function, cast) \
1372 (! FITS_IN_8_BITS(c) \
1373 ? (c) \
31f05a37 1374 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1375 ? (cast)function((cast)(c)) \
31f05a37
KW
1376 : ((((U8)(c)) == MICRO_SIGN) \
1377 ? GREEK_CAPITAL_LETTER_MU \
1378 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1379 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1380 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1381 ? (__ASSERT_(0) (c)) \
1382 : PL_mod_latin1_uc[ (U8) (c) ])))))
1383
1384/* Note that the result can be larger than a byte in a UTF-8 locale. It
1385 * returns a single value, so can't adequately return the fold case of LATIN
1386 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1387 * values "ss"); instead it asserts against that under DEBUGGING, and
1388 * otherwise returns its input */
beab9ebe 1389#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1390 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1391 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1392 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1393 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1394 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1395
84061b6a 1396/* Use the libc versions for these if available. */
f05550c0 1397#if defined(HAS_ISASCII)
84061b6a
KW
1398# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1399#else
1400# define isASCII_LC(c) isASCII(c)
1401#endif
1402
f05550c0 1403#if defined(HAS_ISBLANK)
84061b6a 1404# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1405#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1406# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1407#endif
1408
f05550c0 1409#define _LC_CAST U8
bbce6d69 1410
f05550c0 1411#ifdef WIN32
375f5f06
KW
1412 /* The Windows functions don't bother to follow the POSIX standard, which
1413 * for example says that something can't both be a printable and a control.
1414 * But Windows treats the \t control as a printable, and does such things
1415 * as making superscripts into both digits and punctuation. This tames
1416 * these flaws by assuming that the definitions of both controls and space
1417 * are correct, and then making sure that other definitions don't have
1418 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1419 * Not all possible weirdnesses are checked for, just the ones that were
1420 * detected on actual Microsoft code pages */
1421
b7d90381
KW
1422# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1423# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1424
1425# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1426 && isALPHANUMERIC_LC(c))
1427# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1428 ! isPUNCT_LC(c))
1429# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1430 isALPHANUMERIC_LC(c))
1431# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1432# define isIDFIRST_LC(c) (((c) == '_') \
1433 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1434# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1435# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1436# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1437# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1438# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1439# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1440 && isALPHANUMERIC_LC(c))
f05550c0
BF
1441
1442# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1443# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1444# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1445
1446#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1447 /* For most other platforms */
beab9ebe 1448
f05550c0
BF
1449# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1450# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1451# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1452# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1453# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1454# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1455# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1456# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1457# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1458# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1459# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1460# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1461# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1462
1463
1464# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1465# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1466# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1467
1468#else /* The final fallback position */
1469
b7d90381
KW
1470# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1471# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1472# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1473# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1474# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1475# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1476# define isLOWER_LC(c) (isascii(c) && islower(c))
1477# define isPRINT_LC(c) (isascii(c) && isprint(c))
1478# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1479# define isSPACE_LC(c) (isascii(c) && isspace(c))
1480# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1481# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1482# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1483
1484# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1485# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1486# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1487
f05550c0 1488#endif
55204971 1489
eba68aa0
KW
1490#define isIDCONT(c) isWORDCHAR(c)
1491#define isIDCONT_A(c) isWORDCHAR_A(c)
1492#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1493#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1494#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1495
4650c663
KW
1496/* For internal core Perl use only: the base macros for defining macros like
1497 * isALPHA_uni. 'c' is the code point to check. 'classnum' is the POSIX class
1498 * number defined earlier in this file. _generic_uni() is used for POSIX
1499 * classes where there is a macro or function 'above_latin1' that takes the
1500 * single argument 'c' and returns the desired value. These exist for those
1501 * classes which have simple definitions, avoiding the overhead of a hash
1502 * lookup or inversion list binary search. _generic_swash_uni() can be used
1503 * for classes where that overhead is faster than a direct lookup.
