This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Make FITS_IN_8_BITS() always true under Coverity
[perl5.git] / handy.h
CommitLineData
a0d0e21e 1/* handy.h
a687059c 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
da5d8dbb 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
a687059c 5 *
6e21c824
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8d063cd8 8 *
8d063cd8
LW
9 */
10
4650c663
KW
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
9d745869
KW
14#ifndef HANDY_H /* Guard against nested #inclusion */
15#define HANDY_H
16
85e6fe83 17#if !defined(__STDC__)
378cc40b
LW
18#ifdef NULL
19#undef NULL
20#endif
a687059c 21# define NULL 0
85e6fe83
LW
22#endif
23
24792b8d
NC
24#ifndef PERL_CORE
25# define Null(type) ((type)NULL)
954c1994
GS
26
27/*
ccfc67b7 28=head1 Handy Values
954c1994 29
3609ea0d 30=for apidoc AmU||Nullch
72d33970
FC
31Null character pointer. (No longer available when C<PERL_CORE> is
32defined.)
2307c6d0 33
954c1994 34=for apidoc AmU||Nullsv
72d33970 35Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
36
37=cut
38*/
39
24792b8d
NC
40# define Nullch Null(char*)
41# define Nullfp Null(PerlIO*)
42# define Nullsv Null(SV*)
43#endif
8d063cd8 44
641d3f0b
PP
45#ifdef TRUE
46#undef TRUE
47#endif
48#ifdef FALSE
49#undef FALSE
50#endif
51#define TRUE (1)
52#define FALSE (0)
53
cf3f0ffb
DM
54/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
55 * (compiler permitting) that casting away const-ness will give a warning;
56 * e.g.:
57 *
58 * const SV *sv = ...;
59 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
60 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
61 */
62
b1bc3f34 63#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
f065df04 64# define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
b1bc3f34
NC
65#else
66# define MUTABLE_PTR(p) ((void *) (p))
67#endif
68
a062e10d 69#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 70#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 71#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 72#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 73#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 74#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 75
f789f6a4 76#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
77# include <stdbool.h>
78# ifndef HAS_BOOL
79# define HAS_BOOL 1
80# endif
81#endif
82
8e84507e 83/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
84 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
85 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 86 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
87 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
88 g++ can be identified by __GNUG__.
89 Andy Dougherty February 2000
5d94fbed 90*/
3609ea0d 91#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 92# ifndef PERL_BOOL_AS_CHAR
5d94fbed 93# ifndef HAS_BOOL
c1d22f6b 94# define HAS_BOOL 1
5d94fbed 95# endif
f789f6a4 96# endif
5d94fbed
AD
97#endif
98
99#ifndef HAS_BOOL
f789f6a4
FC
100# ifdef bool
101# undef bool
102# endif
70d5cb32 103# define bool char
c1d22f6b 104# define HAS_BOOL 1
a687059c 105#endif
0d3e774c 106
236415e2
KW
107/* cast-to-bool. A simple (bool) cast may not do the right thing: if bool is
108 * defined as char for example, then the cast from int is
109 * implementation-defined (bool)!!(cbool) in a ternary triggers a bug in xlc on
110 * AIX */
18f5643b 111#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 112
46c6c7e2 113/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
114 * XXX Should really be a Configure probe, with HAS__FUNCTION__
115 * and FUNCTION__ as results.
116 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
46c6c7e2
JH
117#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
118# define FUNCTION__ __func__
119#else
9ea40801 120# if (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
be73ebb0 121 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
46c6c7e2
JH
122# define FUNCTION__ ""
123# else
124# define FUNCTION__ __FUNCTION__ /* Common extension. */
125# endif
126#endif
127
27d4fb96
PP
128/* XXX A note on the perl source internal type system. The
129 original intent was that I32 be *exactly* 32 bits.
130
131 Currently, we only guarantee that I32 is *at least* 32 bits.
132 Specifically, if int is 64 bits, then so is I32. (This is the case
133 for the Cray.) This has the advantage of meshing nicely with
134 standard library calls (where we pass an I32 and the library is
135 expecting an int), but the disadvantage that an I32 is not 32 bits.
136 Andy Dougherty August 1996
24fef2a7 137
dc45a647
MB
138 There is no guarantee that there is *any* integral type with
139 exactly 32 bits. It is perfectly legal for a system to have
140 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 141
dc45a647
MB
142 Similarly, there is no guarantee that I16 and U16 have exactly 16
143 bits.
693762b4 144
8e84507e
NIS
145 For dealing with issues that may arise from various 32/64-bit
146 systems, we will ask Configure to check out
8175356b 147
3609ea0d
JH
148 SHORTSIZE == sizeof(short)
149 INTSIZE == sizeof(int)
150 LONGSIZE == sizeof(long)
dc45a647 151 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 152 PTRSIZE == sizeof(void *)
dc45a647
MB
153 DOUBLESIZE == sizeof(double)
154 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 155
27d4fb96
PP
156*/
157
69512466
JH
158#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
159# include <inttypes.h>
dd0eed91
JH
160# ifdef INT32_MIN_BROKEN
161# undef INT32_MIN
162# define INT32_MIN (-2147483647-1)
163# endif
164# ifdef INT64_MIN_BROKEN
165# undef INT64_MIN
166# define INT64_MIN (-9223372036854775807LL-1)
167# endif
69512466
JH
168#endif
169
8175356b
JH
170typedef I8TYPE I8;
171typedef U8TYPE U8;
172typedef I16TYPE I16;
173typedef U16TYPE U16;
174typedef I32TYPE I32;
175typedef U32TYPE U32;
16d89be8
JH
176
177#ifdef HAS_QUAD
8175356b
JH
178typedef I64TYPE I64;
179typedef U64TYPE U64;
16d89be8 180#endif
8175356b 181
9c39b21c
JH
182/* INT64_C/UINT64_C are C99 from <stdint.h> (so they will not be
183 * available in strict C89 mode), but they are nice, so let's define
184 * them if necessary. */
0d3bd0a9 185#if defined(HAS_QUAD)
9c39b21c
JH
186# undef PeRl_INT64_C
187# undef PeRl_UINT64_C
188/* Prefer the native integer types (int and long) over long long
189 * (which is not C89) and Win32-specific __int64. */
190# if QUADKIND == QUAD_IS_INT && INTSIZE == 8
191# define PeRl_INT64_C(c) (c)
192# define PeRl_UINT64_C(c) CAT2(c,U)
193# endif
194# if QUADKIND == QUAD_IS_LONG && LONGSIZE == 8
195# define PeRl_INT64_C(c) CAT2(c,L)
196# define PeRl_UINT64_C(c) CAT2(c,UL)
197# endif
198# if QUADKIND == QUAD_IS_LONG_LONG && defined(HAS_LONG_LONG)
199# define PeRl_INT64_C(c) CAT2(c,LL)
200# define PeRl_UINT64_C(c) CAT2(c,ULL)
201# endif
202# if QUADKIND == QUAD_IS___INT64
203# define PeRl_INT64_C(c) CAT2(c,I64)
204# define PeRl_UINT64_C(c) CAT2(c,UI64)
205# endif
206# ifndef PeRl_INT64_C
702c92eb
JH
207# define PeRl_INT64_C(c) ((I64)(c)) /* last resort */
208# define PeRl_UINT64_C(c) ((U64)(c))
9c39b21c
JH
209# endif
210/* In OS X the INT64_C/UINT64_C are defined with LL/ULL, which will
211 * not fly with C89-pedantic gcc, so let's undefine them first so that
212 * we can redefine them with our native integer preferring versions. */
8eafc6eb 213# if defined(PERL_DARWIN) && defined(PERL_GCC_PEDANTIC)
9c39b21c
JH
214# undef INT64_C
215# undef UINT64_C
216# endif
217# ifndef INT64_C
218# define INT64_C(c) PeRl_INT64_C(c)
219# endif
220# ifndef UINT64_C
221# define UINT64_C(c) PeRl_UINT64_C(c)
222# endif
e8c95190 223#endif
dc750f2e 224
d8668976 225#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 226
5ff3f7a4
GS
227/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
228 Please search CHAR_MAX in perl.h for further details. */
229#define U8_MAX UINT8_MAX
230#define U8_MIN UINT8_MIN
231
5ff3f7a4
GS
232#define I16_MAX INT16_MAX
233#define I16_MIN INT16_MIN
234#define U16_MAX UINT16_MAX
235#define U16_MIN UINT16_MIN
236
5ff3f7a4
GS
237#define I32_MAX INT32_MAX
238#define I32_MIN INT32_MIN
0e983133
GS
239#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
240# define U32_MAX UINT32_MAX
241#else
242# define U32_MAX 4294967295U
243#endif
5ff3f7a4
GS
244#define U32_MIN UINT32_MIN
245
246#else
247
5c9fa16e
KA
248/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
249 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
250#define U8_MAX PERL_UCHAR_MAX
251#define U8_MIN PERL_UCHAR_MIN
79072805 252
27d4fb96
PP
253#define I16_MAX PERL_SHORT_MAX
254#define I16_MIN PERL_SHORT_MIN
255#define U16_MAX PERL_USHORT_MAX
256#define U16_MIN PERL_USHORT_MIN
79072805 257
c4f23d77 258#if LONGSIZE > 4
27d4fb96
PP
259# define I32_MAX PERL_INT_MAX
260# define I32_MIN PERL_INT_MIN
261# define U32_MAX PERL_UINT_MAX
262# define U32_MIN PERL_UINT_MIN
79072805 263#else
27d4fb96
PP
264# define I32_MAX PERL_LONG_MAX
265# define I32_MIN PERL_LONG_MIN
266# define U32_MAX PERL_ULONG_MAX
267# define U32_MIN PERL_ULONG_MIN
79072805
LW
268#endif
269
5ff3f7a4
GS
270#endif
271
58a9a5d5 272/* log(2) is pretty close to 0.30103, just in case anyone is grepping for it */
fc36a67e
PP
273#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
274#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
275#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
276
88794300 277/* Unused by core; should be deprecated */
ff68c719 278#define Ctl(ch) ((ch) & 037)
8d063cd8 279
4650c663
KW
280/* This is a helper macro to avoid preprocessor issues, replaced by nothing
281 * unless under DEBUGGING, where it expands to an assert of its argument,
282 * followed by a comma (hence the comma operator). If we just used a straight
3e94db23
JH
283 * assert(), we would get a comma with nothing before it when not DEBUGGING.