1504 * _generic_uni() won't compile if 'c' isn't unsigned, as it won't match the
1505 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1506 * have duplicate checks here, so could create versions of the macros that
1507 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1508
1509/* Note that all ignore 'use bytes' */
cd500f2f
KW
1510#define _generic_uni(classnum, above_latin1, c) ((c) < 256 \
1511 ? _generic_isCC(c, classnum) \
1512 : above_latin1(c))
1513#define _generic_swash_uni(classnum, c) ((c) < 256 \
1514 ? _generic_isCC(c, classnum) \
922e8cb4 1515 : _is_uni_FOO(classnum, c))
cd500f2f
KW
1516#define isALPHA_uni(c) _generic_swash_uni(_CC_ALPHA, c)
1517#define isALPHANUMERIC_uni(c) _generic_swash_uni(_CC_ALPHANUMERIC, c)
509fb054
KW
1518#define isASCII_uni(c) isASCII(c)
1519#define isBLANK_uni(c) _generic_uni(_CC_BLANK, is_HORIZWS_cp_high, c)
1520#define isCNTRL_uni(c) isCNTRL_L1(c) /* All controls are in Latin1 */
cd500f2f
KW
1521#define isDIGIT_uni(c) _generic_swash_uni(_CC_DIGIT, c)
1522#define isGRAPH_uni(c) _generic_swash_uni(_CC_GRAPH, c)
eba68aa0 1523#define isIDCONT_uni(c) _generic_uni(_CC_WORDCHAR, _is_uni_perl_idcont, c)
509fb054 1524#define isIDFIRST_uni(c) _generic_uni(_CC_IDFIRST, _is_uni_perl_idstart, c)
cd500f2f
KW
1525#define isLOWER_uni(c) _generic_swash_uni(_CC_LOWER, c)
1526#define isPRINT_uni(c) _generic_swash_uni(_CC_PRINT, c)
aea0c08c 1527
cd500f2f 1528#define isPUNCT_uni(c) _generic_swash_uni(_CC_PUNCT, c)
509fb054 1529#define isSPACE_uni(c) _generic_uni(_CC_SPACE, is_XPERLSPACE_cp_high, c)
779cf272
KW
1530#define isPSXSPC_uni(c) isSPACE_uni(c)
1531
cd500f2f 1532#define isUPPER_uni(c) _generic_swash_uni(_CC_UPPER, c)
509fb054 1533#define isVERTWS_uni(c) _generic_uni(_CC_VERTSPACE, is_VERTWS_cp_high, c)
cd500f2f 1534#define isWORDCHAR_uni(c) _generic_swash_uni(_CC_WORDCHAR, c)
509fb054 1535#define isXDIGIT_uni(c) _generic_uni(_CC_XDIGIT, is_XDIGIT_cp_high, c)
b18192e9 1536
b8d68ded 1537#define toFOLD_uni(c,s,l) to_uni_fold(c,s,l)
509fb054
KW
1538#define toLOWER_uni(c,s,l) to_uni_lower(c,s,l)
1539#define toTITLE_uni(c,s,l) to_uni_title(c,s,l)
1540#define toUPPER_uni(c,s,l) to_uni_upper(c,s,l)
a0ed51b3 1541
4650c663
KW
1542/* For internal core Perl use only: the base macros for defining macros like
1543 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1544 * point, not just 0-255. Like _generic_uni, there are two versions, one for
1545 * simple class definitions; the other for more complex. These are like
1546 * _generic_uni, so see it for more info. */
cd500f2f
KW
1547#define _generic_LC_uvchr(latin1, above_latin1, c) \
1548 (c < 256 ? latin1(c) : above_latin1(c))
1549#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1550 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1551
1552#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1553#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1554 _CC_ALPHANUMERIC, c)
b7d90381
KW
1555#define isASCII_LC_uvchr(c) isASCII_LC(c)
1556#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1557 is_HORIZWS_cp_high, c)
feeab5a9 1558#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1559#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1560#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1561#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1562 _is_uni_perl_idcont, c)
b7d90381 1563#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1564 _is_uni_perl_idstart, c)
1565#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1566#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1567#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1568#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1569#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1570 is_XPERLSPACE_cp_high, c)
cd500f2f 1571#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1572#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1573 _CC_WORDCHAR, c)
b7d90381
KW
1574#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1575 is_XDIGIT_cp_high, c)
e712593e 1576
b7d90381 1577#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1578
4650c663
KW
1579/* For internal core Perl use only: the base macros for defining macros like
1580 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1581 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1582 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1583 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1584 * same representation whether utf8 or not. Note that it assumes that the utf8
1585 * has been validated, and ignores 'use bytes' */
cd500f2f 1586#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1587 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1588 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1589 ? _generic_isCC( \
a62b247b 1590 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
aea0c08c 1591 *((p)+1 )), \
94bb8c36 1592 classnum) \
252810d0 1593 : utf8)
b7d90381
KW
1594/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1595 * 'above_latin1' can be a macro */
cd500f2f
KW
1596#define _generic_func_utf8(classnum, above_latin1, p) \
1597 _generic_utf8(classnum, p, above_latin1(p))
4650c663 1598/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1599 * 'above_latin1' parameter */