284 *
285 * We also use empty definition under Coverity since the __ASSERT__
286 * checks often check for things that Really Cannot Happen, and Coverity
287 * detects that and gets all excited. */
288
289#if defined(DEBUGGING) && !defined(__COVERITY__)
0f092d08
KW
290# define __ASSERT_(statement) assert(statement),
291#else
292# define __ASSERT_(statement)
293#endif
294
3fe05580
MHM
295/*
296=head1 SV-Body Allocation
297
298=for apidoc Ama|SV*|newSVpvs|const char* s
30a15352
KW
299Like C<newSVpvn>, but takes a literal C<NUL>-terminated string instead of a
300string/length pair.
3fe05580 301
84bafc02 302=for apidoc Ama|SV*|newSVpvs_flags|const char* s|U32 flags
30a15352
KW
303Like C<newSVpvn_flags>, but takes a literal C<NUL>-terminated string instead of
304a string/length pair.
84bafc02 305
3fe05580 306=for apidoc Ama|SV*|newSVpvs_share|const char* s
30a15352
KW
307Like C<newSVpvn_share>, but takes a literal C<NUL>-terminated string instead of
308a string/length pair and omits the hash parameter.
3fe05580 309
9dcc53ea 310=for apidoc Am|void|sv_catpvs_flags|SV* sv|const char* s|I32 flags
30a15352
KW
311Like C<sv_catpvn_flags>, but takes a literal C<NUL>-terminated string instead
312of a string/length pair.
9dcc53ea
Z
313
314=for apidoc Am|void|sv_catpvs_nomg|SV* sv|const char* s
315Like C<sv_catpvn_nomg>, but takes a literal string instead of a
316string/length pair.
317
7b4e7469 318=for apidoc Am|void|sv_catpvs|SV* sv|const char* s
3fe05580
MHM
319Like C<sv_catpvn>, but takes a literal string instead of a string/length pair.
320
9dcc53ea
Z
321=for apidoc Am|void|sv_catpvs_mg|SV* sv|const char* s
322Like C<sv_catpvn_mg>, but takes a literal string instead of a
323string/length pair.
324
7b4e7469 325=for apidoc Am|void|sv_setpvs|SV* sv|const char* s
3fe05580
MHM
326Like C<sv_setpvn>, but takes a literal string instead of a string/length pair.
327
9dcc53ea
Z
328=for apidoc Am|void|sv_setpvs_mg|SV* sv|const char* s
329Like C<sv_setpvn_mg>, but takes a literal string instead of a
330string/length pair.
331
332=for apidoc Am|SV *|sv_setref_pvs|const char* s
333Like C<sv_setref_pvn>, but takes a literal string instead of a
334string/length pair.
335
3fe05580
MHM
336=head1 Memory Management
337
338=for apidoc Ama|char*|savepvs|const char* s
30a15352
KW
339Like C<savepvn>, but takes a literal C<NUL>-terminated string instead of a
340string/length pair.
3fe05580 341
9dcc53ea
Z
342=for apidoc Ama|char*|savesharedpvs|const char* s
343A version of C<savepvs()> which allocates the duplicate string in memory
344which is shared between threads.
345
3fe05580
MHM
346=head1 GV Functions
347
348=for apidoc Am|HV*|gv_stashpvs|const char* name|I32 create
349Like C<gv_stashpvn>, but takes a literal string instead of a string/length pair.
350
351=head1 Hash Manipulation Functions
352
353=for apidoc Am|SV**|hv_fetchs|HV* tb|const char* key|I32 lval
354Like C<hv_fetch>, but takes a literal string instead of a string/length pair.
355
356=for apidoc Am|SV**|hv_stores|HV* tb|const char* key|NULLOK SV* val
357Like C<hv_store>, but takes a literal string instead of a string/length pair
358and omits the hash parameter.
359
510966aa
Z
360=head1 Lexer interface
361
362=for apidoc Amx|void|lex_stuff_pvs|const char *pv|U32 flags
363
364Like L</lex_stuff_pvn>, but takes a literal string instead of a
365string/length pair.
366
3fe05580
MHM
367=cut
368*/
369
926bb54c
KW
370/* concatenating with "" ensures that only literal strings are accepted as
371 * argument */
26bfd924 372#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
2efa8cc7 373
926bb54c
KW
374/* note that STR_WITH_LEN() can't be used as argument to macros or functions
375 * that under some configurations might be macros, which means that it requires
376 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
ba3a79e7
GA
377 */
378
379/* STR_WITH_LEN() shortcuts */
380#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
381#define newSVpvs_flags(str,flags) \
382 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 383#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
384#define sv_catpvs_flags(sv, str, flags) \
385 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
386#define sv_catpvs_nomg(sv, str) \
387 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
388#define sv_catpvs(sv, str) \
389 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
390#define sv_catpvs_mg(sv, str) \
391 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 392#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
393#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
394#define sv_setref_pvs(rv, classname, str) \
395 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 396#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
397#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
398#define gv_stashpvs(str, create) \
399 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
400#define gv_fetchpvs(namebeg, add, sv_type) \
401 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
402#define gv_fetchpvn(namebeg, len, add, sv_type) \
403 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
404#define sv_catxmlpvs(dsv, str, utf8) \
405 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
a038e571
NC
406#define hv_fetchs(hv,key,lval) \
407 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
408 (lval) ? (HV_FETCH_JUST_SV | HV_FETCH_LVALUE) \
409 : HV_FETCH_JUST_SV, NULL, 0))
410
411#define hv_stores(hv,key,val) \
412 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
413 (HV_FETCH_ISSTORE|HV_FETCH_JUST_SV), (val), 0))
ba3a79e7 414
510966aa
Z
415#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
416
b96d8cd9
NC
417#define get_cvs(str, flags) \
418 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 419
954c1994 420/*
ccfc67b7
JH
421=head1 Miscellaneous Functions
422
954c1994
GS
423=for apidoc Am|bool|strNE|char* s1|char* s2
424Test two strings to see if they are different. Returns true or
425false.
426
427=for apidoc Am|bool|strEQ|char* s1|char* s2
428Test two strings to see if they are equal. Returns true or false.
429
430=for apidoc Am|bool|strLT|char* s1|char* s2
431Test two strings to see if the first, C<s1>, is less than the second,
432C<s2>. Returns true or false.
433
434=for apidoc Am|bool|strLE|char* s1|char* s2
435Test two strings to see if the first, C<s1>, is less than or equal to the
436second, C<s2>. Returns true or false.
437
438=for apidoc Am|bool|strGT|char* s1|char* s2
439Test two strings to see if the first, C<s1>, is greater than the second,
440C<s2>. Returns true or false.
441
442=for apidoc Am|bool|strGE|char* s1|char* s2
443Test two strings to see if the first, C<s1>, is greater than or equal to
444the second, C<s2>. Returns true or false.
445
446=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
447Test two strings to see if they are different. The C<len> parameter
72d33970 448indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
449wrapper for C<strncmp>).
450
451=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
452Test two strings to see if they are equal. The C<len> parameter indicates
72d33970 453the number of bytes to compare. Returns true or false. (A wrapper for
954c1994
GS
454C<strncmp>).
455
bd18bd40
KW
456=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
457Test two buffers (which may contain embedded C<NUL> characters, to see if they
458are equal. The C<len> parameter indicates the number of bytes to compare.
459Returns zero if equal, or non-zero if non-equal.
460
461=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
462Test two buffers (which may contain embedded C<NUL> characters, to see if they
463are not equal. The C<len> parameter indicates the number of bytes to compare.
464Returns zero if non-equal, or non-zero if equal.
465
954c1994
GS
466=cut
467*/
468
8d063cd8
LW
469#define strNE(s1,s2) (strcmp(s1,s2))
470#define strEQ(s1,s2) (!strcmp(s1,s2))
471#define strLT(s1,s2) (strcmp(s1,s2) < 0)
472#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
473#define strGT(s1,s2) (strcmp(s1,s2) > 0)
474#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
475#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
476#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
378cc40b 477
36477c24
PP
478#ifdef HAS_MEMCMP
479# define memNE(s1,s2,l) (memcmp(s1,s2,l))
480# define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
481#else
482# define memNE(s1,s2,l) (bcmp(s1,s2,l))
483# define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
484#endif
485
568a785a 486#define memEQs(s1, l, s2) \
8e6497b6 487 (sizeof(s2)-1 == l && memEQ(s1, ("" s2 ""), (sizeof(s2)-1)))
568a785a
NC
488#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
489
bbce6d69
PP
490/*
491 * Character classes.