922e8cb4 1600#define _generic_swash_utf8(classnum, p) \
cd500f2f 1601 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
922e8cb4 1602
cc8ab7c0 1603/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1604 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1605 * class is TRUE for. Hence it can skip the tests for this range.
1606 * 'above_latin1' should include its arguments */
3ded5eb0 1607#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1608 (UTF8_IS_INVARIANT(*(p)) \
1609 ? _generic_isCC(*(p), classnum) \
1610 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1611 ? above_latin1 \
cc8ab7c0
KW
1612 : 0)
1613
84238efa
KW
1614/* NOTE that some of these macros have very similar ones in regcharclass.h.
1615 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1616 * there, differing in name only by an underscore from the one here
1617 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1618 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1619 * points; the regcharclass.h ones are implemented as a series of
1620 * "if-else-if-else ..." */
1621
b7d90381
KW
1622#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1623#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1624#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
509fb054
KW
1625 utf8, the non-utf8 macro works
1626 */
b7d90381 1627#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2
KW
1628
1629#ifdef EBCDIC
1630 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1631 * more efficient macro instead of the more general one */
88794300 1632# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
e8fa43e2
KW
1633#else
1634# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1635#endif
1636
cd500f2f
KW
1637#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1638 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1639#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1640#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1641 _is_utf8_perl_idcont, p)
e5dcd934 1642
c11ff943
KW
1643/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1644 * IDFIRST is an alnum. See
c8362b00 1645 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1646 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1647 * This used to be not the XID version, but we decided to go with the more
1648 * modern Unicode definition */
b7d90381 1649#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1650 _is_utf8_perl_idstart, p)
e5dcd934 1651
b7d90381
KW
1652#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1653#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1654#define isPSXSPC_utf8(p) isSPACE_utf8(p)
1655#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1656#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1657#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1658#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1659#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1660#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
cd500f2f 1661 is_XDIGIT_high(p))
a0ed51b3 1662
7d2b61dd 1663#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1664#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1665#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1666#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1667
4650c663
KW
1668/* For internal core Perl use only: the base macros for defining macros like
1669 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1670 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1671 * 'macro' parameter */
94bb8c36
KW
1672#define _generic_LC_utf8(macro, p, utf8) \
1673 (UTF8_IS_INVARIANT(*(p)) \
1674 ? macro(*(p)) \
1675 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
a62b247b 1676 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
9201a572
KW
1677 : utf8)
1678
cd500f2f
KW
1679#define _generic_LC_swash_utf8(macro, classnum, p) \
1680 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1681#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1682 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1683
b7d90381 1684#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
cd500f2f 1685 _CC_ALPHANUMERIC, p)
b7d90381
KW
1686#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1687#define isASCII_LC_utf8(p) isASCII_LC(*p)
1688#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1689 is_HORIZWS_high, p)
1690#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1691#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1692#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1693#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1694 _is_utf8_perl_idcont, p)
1695#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1696 _is_utf8_perl_idstart, p)
1697#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1698#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1699#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1700#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1701#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1702 is_XPERLSPACE_high, p)
1703#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
cd500f2f
KW
1704#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1705 _CC_WORDCHAR, p)
b7d90381
KW
1706#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1707 is_XDIGIT_high, p)
aaa51d5e 1708
fbc19f27
KW
1709/* Macros for backwards compatibility and for completeness when the ASCII and
1710 * Latin1 values are identical */
b7d90381
KW
1711#define isALPHAU(c) isALPHA_L1(c)
1712#define isDIGIT_L1(c) isDIGIT_A(c)
1713#define isOCTAL(c) isOCTAL_A(c)
1714#define isOCTAL_L1(c) isOCTAL_A(c)
1715#define isXDIGIT_L1(c) isXDIGIT_A(c)
1716#define isALNUM(c) isWORDCHAR(c)
1717#define isALNUMU(c) isWORDCHAR_L1(c)
1718#define isALNUM_LC(c) isWORDCHAR_LC(c)
1719#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 1720#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
1721#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1722#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1723#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1724#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1725#define isALNUMC(c) isALPHANUMERIC(c)
1726#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1727#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 1728#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 1729#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 1730#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1731
2bd1cbf6
KW
1732/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1733 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1734 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1735 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1736 * the outlier from the block that contains the other controls, just like
1737 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1738 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
1739 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1740 * etc. */
2bd1cbf6 1741#ifndef EBCDIC
75763b3a 1742# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 1743#else
75763b3a
KW
1744# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1745 ((isPRINT_A(c)) \
1746 ? (UNLIKELY((c) == '?') \
1747 ? QUESTION_MARK_CTRL \
1748 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1749 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1750 ? '?' \
1751 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 1752#endif
bbce6d69 1753
dea28490
JJ
1754/* Line numbers are unsigned, 32 bits. */
1755typedef U32 line_t;
e5dcd934 1756#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 1757
91152fc1
DG
1758/* Helpful alias for version prescan */
1759#define is_LAX_VERSION(a,b) \
1760 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1761
1762#define is_STRICT_VERSION(a,b) \
1763 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1764
1765#define BADVERSION(a,b,c) \
1766 if (b) { \
1767 *b = c; \
1768 } \
1769 return a;
8c52afec 1770
cb27eebd
KW
1771/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1772 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1773 * which is advanced. The input is validated only by an assert() in DEBUGGING
1774 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1775 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1776#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1777 ? (c) \
1778 : ((c) + 9))))
902008b8
KW
1779#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1780 ? (*(s)++) \
1781 : (*(s)++ + 9))))
95a59cab 1782
cb27eebd
KW
1783/* Converts a character known to represent an octal digit (0-7) to its numeric
1784 * value. The input is validated only by an assert() in DEBUGGING builds. In
1785 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1786#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1787
305b8651
KW
1788/* Efficiently returns a boolean as to if two native characters are equivalent
1789 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1790 * the ALPHA in the name is to remind you of that. This is asserted() in
1791 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1792 * works (on valid input) for both non- and UTF-8-encoded bytes.
1793 *
1794 * When one of the inputs is a compile-time constant and gets folded by the
1795 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1796 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1797 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
1798 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1799 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1800 * just a single 0, in the bit position where the upper- and lowercase differ.
1801 * */
305b8651
KW
1802#define isALPHA_FOLD_EQ(c1, c2) \
1803 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1804 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1805#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1806
8e84507e 1807/*
ccfc67b7
JH
1808=head1 Memory Management
1809
a02a5408 1810=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
1811The XSUB-writer's interface to the C C<malloc> function.
1812
0d7b2759
MH
1813Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1814
c5008215
JC
1815In 5.9.3, Newx() and friends replace the older New() API, and drops
1816the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 1817themselves. This aid has been superseded by a new build option,
d10b4965 1818PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
1819there for use in XS modules supporting older perls.