492 *
493 * Unfortunately, the introduction of locales means that we
494 * can't trust isupper(), etc. to tell the truth. And when
495 * it comes to /\w+/ with tainting enabled, we *must* be able
496 * to trust our character classes.
497 *
498 * Therefore, the default tests in the text of Perl will be
499 * independent of locale. Any code that wants to depend on
500 * the current locale will use the tests that begin with "lc".
501 */
502
2304df62
AD
503#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
504# ifndef CTYPE256
505# define CTYPE256
506# endif
507#endif
508
954c1994 509/*
ccfc67b7 510
dcccc8ff 511=head1 Character classification
243effed
KW
512This section is about functions (really macros) that classify characters
513into types, such as punctuation versus alphabetic, etc. Most of these are
514analogous to regular expression character classes. (See
515L<perlrecharclass/POSIX Character Classes>.) There are several variants for
516each class. (Not all macros have all variants; each item below lists the
517ones valid for it.) None are affected by C<use bytes>, and only the ones
518with C<LC> in the name are affected by the current locale.
519
520The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
521C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
522represented by that octet is (or on non-ASCII platforms, corresponds to) an
523ASCII character in the named class based on platform, Unicode, and Perl rules.
524If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 525
6aff1f14
KW
526Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
527with no suffix C<"_A">.
4b9734bf
KW
528
529Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
530onto the platform. That is, the code points that are ASCII are unaffected,
531since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
532as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
533true when called with the code point 0xDF, which is a word character in both
4650c663 534ASCII and EBCDIC (though it represents different characters in each).
243effed
KW
535
536Variant C<isFOO_uni> is like the C<isFOO_L1> variant, but accepts any UV code
537point as input. If the code point is larger than 255, Unicode rules are used
538to determine if it is in the character class. For example,
6aff1f14
KW
539C<isWORDCHAR_uni(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
540WITH MACRON in Unicode, and is a word character.
243effed
KW
541
542Variant C<isFOO_utf8> is like C<isFOO_uni>, but the input is a pointer to a
543(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
6aff1f14
KW
544classification of just the first (possibly multi-byte) character in the string
545is tested.
243effed 546
31f05a37
KW
547Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
548result is based on the current locale, which is what C<LC> in the name stands
549for. If Perl can determine that the current locale is a UTF-8 locale, it uses
550the published Unicode rules; otherwise, it uses the C library function that
551gives the named classification. For example, C<isDIGIT_LC()> when not in a
552UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
553returned if the input won't fit into an octet. On some platforms where the C
554library function is known to be defective, Perl changes its result to follow
555the POSIX standard's rules.
243effed
KW
556
557Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
558returns the same as C<isFOO_LC> for input code points less than 256, and
559returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
560
b7d90381
KW
561Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer
562to a (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
6aff1f14
KW
563classification of just the first (possibly multi-byte) character in the string
564is tested.
ccfc67b7 565
954c1994 566=for apidoc Am|bool|isALPHA|char ch
2787a470 567Returns a boolean indicating whether the specified character is an
6aff1f14 568alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
569See the L<top of this section|/Character classification> for an explanation of
570variants
e5ad6aba 571C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uni>, C<isALPHA_utf8>, C<isALPHA_LC>,
243effed 572C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
8a58bdcf 573
15861f94
KW
574=for apidoc Am|bool|isALPHANUMERIC|char ch
575Returns a boolean indicating whether the specified character is a either an
6aff1f14 576alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
577See the L<top of this section|/Character classification> for an explanation of
578variants
15861f94 579C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uni>,
e5ad6aba 580C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
15861f94
KW
581C<isALPHANUMERIC_LC_utf8>.
582
8a58bdcf
KW
583=for apidoc Am|bool|isASCII|char ch
584Returns a boolean indicating whether the specified character is one of the 128
243effed 585characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 586On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
587character corresponds to an ASCII character. Variants C<isASCII_A()> and
588C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
589See the L<top of this section|/Character classification> for an explanation of
590variants
243effed
KW
591C<isASCII_uni>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
592C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
593library routine C<isascii()>. In these cases, the variants whose names contain
594C<LC> are the same as the corresponding ones without.
595
d98532ea
KW
596Also note, that because all ASCII characters are UTF-8 invariant (meaning they
597have the exact same representation (always a single byte) whether encoded in
598UTF-8 or not), C<isASCII> will give the correct results when called with any
599byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
600work properly on any string encoded or not in UTF-8.
601
243effed
KW
602=for apidoc Am|bool|isBLANK|char ch
603Returns a boolean indicating whether the specified character is a
6aff1f14 604character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
605See the L<top of this section|/Character classification> for an explanation of
606variants
e5ad6aba 607C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uni>, C<isBLANK_utf8>, C<isBLANK_LC>,
243effed
KW
608C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
609platforms do not have the C library routine C<isblank()>. In these cases, the
610variants whose names contain C<LC> are the same as the corresponding ones
611without.
612
613=for apidoc Am|bool|isCNTRL|char ch
614Returns a boolean indicating whether the specified character is a
6aff1f14 615control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
616See the L<top of this section|/Character classification> for an explanation of
617variants
e5ad6aba 618C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uni>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
6aff1f14
KW
619C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
620On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 621
954c1994 622=for apidoc Am|bool|isDIGIT|char ch
2787a470 623Returns a boolean indicating whether the specified character is a
6aff1f14 624digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 625Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
626See the L<top of this section|/Character classification> for an explanation of
627variants
e5ad6aba 628C<isDIGIT_uni>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
243effed
KW
629C<isDIGIT_LC_utf8>.
630
631=for apidoc Am|bool|isGRAPH|char ch
632Returns a boolean indicating whether the specified character is a
6aff1f14 633graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff
KW
634See the L<top of this section|/Character classification> for an explanation of
635variants
e5ad6aba 636C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uni>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
243effed 637C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
954c1994 638
0c82b6df 639=for apidoc Am|bool|isLOWER|char ch
2787a470 640Returns a boolean indicating whether the specified character is a
6aff1f14 641lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
642See the L<top of this section|/Character classification> for an explanation of
643variants
e5ad6aba 644C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uni>, C<isLOWER_utf8>, C<isLOWER_LC>,
243effed 645C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
0c82b6df 646
c99e91e9 647=for apidoc Am|bool|isOCTAL|char ch
2787a470 648Returns a boolean indicating whether the specified character is an
6aff1f14 649octal digit, [0-7].
243effed
KW
650The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
651C<isOCTAL>.
652
653=for apidoc Am|bool|isPUNCT|char ch
654Returns a boolean indicating whether the specified character is a
6aff1f14
KW
655punctuation character, analogous to C<m/[[:punct:]]/>.
656Note that the definition of what is punctuation isn't as
243effed
KW
657straightforward as one might desire. See L<perlrecharclass/POSIX Character
658Classes> for details.
dcccc8ff
KW
659See the L<top of this section|/Character classification> for an explanation of
660variants
e5ad6aba 661C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uni>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
243effed 662C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
c99e91e9 663
0c82b6df 664=for apidoc Am|bool|isSPACE|char ch
2787a470 665Returns a boolean indicating whether the specified character is a
6aff1f14 666whitespace character. This is analogous
398d098a 667to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 668this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
669locale forms of this macro (the ones with C<LC> in their names) matched
670precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
671in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
672(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
673See the L<top of this section|/Character classification> for an explanation of
674variants
e5ad6aba 675C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uni>, C<isSPACE_utf8>, C<isSPACE_LC>,
243effed 676C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
0c82b6df 677
398d098a
KW
678=for apidoc Am|bool|isPSXSPC|char ch
679(short for Posix Space)
779cf272
KW
680Starting in 5.18, this is identical in all its forms to the
681corresponding C<isSPACE()> macros.
398d098a
KW
682The locale forms of this macro are identical to their corresponding
683C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
684non-locale forms differ from their C<isSPACE()> forms only in that the
685C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
686Otherwise they are identical. Thus this macro is analogous to what
687C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 688See the L<top of this section|/Character classification> for an explanation of
b7d90381
KW
689variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uni>, C<isPSXSPC_utf8>,
690C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
398d098a 691
954c1994 692=for apidoc Am|bool|isUPPER|char ch
2787a470 693Returns a boolean indicating whether the specified character is an
6aff1f14 694uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff
KW
695See the L<top of this section|/Character classification> for an explanation of
696variants
e5ad6aba 697C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uni>, C<isUPPER_utf8>, C<isUPPER_LC>,
243effed 698C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
954c1994 699
243effed 700=for apidoc Am|bool|isPRINT|char ch
8eea39dd 701Returns a boolean indicating whether the specified character is a
6aff1f14 702printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
703See the L<top of this section|/Character classification> for an explanation of
704variants
e5ad6aba 705C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uni>, C<isPRINT_utf8>, C<isPRINT_LC>,
243effed
KW
706C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
707
708=for apidoc Am|bool|isWORDCHAR|char ch
709Returns a boolean indicating whether the specified character is a character
710that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
711in a regular expression. A word character is an alphabetic character, a
712decimal digit, a connecting punctuation character (such as an underscore), or
713a "mark" character that attaches to one of those (like some sort of accent).
714C<isALNUM()> is a synonym provided for backward compatibility, even though a
715word character includes more than the standard C language meaning of
716alphanumeric.
dcccc8ff
KW
717See the L<top of this section|/Character classification> for an explanation of
718variants
1a83413c
KW
719C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uni>, and C<isWORDCHAR_utf8>.