1820
a02a5408 1821=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 1822The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 1823cast. See also C<L</Newx>>.
954c1994 1824
0d7b2759
MH
1825Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1826
a02a5408 1827=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 1828The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 1829memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 1830
0d7b2759
MH
1831Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1832
954c1994
GS
1833=for apidoc Am|void|Renew|void* ptr|int nitems|type
1834The XSUB-writer's interface to the C C<realloc> function.
1835
0d7b2759
MH
1836Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1837
954c1994
GS
1838=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1839The XSUB-writer's interface to the C C<realloc> function, with
1840cast.
1841
0d7b2759
MH
1842Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1843
49b8b560 1844=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
1845The XSUB-writer's interface to the C C<free> function.
1846
0d7b2759
MH
1847This should B<ONLY> be used on memory obtained using L<"Newx"> and friends.
1848
954c1994
GS
1849=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1850The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 1851source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1852C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 1853
e90e2364 1854=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 1855Like C<Move> but returns C<dest>. Useful
72d33970 1856for encouraging compilers to tail-call
e90e2364
NC
1857optimise.
1858
954c1994
GS
1859=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1860The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 1861source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1862C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 1863
e90e2364
NC
1864=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1865
796b6530 1866Like C<Copy> but returns C<dest>. Useful
72d33970 1867for encouraging compilers to tail-call
e90e2364
NC
1868optimise.
1869
954c1994
GS
1870=for apidoc Am|void|Zero|void* dest|int nitems|type
1871
1872The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1873destination, C<nitems> is the number of items, and C<type> is the type.
1874
e90e2364
NC
1875=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1876
72d33970
FC
1877Like C<Zero> but returns dest. Useful
1878for encouraging compilers to tail-call
e90e2364
NC
1879optimise.
1880
da5d8dbb 1881=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 1882This is an architecture-independent macro to copy one structure to another.
954c1994 1883
7e337ee0
JH
1884=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1885
1886Fill up memory with a byte pattern (a byte repeated over and over
1887again) that hopefully catches attempts to access uninitialized memory.
1888
1889=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1890
1891PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1892
1c12ffb4 1893=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
1894
1895PoisonWith(0xEF) for catching access to freed memory.
1896
9965345d
JH
1897=for apidoc Am|void|Poison|void* dest|int nitems|type
1898
7e337ee0 1899PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
1900
1901=cut */
954c1994 1902
561b68a9
SH
1903/* Maintained for backwards-compatibility only. Use newSV() instead. */
1904#ifndef PERL_CORE
ff06c60c 1905#define NEWSV(x,len) newSV(len)
561b68a9 1906#endif
ff06c60c 1907
19a94d75
JH
1908#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1909
e6bdf523 1910
27d5b266 1911#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
1912
1913/* This expression will be constant-folded at compile time. It checks
1914 * whether or not the type of the count n is so small (e.g. U8 or U16, or
1915 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
1916 * As well as avoiding the need for a run-time check in some cases, it's
1917 * designed to avoid compiler warnings like:
1918 * comparison is always false due to limited range of data type
1919 */
1920
1921# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
1922 (sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
1923
88f9f128 1924/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
1925 * compiler warnings. We *want* to write the expression as
1926 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
1927 * (for some compile-time constant C), but even when the LHS
1928 * constant-folds to false at compile-time, g++ insists on emitting
1929 * warnings about the RHS (e.g. "comparison is always false"), so instead
1930 * we write it as
e6bdf523 1931 *
d98e5cde 1932 * (cond ? n : X) > C
88f9f128 1933 *
d98e5cde
DM
1934 * where X is a constant with X > C always false. Choosing a value for X
1935 * is tricky. If 0, some compilers will complain about 0 > C always being
1936 * false; if 1, Coverity complains when n happens to be the constant value
1937 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
1938 * for X and hope that nothing else whines.
e6bdf523
DM
1939 */
1940
1941# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
1942 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
1943 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
1944
1945# define MEM_WRAP_CHECK(n,t) \
1946 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1947 && (croak_memory_wrap(),0))
1948
1949# define MEM_WRAP_CHECK_1(n,t,a) \
1950 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1951 && (Perl_croak_nocontext("%s",(a)),0))
1952
8b44ba4c 1953#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 1954
d1decf2b 1955#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
1956#else
1957
410319be
NC
1958#define MEM_WRAP_CHECK(n,t)
1959#define MEM_WRAP_CHECK_1(n,t,a)
1960#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
1961#define MEM_WRAP_CHECK_(n,t)
1962
6fff79ce 1963#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 1964
1936d2a7 1965#endif
8b44ba4c 1966
fe4f188c 1967#ifdef PERL_MEM_LOG
46c6c7e2 1968/*
9f653bb5 1969 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
1970 * go through functions, which are handy for debugging breakpoints, but
1971 * which more importantly get the immediate calling environment (file and
e352bcff
JH
1972 * line number, and C function name if available) passed in. This info can
1973 * then be used for logging the calls, for which one gets a sample
73d1d973 1974 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 1975 *
46c6c7e2 1976 * Known problems:
94e892a6 1977 * - not all memory allocs get logged, only those
46c6c7e2 1978 * that go through Newx() and derivatives (while all
94e892a6 1979 * Safefrees do get logged)
46c6c7e2
JH
1980 * - __FILE__ and __LINE__ do not work everywhere
1981 * - __func__ or __FUNCTION__ even less so
1982 * - I think more goes on after the perlio frees but
1983 * the thing is that STDERR gets closed (as do all
1984 * the file descriptors)
1985 * - no deeper calling stack than the caller of the Newx()
1986 * or the kind, but do I look like a C reflection/introspection
1987 * utility to you?
1988 * - the function prototypes for the logging functions
1989 * probably should maybe be somewhere else than handy.h
1990 * - one could consider inlining (macrofying) the logging
1991 * for speed, but I am too lazy
1992 * - one could imagine recording the allocations in a hash,
1993 * (keyed by the allocation address?), and maintain that
1994 * through reallocs and frees, but how to do that without
1995 * any News() happening...?
73d1d973 1996 * - lots of -Ddefines to get useful/controllable output
b953482e 1997 * - lots of ENV reads
46c6c7e2
JH
1998 */
1999
fc6bde6f 2000PERL_CALLCONV Malloc_t Perl_mem_log_alloc(const UV n, const UV typesize, const char *type_name, Malloc_t newalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 2001
fc6bde6f 2002PERL_CALLCONV Malloc_t Perl_mem_log_realloc(const UV n, const UV typesize, const char *type_name, Malloc_t oldalloc, Malloc_t newalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 2003
fc6bde6f 2004PERL_CALLCONV Malloc_t Perl_mem_log_free(Malloc_t oldalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 2005
0b0ab801 2006# ifdef PERL_CORE
73d1d973 2007# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2008enum mem_log_type {
2009 MLT_ALLOC,
2010 MLT_REALLOC,
d7a2c63c
MHM
2011 MLT_FREE,
2012 MLT_NEW_SV,
2013 MLT_DEL_SV
0b0ab801
MHM
2014};
2015# endif
12754f92 2016# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2017void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2018void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2019# endif
0b0ab801
MHM
2020# endif
2021
fe4f188c
JH
2022#endif
2023
2024#ifdef PERL_MEM_LOG
d1401ee9
MHM
2025#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2026#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2027#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2028#endif
2029
2030#ifndef MEM_LOG_ALLOC
2031#define MEM_LOG_ALLOC(n,t,a) (a)
2032#endif
2033#ifndef MEM_LOG_REALLOC
2034#define MEM_LOG_REALLOC(n,t,v,a) (a)
2035#endif
2036#ifndef MEM_LOG_FREE
2037#define MEM_LOG_FREE(a) (a)
2038#endif
2039
d1401ee9
MHM
2040#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2041#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2042#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2043
2044#ifndef PERL_CORE
a02a5408
JC
2045/* pre 5.9.x compatibility */
2046#define New(x,v,n,t) Newx(v,n,t)