720C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
721described there, but additionally include the platform's native underscore.
8a58bdcf
KW
722
723=for apidoc Am|bool|isXDIGIT|char ch
724Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
725digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
726and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
727See the L<top of this section|/Character classification> for an explanation of
728variants
243effed
KW
729C<isXDIGIT_uni>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
730C<isXDIGIT_LC_utf8>.
731
3c3ecf18
KW
732=for apidoc Am|bool|isIDFIRST|char ch
733Returns a boolean indicating whether the specified character can be the first
734character of an identifier. This is very close to, but not quite the same as
735the official Unicode property C<XID_Start>. The difference is that this
736returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
737See the L<top of this section|/Character classification> for an explanation of
738variants
3c3ecf18
KW
739C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uni>, C<isIDFIRST_utf8>,
740C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
741
742=for apidoc Am|bool|isIDCONT|char ch
743Returns a boolean indicating whether the specified character can be the
744second or succeeding character of an identifier. This is very close to, but
745not quite the same as the official Unicode property C<XID_Continue>. The
746difference is that this returns true only if the input character also matches
dcccc8ff
KW
747L</isWORDCHAR>. See the L<top of this section|/Character classification> for
748an
3c3ecf18
KW
749explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uni>,
750C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
751C<isIDCONT_LC_utf8>.
752
243effed 753=head1 Miscellaneous Functions
8eea39dd 754
95a59cab 755=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 756Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
757Behaviour is only well defined when isXDIGIT(*str) is true.
758
e7c1e6c1
KW
759=head1 Character case changing
760
1f607577
KW
761=for apidoc Am|U8|toUPPER|U8 ch
762Converts the specified character to uppercase. If the input is anything but an
763ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 764C<toUPPER_A> is equivalent.
954c1994 765
1f607577
KW
766=for apidoc Am|UV|toUPPER_uni|UV cp|U8* s|STRLEN* lenp
767Converts the Unicode code point C<cp> to its uppercase version, and
768stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
769that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
770bytes since the uppercase version may be longer than the original character.
771
772The first code point of the uppercased version is returned
773(but note, as explained just above, that there may be more.)
774
775=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
776Converts the UTF-8 encoded character at C<p> to its uppercase version, and
777stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
778that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
779bytes since the uppercase version may be longer than the original character.
780
781The first code point of the uppercased version is returned
782(but note, as explained just above, that there may be more.)
783
784The input character at C<p> is assumed to be well-formed.
785
25200305
KW
786=for apidoc Am|U8|toFOLD|U8 ch
787Converts the specified character to foldcase. If the input is anything but an
788ASCII uppercase character, that input character itself is returned. Variant
789C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
790Latin1 range, as the full generality of L</toFOLD_uni> is needed there.)
791
1f607577
KW
792=for apidoc Am|UV|toFOLD_uni|UV cp|U8* s|STRLEN* lenp
793Converts the Unicode code point C<cp> to its foldcase version, and
794stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
795that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
796bytes since the foldcase version may be longer than the original character.
797
798The first code point of the foldcased version is returned
799(but note, as explained just above, that there may be more.)
800
801=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
802Converts the UTF-8 encoded character at C<p> to its foldcase version, and
803stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
804that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
805bytes since the foldcase version may be longer than the original character.
806
807The first code point of the foldcased version is returned
808(but note, as explained just above, that there may be more.)
809
810The input character at C<p> is assumed to be well-formed.
811
812=for apidoc Am|U8|toLOWER|U8 ch
813Converts the specified character to lowercase. If the input is anything but an
814ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 815C<toLOWER_A> is equivalent.
954c1994 816
1f607577 817=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
818Converts the specified Latin1 character to lowercase. The results are
819undefined if the input doesn't fit in a byte.
1f607577
KW
820
821=for apidoc Am|U8|toLOWER_LC|U8 ch
822Converts the specified character to lowercase using the current locale's rules,
823if possible; otherwise returns the input character itself.
824
825=for apidoc Am|UV|toLOWER_uni|UV cp|U8* s|STRLEN* lenp
826Converts the Unicode code point C<cp> to its lowercase version, and
827stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
828that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
829bytes since the lowercase version may be longer than the original character.
830
831The first code point of the lowercased version is returned
832(but note, as explained just above, that there may be more.)
833
834=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
835Converts the UTF-8 encoded character at C<p> to its lowercase version, and
836stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
837that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
838bytes since the lowercase version may be longer than the original character.
839
840The first code point of the lowercased version is returned
841(but note, as explained just above, that there may be more.)
842
843The input character at C<p> is assumed to be well-formed.
844
25200305
KW
845=for apidoc Am|U8|toTITLE|U8 ch
846Converts the specified character to titlecase. If the input is anything but an
847ASCII lowercase character, that input character itself is returned. Variant
b7d90381
KW
848C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
849range, as the full generality of L</toTITLE_uni> is needed there. Titlecase is
850not a concept used in locale handling, so there is no functionality for that.)
25200305 851
1f607577
KW
852=for apidoc Am|UV|toTITLE_uni|UV cp|U8* s|STRLEN* lenp
853Converts the Unicode code point C<cp> to its titlecase version, and
854stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
855that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
856bytes since the titlecase version may be longer than the original character.
857
858The first code point of the titlecased version is returned
859(but note, as explained just above, that there may be more.)
860
861=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
862Converts the UTF-8 encoded character at C<p> to its titlecase version, and
863stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
864that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
865bytes since the titlecase version may be longer than the original character.
866
867The first code point of the titlecased version is returned
868(but note, as explained just above, that there may be more.)
869
870The input character at C<p> is assumed to be well-formed.
871
954c1994 872=cut
353c9b6f 873
1f607577 874XXX Still undocumented isVERTWS_uni and _utf8; it's unclear what their names
25200305 875really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 876
8a58bdcf 877Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
878patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
879
954c1994
GS
880*/
881
7c062697
KW
882/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
883 * is known only in the perl core, and this macro can be called from outside
884 * that */
885#ifdef HAS_QUAD
886# define WIDEST_UTYPE U64TYPE
887#else
888# define WIDEST_UTYPE U32
889#endif
890
3912bc88
KW
891/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
892 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
893 * bits of information are lost even on a 64-bit machine, but to get the
894 * compiler to optimize it out if possible. This is because Configure makes
895 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
896 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
897 *
898 * For Coverity, be always true, because otherwise Coverity thinks
899 * it finds several expressions that are always true, independent
900 * of operands. Well, they are, but that is kind of the point.
220c71bf 901 */
7e75d1a1 902#ifndef __COVERITY__
220c71bf 903#define FITS_IN_8_BITS(c) ((sizeof(c) == 1) || !(((WIDEST_UTYPE)(c)) & ~0xFF))
7e75d1a1
JH
904#else
905#define FITS_IN_8_BITS(c) (1)
906#endif
cf301eb7 907
41f43cc2 908#ifdef EBCDIC
b6340bd0 909# ifndef _ALL_SOURCE
0852beac
KW
910 /* The native libc isascii() et.al. functions return the wrong results
911 * on at least z/OS unless this is defined. */
b6340bd0
KW
912# error _ALL_SOURCE should probably be defined
913# endif
41f43cc2 914#else
0852beac
KW
915 /* There is a simple definition of ASCII for ASCII platforms. But the
916 * EBCDIC one isn't so simple, so is defined using table look-up like the
917 * other macros below */
41f43cc2
KW
918# define isASCII(c) ((WIDEST_UTYPE)(c) < 128)
919#endif
920
c2da0b36
KW
921/* The lower 3 bits in both the ASCII and EBCDIC representations of '0' are 0,
922 * and the 8 possible permutations of those bits exactly comprise the 8 octal
923 * digits */
3d33b3e3 924#define isOCTAL_A(c) cBOOL(FITS_IN_8_BITS(c) && (0xF8 & (c)) == '0')
c2da0b36 925
9fb1bf9d 926#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 927
a500dc72
KW
928/* Character class numbers. For internal core Perl use only. The ones less
929 * than 32 are used in PL_charclass[] and the ones up through the one that
930 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
931 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
932 * their actual definitions are here. If that file has a name not used here,
933 * it won't compile.
1709d539
KW
934 *
935 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
936 * frequency of their use. This gives a slight edge to exiting a loop earlier
937 * (in reginclass() in regexec.c) */
1709d539
KW
938# define _CC_WORDCHAR 0 /* \w and [:word:] */
939# define _CC_DIGIT 1 /* \d and [:digit:] */
940# define _CC_ALPHA 2 /* [:alpha:] */
941# define _CC_LOWER 3 /* [:lower:] */
942# define _CC_UPPER 4 /* [:upper:] */
943# define _CC_PUNCT 5 /* [:punct:] */
944# define _CC_PRINT 6 /* [:print:] */
15861f94 945# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 946# define _CC_GRAPH 8 /* [:graph:] */
b0d691b2 947# define _CC_CASED 9 /* [:lower:] and [:upper:] under /i */
1709d539 948
b0d691b2 949#define _FIRST_NON_SWASH_CC 10
1709d539
KW
950/* The character classes above are implemented with swashes. The second group
951 * (just below) contains the ones implemented without. These are also sorted
952 * in rough order of the frequency of their use, except that \v should be last,
953 * as it isn't a real Posix character class, and some (small) inefficiencies in
954 * regular expression handling would be introduced by putting it in the middle
955 * of those that are. Also, cntrl and ascii come after the others as it may be
956 * useful to group these which have no members that match above Latin1, (or
957 * above ASCII in the latter case) */
958
779cf272 959# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
960# define _CC_BLANK 11 /* [:blank:] */
961# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
962# define _CC_CNTRL 13 /* [:cntrl:] */
963# define _CC_ASCII 14 /* [:ascii:] */
964# define _CC_VERTSPACE 15 /* \v */
1709d539 965
a0947d7b
KW
966# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
967
1709d539 968/* The members of the third group below do not need to be coordinated with data
3ffc8c70 969 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
970# define _CC_IDFIRST 16
971# define _CC_CHARNAME_CONT 17
972# define _CC_NONLATIN1_FOLD 18
973# define _CC_NONLATIN1_SIMPLE_FOLD 19
974# define _CC_QUOTEMETA 20
975# define _CC_NON_FINAL_FOLD 21
976# define _CC_IS_IN_SOME_FOLD 22
977# define _CC_MNEMONIC_CNTRL 23
073c22b3
KW
978
979/* This next group is only used on EBCDIC platforms, so theoretically could be
980 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
981# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
982# define _CC_UTF8_IS_START 29
983# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
984# define _CC_UTF8_IS_CONTINUATION 31
985/* Unused: 24-27
f4cdb42c
KW
986 * If more bits are needed, one could add a second word for non-64bit
987 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
988 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
989 * is used only for optimization (as of this writing), and differs in the
990 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 991 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
992 * would just cause /i regexes which match them to run less efficiently.
993 * Similarly the EBCDIC-only bits are used just for speed, and could be
994 * replaced by other means */
96ac0975 995
3a371f2f
KW
996#if defined(PERL_CORE) || defined(PERL_EXT)
997/* An enum version of the character class numbers, to help compilers
998 * optimize */
999typedef enum {
3a371f2f 1000 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1001 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1002 _CC_ENUM_ASCII = _CC_ASCII,
1003 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1004 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1005 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1006 _CC_ENUM_DIGIT = _CC_DIGIT,
1007 _CC_ENUM_GRAPH = _CC_GRAPH,
1008 _CC_ENUM_LOWER = _CC_LOWER,
1009 _CC_ENUM_PRINT = _CC_PRINT,
1010 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1011 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1012 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1013 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1014 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1015 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1016} _char_class_number;
1017#endif
1018
63c61c3f 1019#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1020#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1021
b7d90381
KW
1022#if defined(PERL_IN_UTF8_C) \
1023 || defined(PERL_IN_REGCOMP_C) \
1024 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1025# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1026 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1027 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1028 #error Need to adjust order of swash_property_names[]
1029# endif
1030
b44daeeb
KW
1031/* This is declared static in each of the few files that this is #defined for
1032 * to keep them from being publicly accessible. Hence there is a small amount
1033 * of wasted space */
1034
922e8cb4
KW
1035static const char* const swash_property_names[] = {
1036 "XPosixWord",
1037 "XPosixDigit",
1038 "XPosixAlpha",
1039 "XPosixLower",
1040 "XPosixUpper",
1041 "XPosixPunct",
1042 "XPosixPrint",
1043 "XPosixAlnum",
b0d691b2
KW
1044 "XPosixGraph",
1045 "Cased"
922e8cb4
KW
1046};
1047#endif
1048
6635f04f 1049START_EXTERN_C
96ac0975
NC
1050# ifdef DOINIT
1051EXTCONST U32 PL_charclass[] = {
1052# include "l1_char_class_tab.h"
1053};
1054
1055# else /* ! DOINIT */
1056EXTCONST U32 PL_charclass[];
1057# endif
6635f04f 1058END_EXTERN_C
96ac0975 1059
265c1f46 1060 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1061# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1062
1063 /* For internal core Perl use only: the base macro for defining macros like
1064 * isALPHA */
ff7ecfc3 1065# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1066 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1067
f4cdb42c
KW
1068 /* The mask for the _A versions of the macros; it just adds in the bit for
1069 * ASCII. */
1070# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1071
4650c663
KW
1072 /* For internal core Perl use only: the base macro for defining macros like
1073 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1074 * the ASCII bit are present */
b7d90381
KW
1075# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1076 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1077 == _CC_mask_A(classnum)))
f4cdb42c 1078
f4cdb42c 1079# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1080# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1081# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1082# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1083# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1084# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1085# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1086# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1087# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1088# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1089# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1090# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1091# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1092 */
d95f8b6a 1093# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1094# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1095# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1096# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1097
1098 /* continuation character for legal NAME in \N{NAME} */
1099# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1100
1101# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1102# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1103# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1104# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1105# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1106# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1107# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1108# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1109# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1110# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1111
0852beac
KW
1112# ifdef EBCDIC
1113# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1114# endif
1115
f12c0118
KW
1116 /* Participates in a single-character fold with a character above 255 */
1117# 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)))
1118
1119 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1120# 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 1121
4eeeb416 1122# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1123# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1124 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1125# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1126 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1127# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1128 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1129#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1130
1131 /* If we don't have perl.h, we are compiling a utility program. Below we
1132 * hard-code various macro definitions that wouldn't otherwise be available
0852beac
KW
1133 * to it. Most are coded based on first principals. First some ones common
1134 * to both ASCII and EBCDIC */
1135# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1136# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1137# define isSPACE_A(c) (isBLANK_A(c) \
1138 || (c) == '\n' \
1139 || (c) == '\r' \
1140 || (c) == '\v' \
1141 || (c) == '\f')
1142# ifdef EBCDIC /* There are gaps between 'i' and 'j'; 'r' and 's'. Same
1143 for uppercase. This is ordered to exclude most things
1144 early */
1145# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1146 && ((c) <= 'i' \
1147 || ((c) >= 'j' && (c) <= 'r') \
1148 || (c) >= 's'))
1149# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1150 && ((c) <= 'I' \
1151 || ((c) >= 'J' && (c) <= 'R') \
1152 || (c) >= 'S'))
1153# else /* ASCII platform. */
9fb1bf9d 1154# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z')
2181da29 1155# define isUPPER_A(c) ((c) <= 'Z' && (c) >= 'A')
0852beac
KW
1156# endif
1157
1158 /* Some more ASCII, non-ASCII common definitions */
a4d7a999
KW
1159# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1160# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1161# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac
KW
1162# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1163# define isXDIGIT_A(c) (isDIGIT_A(c) \
1164 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1165 || ((c) <= 'F' && (c) >= 'A'))
3ded5eb0 1166
0852beac
KW
1167# ifdef EBCDIC
1168# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1169 || (c) == '#' || (c) == '$' || (c) == '%' \
1170 || (c) == '&' || (c) == '\'' || (c) == '(' \
1171 || (c) == ')' || (c) == '*' || (c) == '+' \
1172 || (c) == ',' || (c) == '.' || (c) == '/' \
1173 || (c) == ':' || (c) == ';' || (c) == '<' \
1174 || (c) == '=' || (c) == '>' || (c) == '?' \
1175 || (c) == '@' || (c) == '[' || (c) == '\\' \
1176 || (c) == ']' || (c) == '^' || (c) == '_' \
1177 || (c) == '`' || (c) == '{' || (c) == '|' \
1178 || (c) == '}' || (c) == '~')
1179# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1180# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1181
1182# ifdef QUESTION_MARK_CTRL
1183# define _isQMC(c) ((c) == QUESTION_MARK_CTRL)
1184# else
1185# define _isQMC(c) 0
1186# endif
1187
1188 /* I (khw) can't think of a way to define all the ASCII controls
1189 * without resorting to a libc (locale-sensitive) call. But we know
1190 * that all controls but the question-mark one are in the range 0-0x3f.