2047#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2048#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2049#endif
a02a5408 2050
ff68c719 2051#define Renew(v,n,t) \
d1401ee9 2052 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2053#define Renewc(v,n,t,c) \
d1401ee9 2054 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2055
2056#ifdef PERL_POISON
2057#define Safefree(d) \
06c0cc96 2058 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2059#else
fe4f188c 2060#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2061#endif
55497cff 2062
8b44ba4c
NC
2063#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2064#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2065#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2066
8b44ba4c
NC
2067#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2068#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2069#ifdef HAS_MEMSET
8b44ba4c 2070#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2071#else
8b44ba4c
NC
2072/* Using bzero(), which returns void. */
2073#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2074#endif
2075
7e337ee0
JH
2076#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2077#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2078#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2079#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2080
caa674f3
DD
2081#ifdef PERL_POISON
2082# define PERL_POISON_EXPR(x) x
2083#else
2084# define PERL_POISON_EXPR(x)
2085#endif
2086
2304df62 2087#ifdef USE_STRUCT_COPY
ff68c719 2088#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2089#else
2090#define StructCopy(s,d,t) Copy(s,d,1,t)
2091#endif
2cc61e15 2092
c3caa5c3
JH
2093/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2094 * want your zero-based indices to be less than but not equal to).
2095 *
2096 * C_ARRAY_END is one past the last: half-open/half-closed range,
2097 * not last-inclusive range. */
622913ab 2098#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2099#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2100
2cc61e15
DD
2101#ifdef NEED_VA_COPY
2102# ifdef va_copy
2103# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2104# else
a1866d1b
JH
2105# if defined(__va_copy)
2106# define Perl_va_copy(s, d) __va_copy(d, s)
2107# else
2108# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2109# endif
2cc61e15
DD
2110# endif
2111#endif
2112
472d47bc
SB
2113/* convenience debug macros */
2114#ifdef USE_ITHREADS
2115#define pTHX_FORMAT "Perl interpreter: 0x%p"
2116#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2117#define pTHX_VALUE_ (void *)my_perl,
2118#define pTHX_VALUE (void *)my_perl
2119#define pTHX__VALUE_ ,(void *)my_perl,
2120#define pTHX__VALUE ,(void *)my_perl
472d47bc 2121#else
3609ea0d 2122#define pTHX_FORMAT
472d47bc 2123#define pTHX__FORMAT
3609ea0d 2124#define pTHX_VALUE_
472d47bc 2125#define pTHX_VALUE
3609ea0d 2126#define pTHX__VALUE_
472d47bc
SB
2127#define pTHX__VALUE
2128#endif /* USE_ITHREADS */
3609ea0d 2129
2acdbac1
NC
2130/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2131 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2132 CPAN::Unpack show any users outside the core. */
2133#ifdef PERL_CORE
b7d90381
KW
2134# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2135 "Use of " s " is deprecated")
2acdbac1
NC
2136#endif
2137
dfff4baf
BF
2138/* Internal macros to deal with gids and uids */
2139#ifdef PERL_CORE
2140
2141# if Uid_t_size > IVSIZE
2142# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2143# define SvUID(sv) SvNV(sv)
2144# else
2145# if Uid_t_sign <= 0
2146# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2147# define SvUID(sv) SvIV(sv)
2148# else
2149# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2150# define SvUID(sv) SvUV(sv)
2151# endif
2152# endif /* Uid_t_size */
2153
2154# if Gid_t_size > IVSIZE
2155# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2156# define SvGID(sv) SvNV(sv)
2157# else
2158# if Gid_t_sign <= 0
2159# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2160# define SvGID(sv) SvIV(sv)
2161# else
2162# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2163# define SvGID(sv) SvUV(sv)
2164# endif
2165# endif /* Gid_t_size */
2166
2167#endif
2168
9d745869
KW
2169#endif /* HANDY_H */
2170
e9a8c099 2171/*
14d04a33 2172 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2173 */