1191 * This makes sure that all the controls that have names are included,
1192 * and all controls that are also considered ASCII in the locale. This
1193 * may include more or fewer than what it actually should, but the
1194 * wrong ones are less-important controls, so likely won't impact
1195 * things (keep in mind that this is compiled only if perl.h isn't
1196 * available). The question mark control is included if available */
1197# define isCNTRL_A(c) (((c) < 0x40 && isascii(c)) \
1198 || (c) == '\0' || (c) == '\a' || (c) == '\b' \
1199 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1200 || (c) == '\t' || (c) == '\v' || _isQMC(c))
1201
1202# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1203# else /* ASCII platform; things are simpler, and isASCII has already
1204 been defined */
1205# define isGRAPH_A(c) (((c) > ' ' && (c) < 127))
1206# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1207# define isPUNCT_A(c) (isGRAPH_A(c) && (! isALPHANUMERIC_A(c)))
1208# define isCNTRL_A(c) (isASCII(c) && (! isPRINT_A(c)))
1209# endif
1210
3ffc8c70 1211 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1212 * during debugging, and it seems best to keep them. We may be called
1213 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1214 * do anything anyway, so make it not a problem */
1215# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1216# define NATIVE_TO_LATIN1(ch) (ch)
1217# endif
3ded5eb0
KW
1218# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1219# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1220# define isBLANK_L1(c) (isBLANK_A(c) \
1221 || (FITS_IN_8_BITS(c) \
1222 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1223# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1224# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1225# define isLOWER_L1(c) (isLOWER_A(c) \
1226 || (FITS_IN_8_BITS(c) \
1227 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1228 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1229 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1230 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1231 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1232# define isPRINT_L1(c) (isPRINT_A(c) \
1233 || (FITS_IN_8_BITS(c) \
1234 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1235# define isPUNCT_L1(c) (isPUNCT_A(c) \
1236 || (FITS_IN_8_BITS(c) \
1237 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1238 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1239 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1240 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1241 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1242 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1243 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1244# define isSPACE_L1(c) (isSPACE_A(c) \
1245 || (FITS_IN_8_BITS(c) \
1246 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1247 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1248# define isUPPER_L1(c) (isUPPER_A(c) \
1249 || (FITS_IN_8_BITS(c) \
1250 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1251 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1252 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1253# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1254# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1255# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1256 || isBLANK_L1(c) \
1257 || (c) == '-' \
1258 || (c) == '(' \
1259 || (c) == ')')
1260 /* The following are not fully accurate in the above-ASCII range. I (khw)
1261 * don't think it's necessary to be so for the purposes where this gets
1262 * compiled */
1263# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1264# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1265
1266 /* And these aren't accurate at all. They are useful only for above
1267 * Latin1, which utilities and bootstrapping don't deal with */
1268# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1269# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1270# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1271
1272 /* Many of the macros later in this file are defined in terms of these. By
1273 * implementing them with a function, which converts the class number into
1274 * a call to the desired macro, all of the later ones work. However, that
1275 * function won't be actually defined when building a utility program (no
1276 * perl.h), and so a compiler error will be generated if one is attempted
1277 * to be used. And the above-Latin1 code points require Unicode tables to
1278 * be present, something unlikely to be the case when bootstrapping */
1279# define _generic_isCC(c, classnum) \
1280 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1281# define _generic_isCC_A(c, classnum) \
1282 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1283#endif /* End of no perl.h H_PERL */
8a58bdcf 1284
e66b99e9
KW
1285#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1286#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1287#define isASCII_A(c) isASCII(c)
1288#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1289#define isBLANK(c) isBLANK_A(c)
1290#define isCNTRL(c) isCNTRL_A(c)
1291#define isDIGIT(c) isDIGIT_A(c)
1292#define isGRAPH(c) isGRAPH_A(c)
1293#define isIDFIRST(c) isIDFIRST_A(c)
1294#define isLOWER(c) isLOWER_A(c)
1295#define isPRINT(c) isPRINT_A(c)
779cf272 1296#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1297#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1298#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1299#define isPUNCT(c) isPUNCT_A(c)
1300#define isSPACE(c) isSPACE_A(c)
1301#define isUPPER(c) isUPPER_A(c)
1302#define isWORDCHAR(c) isWORDCHAR_A(c)
1303#define isXDIGIT(c) isXDIGIT_A(c)
1304
1305/* ASCII casing. These could also be written as
1306 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1307 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1308 which uses table lookup and mask instead of subtraction. (This would
1309 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1310#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1311#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1312
25200305
KW
1313/* In the ASCII range, these are equivalent to what they're here defined to be.
1314 * But by creating these definitions, other code doesn't have to be aware of
1315 * this detail */
1316#define toFOLD(c) toLOWER(c)
25200305
KW
1317#define toTITLE(c) toUPPER(c)
1318
c753c8d3
KW
1319#define toLOWER_A(c) toLOWER(c)
1320#define toUPPER_A(c) toUPPER(c)
25200305
KW
1321#define toFOLD_A(c) toFOLD(c)
1322#define toTITLE_A(c) toTITLE(c)
1a0901db 1323
4650c663 1324/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1325#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1326 ? (c) \
f4cd282c 1327 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1328#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1329
1a0901db 1330/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1331 * all mapped to one of them, and these need special handling; returns the
1332 * input itself if is out-of-range */
b2bf251f 1333#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1334 ? (c) \
f4cd282c 1335 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1336#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1337
beab9ebe
KW
1338/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1339
1340/* For internal core Perl use only: the base macro for defining macros like
1341 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1342 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1343 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1344 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1345 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1346 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1347 * code points under locale rules, use the macros based on _generic_LC_uvchr
1348 * instead of this. */
beab9ebe
KW
1349#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1350 (! FITS_IN_8_BITS(c) \
1351 ? 0 \
31f05a37
KW
1352 : IN_UTF8_CTYPE_LOCALE \
1353 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1354 : cBOOL(non_utf8))
1355
1356/* For internal core Perl use only: a helper macro for defining macros like
1357 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1358 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1359 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1360 * _generic_LC_base for more info */
1361#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1362 _generic_LC_base(c,utf8_locale_classnum, \
1363 non_utf8_func( (_LC_CAST) (c)))
1364
1365/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1366 * 'c' is the platform's native underscore character */
1367#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1368 _generic_LC_base(c, utf8_locale_classnum, \
1369 (non_utf8_func( (_LC_CAST) (c)) \
1370 || (char)(c) == '_'))
1371
1372/* These next three are also for internal core Perl use only: case-change
1373 * helper macros */
1374#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1375 ? (c) \
31f05a37
KW
1376 : (IN_UTF8_CTYPE_LOCALE) \
1377 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1378 : (cast)function((cast)(c)))
beab9ebe 1379
31f05a37
KW
1380/* Note that the result can be larger than a byte in a UTF-8 locale. It
1381 * returns a single value, so can't adequately return the upper case of LATIN
1382 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1383 * values "SS"); instead it asserts against that under DEBUGGING, and
1384 * otherwise returns its input */
beab9ebe
KW
1385#define _generic_toUPPER_LC(c, function, cast) \
1386 (! FITS_IN_8_BITS(c) \
1387 ? (c) \
31f05a37 1388 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1389 ? (cast)function((cast)(c)) \
31f05a37
KW
1390 : ((((U8)(c)) == MICRO_SIGN) \
1391 ? GREEK_CAPITAL_LETTER_MU \
1392 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1393 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1394 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1395 ? (__ASSERT_(0) (c)) \
1396 : PL_mod_latin1_uc[ (U8) (c) ])))))
1397
1398/* Note that the result can be larger than a byte in a UTF-8 locale. It
1399 * returns a single value, so can't adequately return the fold case of LATIN
1400 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1401 * values "ss"); instead it asserts against that under DEBUGGING, and
1402 * otherwise returns its input */
beab9ebe 1403#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1404 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1405 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1406 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1407 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1408 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1409
84061b6a 1410/* Use the libc versions for these if available. */
f05550c0 1411#if defined(HAS_ISASCII)
84061b6a
KW
1412# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1413#else
1414# define isASCII_LC(c) isASCII(c)
1415#endif
1416
f05550c0 1417#if defined(HAS_ISBLANK)
84061b6a 1418# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1419#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1420# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1421#endif
1422
f05550c0 1423#define _LC_CAST U8
bbce6d69 1424
f05550c0 1425#ifdef WIN32
375f5f06
KW
1426 /* The Windows functions don't bother to follow the POSIX standard, which
1427 * for example says that something can't both be a printable and a control.
1428 * But Windows treats the \t control as a printable, and does such things
1429 * as making superscripts into both digits and punctuation. This tames
1430 * these flaws by assuming that the definitions of both controls and space
1431 * are correct, and then making sure that other definitions don't have
1432 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1433 * Not all possible weirdnesses are checked for, just the ones that were
1434 * detected on actual Microsoft code pages */
1435
b7d90381
KW
1436# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1437# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1438
1439# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1440 && isALPHANUMERIC_LC(c))
1441# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1442 ! isPUNCT_LC(c))
1443# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1444 isALPHANUMERIC_LC(c))
1445# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1446# define isIDFIRST_LC(c) (((c) == '_') \
1447 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1448# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1449# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1450# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1451# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1452# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1453# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1454 && isALPHANUMERIC_LC(c))
f05550c0
BF
1455
1456# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1457# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1458# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1459
1460#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1461 /* For most other platforms */
beab9ebe 1462
f05550c0
BF
1463# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1464# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1465# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1466# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1467# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1468# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1469# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1470# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1471# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1472# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1473# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1474# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1475# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1476
1477
1478# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1479# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1480# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1481
1482#else /* The final fallback position */
1483
b7d90381
KW
1484# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1485# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1486# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1487# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1488# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1489# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1490# define isLOWER_LC(c) (isascii(c) && islower(c))
1491# define isPRINT_LC(c) (isascii(c) && isprint(c))
1492# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1493# define isSPACE_LC(c) (isascii(c) && isspace(c))
1494# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1495# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1496# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1497
1498# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1499# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1500# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1501
f05550c0 1502#endif
55204971 1503
eba68aa0
KW
1504#define isIDCONT(c) isWORDCHAR(c)
1505#define isIDCONT_A(c) isWORDCHAR_A(c)
1506#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1507#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1508#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1509
4650c663
KW
1510/* For internal core Perl use only: the base macros for defining macros like
1511 * isALPHA_uni. 'c' is the code point to check. 'classnum' is the POSIX class
1512 * number defined earlier in this file. _generic_uni() is used for POSIX
1513 * classes where there is a macro or function 'above_latin1' that takes the
1514 * single argument 'c' and returns the desired value. These exist for those
1515 * classes which have simple definitions, avoiding the overhead of a hash
1516 * lookup or inversion list binary search. _generic_swash_uni() can be used
1517 * for classes where that overhead is faster than a direct lookup.
1518 * _generic_uni() won't compile if 'c' isn't unsigned, as it won't match the
1519 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1520 * have duplicate checks here, so could create versions of the macros that
1521 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1522
1523/* Note that all ignore 'use bytes' */
cd500f2f
KW
1524#define _generic_uni(classnum, above_latin1, c) ((c) < 256 \
1525 ? _generic_isCC(c, classnum) \
1526 : above_latin1(c))
1527#define _generic_swash_uni(classnum, c) ((c) < 256 \
1528 ? _generic_isCC(c, classnum) \
922e8cb4 1529 : _is_uni_FOO(classnum, c))
cd500f2f
KW
1530#define isALPHA_uni(c) _generic_swash_uni(_CC_ALPHA, c)
1531#define isALPHANUMERIC_uni(c) _generic_swash_uni(_CC_ALPHANUMERIC, c)
509fb054
KW
1532#define isASCII_uni(c) isASCII(c)
1533#define isBLANK_uni(c) _generic_uni(_CC_BLANK, is_HORIZWS_cp_high, c)
1534#define isCNTRL_uni(c) isCNTRL_L1(c) /* All controls are in Latin1 */
cd500f2f
KW
1535#define isDIGIT_uni(c) _generic_swash_uni(_CC_DIGIT, c)
1536#define isGRAPH_uni(c) _generic_swash_uni(_CC_GRAPH, c)
eba68aa0 1537#define isIDCONT_uni(c) _generic_uni(_CC_WORDCHAR, _is_uni_perl_idcont, c)
509fb054 1538#define isIDFIRST_uni(c) _generic_uni(_CC_IDFIRST, _is_uni_perl_idstart, c)
cd500f2f
KW
1539#define isLOWER_uni(c) _generic_swash_uni(_CC_LOWER, c)
1540#define isPRINT_uni(c) _generic_swash_uni(_CC_PRINT, c)
aea0c08c 1541
cd500f2f 1542#define isPUNCT_uni(c) _generic_swash_uni(_CC_PUNCT, c)
509fb054 1543#define isSPACE_uni(c) _generic_uni(_CC_SPACE, is_XPERLSPACE_cp_high, c)
779cf272
KW
1544#define isPSXSPC_uni(c) isSPACE_uni(c)
1545
cd500f2f 1546#define isUPPER_uni(c) _generic_swash_uni(_CC_UPPER, c)
509fb054 1547#define isVERTWS_uni(c) _generic_uni(_CC_VERTSPACE, is_VERTWS_cp_high, c)
cd500f2f 1548#define isWORDCHAR_uni(c) _generic_swash_uni(_CC_WORDCHAR, c)
509fb054 1549#define isXDIGIT_uni(c) _generic_uni(_CC_XDIGIT, is_XDIGIT_cp_high, c)
b18192e9 1550
b8d68ded 1551#define toFOLD_uni(c,s,l) to_uni_fold(c,s,l)
509fb054
KW
1552#define toLOWER_uni(c,s,l) to_uni_lower(c,s,l)
1553#define toTITLE_uni(c,s,l) to_uni_title(c,s,l)
1554#define toUPPER_uni(c,s,l) to_uni_upper(c,s,l)
a0ed51b3 1555
4650c663
KW
1556/* For internal core Perl use only: the base macros for defining macros like
1557 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1558 * point, not just 0-255. Like _generic_uni, there are two versions, one for
1559 * simple class definitions; the other for more complex. These are like
1560 * _generic_uni, so see it for more info. */
cd500f2f
KW
1561#define _generic_LC_uvchr(latin1, above_latin1, c) \
1562 (c < 256 ? latin1(c) : above_latin1(c))
1563#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1564 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1565
1566#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1567#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1568 _CC_ALPHANUMERIC, c)
b7d90381
KW
1569#define isASCII_LC_uvchr(c) isASCII_LC(c)
1570#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1571 is_HORIZWS_cp_high, c)
feeab5a9 1572#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1573#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1574#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1575#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1576 _is_uni_perl_idcont, c)
b7d90381 1577#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1578 _is_uni_perl_idstart, c)
1579#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1580#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1581#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1582#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1583#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1584 is_XPERLSPACE_cp_high, c)
cd500f2f 1585#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1586#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1587 _CC_WORDCHAR, c)
b7d90381
KW
1588#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1589 is_XDIGIT_cp_high, c)
e712593e 1590
b7d90381 1591#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1592
4650c663
KW
1593/* For internal core Perl use only: the base macros for defining macros like
1594 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1595 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1596 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1597 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1598 * same representation whether utf8 or not. Note that it assumes that the utf8
1599 * has been validated, and ignores 'use bytes' */
cd500f2f 1600#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1601 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1602 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1603 ? _generic_isCC( \
a62b247b 1604 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
aea0c08c 1605 *((p)+1 )), \
94bb8c36 1606 classnum) \
252810d0 1607 : utf8)
b7d90381
KW
1608/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1609 * 'above_latin1' can be a macro */
cd500f2f
KW
1610#define _generic_func_utf8(classnum, above_latin1, p) \
1611 _generic_utf8(classnum, p, above_latin1(p))
4650c663 1612/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1613 * 'above_latin1' parameter */
922e8cb4 1614#define _generic_swash_utf8(classnum, p) \
cd500f2f 1615 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
922e8cb4 1616
cc8ab7c0 1617/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1618 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1619 * class is TRUE for. Hence it can skip the tests for this range.
1620 * 'above_latin1' should include its arguments */
3ded5eb0 1621#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1622 (UTF8_IS_INVARIANT(*(p)) \
1623 ? _generic_isCC(*(p), classnum) \
1624 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1625 ? above_latin1 \
cc8ab7c0
KW
1626 : 0)
1627
84238efa
KW
1628/* NOTE that some of these macros have very similar ones in regcharclass.h.
1629 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1630 * there, differing in name only by an underscore from the one here
1631 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1632 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1633 * points; the regcharclass.h ones are implemented as a series of
1634 * "if-else-if-else ..." */
1635
b7d90381
KW
1636#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1637#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1638#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
509fb054
KW
1639 utf8, the non-utf8 macro works
1640 */
b7d90381 1641#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2
KW
1642
1643#ifdef EBCDIC
1644 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1645 * more efficient macro instead of the more general one */
88794300 1646# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
e8fa43e2
KW
1647#else
1648# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1649#endif
1650
cd500f2f
KW
1651#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1652 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1653#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1654#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1655 _is_utf8_perl_idcont, p)
e5dcd934 1656
c11ff943
KW
1657/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1658 * IDFIRST is an alnum. See
c8362b00 1659 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1660 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1661 * This used to be not the XID version, but we decided to go with the more
1662 * modern Unicode definition */
b7d90381 1663#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1664 _is_utf8_perl_idstart, p)
e5dcd934 1665
b7d90381
KW
1666#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1667#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1668#define isPSXSPC_utf8(p) isSPACE_utf8(p)
1669#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1670#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1671#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1672#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1673#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1674#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
cd500f2f 1675 is_XDIGIT_high(p))
a0ed51b3 1676
7d2b61dd 1677#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1678#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1679#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1680#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1681
4650c663
KW
1682/* For internal core Perl use only: the base macros for defining macros like
1683 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1684 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1685 * 'macro' parameter */
94bb8c36
KW
1686#define _generic_LC_utf8(macro, p, utf8) \
1687 (UTF8_IS_INVARIANT(*(p)) \
1688 ? macro(*(p)) \
1689 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
a62b247b 1690 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
9201a572
KW
1691 : utf8)
1692
cd500f2f
KW
1693#define _generic_LC_swash_utf8(macro, classnum, p) \
1694 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1695#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1696 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1697
b7d90381 1698#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
cd500f2f 1699 _CC_ALPHANUMERIC, p)
b7d90381
KW
1700#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1701#define isASCII_LC_utf8(p) isASCII_LC(*p)
1702#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1703 is_HORIZWS_high, p)
1704#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1705#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1706#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1707#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1708 _is_utf8_perl_idcont, p)
1709#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1710 _is_utf8_perl_idstart, p)
1711#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1712#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1713#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1714#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1715#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1716 is_XPERLSPACE_high, p)
1717#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
cd500f2f
KW
1718#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1719 _CC_WORDCHAR, p)
b7d90381
KW
1720#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1721 is_XDIGIT_high, p)
aaa51d5e 1722
fbc19f27
KW
1723/* Macros for backwards compatibility and for completeness when the ASCII and
1724 * Latin1 values are identical */
b7d90381
KW
1725#define isALPHAU(c) isALPHA_L1(c)
1726#define isDIGIT_L1(c) isDIGIT_A(c)
1727#define isOCTAL(c) isOCTAL_A(c)
1728#define isOCTAL_L1(c) isOCTAL_A(c)
1729#define isXDIGIT_L1(c) isXDIGIT_A(c)
1730#define isALNUM(c) isWORDCHAR(c)
1731#define isALNUMU(c) isWORDCHAR_L1(c)
1732#define isALNUM_LC(c) isWORDCHAR_LC(c)
1733#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 1734#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
1735#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1736#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1737#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1738#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1739#define isALNUMC(c) isALPHANUMERIC(c)
1740#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1741#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 1742#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 1743#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 1744#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1745
2bd1cbf6
KW
1746/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1747 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1748 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1749 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1750 * the outlier from the block that contains the other controls, just like
1751 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1752 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
1753 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1754 * etc. */
2bd1cbf6 1755#ifndef EBCDIC
75763b3a 1756# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 1757#else
75763b3a
KW
1758# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1759 ((isPRINT_A(c)) \
1760 ? (UNLIKELY((c) == '?') \
1761 ? QUESTION_MARK_CTRL \
1762 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1763 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1764 ? '?' \
1765 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 1766#endif
bbce6d69 1767
dea28490
JJ
1768/* Line numbers are unsigned, 32 bits. */
1769typedef U32 line_t;
e5dcd934 1770#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 1771
91152fc1
DG
1772/* Helpful alias for version prescan */
1773#define is_LAX_VERSION(a,b) \
1774 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1775
1776#define is_STRICT_VERSION(a,b) \
1777 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1778
1779#define BADVERSION(a,b,c) \
1780 if (b) { \
1781 *b = c; \
1782 } \
1783 return a;
8c52afec 1784
cb27eebd
KW
1785/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1786 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1787 * which is advanced. The input is validated only by an assert() in DEBUGGING
1788 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1789 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1790#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1791 ? (c) \
1792 : ((c) + 9))))
902008b8
KW
1793#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1794 ? (*(s)++) \
1795 : (*(s)++ + 9))))
95a59cab 1796
cb27eebd
KW
1797/* Converts a character known to represent an octal digit (0-7) to its numeric
1798 * value. The input is validated only by an assert() in DEBUGGING builds. In
1799 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1800#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1801
305b8651
KW
1802/* Efficiently returns a boolean as to if two native characters are equivalent
1803 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1804 * the ALPHA in the name is to remind you of that. This is asserted() in
1805 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1806 * works (on valid input) for both non- and UTF-8-encoded bytes.
1807 *
1808 * When one of the inputs is a compile-time constant and gets folded by the
1809 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1810 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1811 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
1812 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1813 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1814 * just a single 0, in the bit position where the upper- and lowercase differ.
1815 * */
305b8651
KW
1816#define isALPHA_FOLD_EQ(c1, c2) \
1817 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1818 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1819#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1820
8e84507e 1821/*
ccfc67b7
JH
1822=head1 Memory Management
1823
a02a5408 1824=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
1825The XSUB-writer's interface to the C C<malloc> function.
1826
0d7b2759
MH
1827Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1828
c5008215
JC
1829In 5.9.3, Newx() and friends replace the older New() API, and drops
1830the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 1831themselves. This aid has been superseded by a new build option,
d10b4965 1832PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
1833there for use in XS modules supporting older perls.
1834
a02a5408 1835=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 1836The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 1837cast. See also C<L</Newx>>.
954c1994 1838
0d7b2759
MH
1839Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1840
a02a5408 1841=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 1842The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 1843memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 1844
0d7b2759
MH
1845Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1846
954c1994
GS
1847=for apidoc Am|void|Renew|void* ptr|int nitems|type
1848The XSUB-writer's interface to the C C<realloc> function.
1849
0d7b2759
MH
1850Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1851
954c1994
GS
1852=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1853The XSUB-writer's interface to the C C<realloc> function, with
1854cast.
1855
0d7b2759
MH
1856Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1857
49b8b560 1858=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
1859The XSUB-writer's interface to the C C<free> function.
1860
0d7b2759
MH
1861This should B<ONLY> be used on memory obtained using L<"Newx"> and friends.
1862
954c1994
GS
1863=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1864The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 1865source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1866C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 1867
e90e2364 1868=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 1869Like C<Move> but returns C<dest>. Useful
72d33970 1870for encouraging compilers to tail-call
e90e2364
NC
1871optimise.
1872
954c1994
GS
1873=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1874The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 1875source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1876C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 1877
e90e2364
NC
1878=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1879
796b6530 1880Like C<Copy> but returns C<dest>. Useful
72d33970 1881for encouraging compilers to tail-call
e90e2364
NC
1882optimise.
1883
954c1994
GS
1884=for apidoc Am|void|Zero|void* dest|int nitems|type
1885
1886The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1887destination, C<nitems> is the number of items, and C<type> is the type.
1888
e90e2364
NC
1889=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1890
72d33970
FC
1891Like C<Zero> but returns dest. Useful
1892for encouraging compilers to tail-call
e90e2364
NC
1893optimise.
1894
da5d8dbb 1895=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 1896This is an architecture-independent macro to copy one structure to another.
954c1994 1897
7e337ee0
JH
1898=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1899
1900Fill up memory with a byte pattern (a byte repeated over and over
1901again) that hopefully catches attempts to access uninitialized memory.
1902
1903=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1904
1905PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1906
1c12ffb4 1907=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
1908
1909PoisonWith(0xEF) for catching access to freed memory.
1910
9965345d
JH
1911=for apidoc Am|void|Poison|void* dest|int nitems|type
1912
7e337ee0 1913PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
1914
1915=cut */
954c1994 1916
561b68a9
SH
1917/* Maintained for backwards-compatibility only. Use newSV() instead. */
1918#ifndef PERL_CORE
ff06c60c 1919#define NEWSV(x,len) newSV(len)
561b68a9 1920#endif
ff06c60c 1921
19a94d75
JH
1922#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1923
e6bdf523 1924
27d5b266 1925#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
1926
1927/* This expression will be constant-folded at compile time. It checks
1928 * whether or not the type of the count n is so small (e.g. U8 or U16, or
1929 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
1930 * As well as avoiding the need for a run-time check in some cases, it's
1931 * designed to avoid compiler warnings like:
1932 * comparison is always false due to limited range of data type
73e8ff00
DM
1933 * It's mathematically equivalent to
1934 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
1935 */
1936
1937# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
73e8ff00
DM
1938 ( sizeof(MEM_SIZE) < sizeof(n) \
1939 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
e6bdf523 1940
88f9f128 1941/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
1942 * compiler warnings. We *want* to write the expression as
1943 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
1944 * (for some compile-time constant C), but even when the LHS
1945 * constant-folds to false at compile-time, g++ insists on emitting
1946 * warnings about the RHS (e.g. "comparison is always false"), so instead
1947 * we write it as
e6bdf523 1948 *
d98e5cde 1949 * (cond ? n : X) > C
88f9f128 1950 *
d98e5cde
DM
1951 * where X is a constant with X > C always false. Choosing a value for X
1952 * is tricky. If 0, some compilers will complain about 0 > C always being
1953 * false; if 1, Coverity complains when n happens to be the constant value
1954 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
1955 * for X and hope that nothing else whines.
e6bdf523
DM
1956 */
1957
1958# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
1959 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
1960 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
1961
1962# define MEM_WRAP_CHECK(n,t) \
1963 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1964 && (croak_memory_wrap(),0))
1965
1966# define MEM_WRAP_CHECK_1(n,t,a) \
1967 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1968 && (Perl_croak_nocontext("%s",(a)),0))
1969
8b44ba4c 1970#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 1971
d1decf2b 1972#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
1973#else
1974
410319be
NC
1975#define MEM_WRAP_CHECK(n,t)
1976#define MEM_WRAP_CHECK_1(n,t,a)
1977#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
1978#define MEM_WRAP_CHECK_(n,t)
1979
6fff79ce 1980#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 1981
1936d2a7 1982#endif
8b44ba4c 1983
fe4f188c 1984#ifdef PERL_MEM_LOG
46c6c7e2 1985/*
9f653bb5 1986 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
1987 * go through functions, which are handy for debugging breakpoints, but
1988 * which more importantly get the immediate calling environment (file and
e352bcff
JH
1989 * line number, and C function name if available) passed in. This info can
1990 * then be used for logging the calls, for which one gets a sample
73d1d973 1991 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 1992 *
46c6c7e2 1993 * Known problems:
94e892a6 1994 * - not all memory allocs get logged, only those
46c6c7e2 1995 * that go through Newx() and derivatives (while all
94e892a6 1996 * Safefrees do get logged)
46c6c7e2
JH
1997 * - __FILE__ and __LINE__ do not work everywhere
1998 * - __func__ or __FUNCTION__ even less so
1999 * - I think more goes on after the perlio frees but
2000 * the thing is that STDERR gets closed (as do all
2001 * the file descriptors)
2002 * - no deeper calling stack than the caller of the Newx()
2003 * or the kind, but do I look like a C reflection/introspection
2004 * utility to you?
2005 * - the function prototypes for the logging functions
2006 * probably should maybe be somewhere else than handy.h
2007 * - one could consider inlining (macrofying) the logging
2008 * for speed, but I am too lazy
2009 * - one could imagine recording the allocations in a hash,
2010 * (keyed by the allocation address?), and maintain that
2011 * through reallocs and frees, but how to do that without
2012 * any News() happening...?
73d1d973 2013 * - lots of -Ddefines to get useful/controllable output
b953482e 2014 * - lots of ENV reads
46c6c7e2
JH
2015 */
2016
fc6bde6f 2017PERL_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 2018
fc6bde6f 2019PERL_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 2020
fc6bde6f 2021PERL_CALLCONV Malloc_t Perl_mem_log_free(Malloc_t oldalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 2022
0b0ab801 2023# ifdef PERL_CORE
73d1d973 2024# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2025enum mem_log_type {
2026 MLT_ALLOC,
2027 MLT_REALLOC,
d7a2c63c
MHM
2028 MLT_FREE,
2029 MLT_NEW_SV,
2030 MLT_DEL_SV
0b0ab801
MHM
2031};
2032# endif
12754f92 2033# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2034void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2035void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2036# endif
0b0ab801
MHM
2037# endif
2038
fe4f188c
JH
2039#endif
2040
2041#ifdef PERL_MEM_LOG
d1401ee9
MHM
2042#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2043#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2044#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2045#endif
2046
2047#ifndef MEM_LOG_ALLOC
2048#define MEM_LOG_ALLOC(n,t,a) (a)
2049#endif
2050#ifndef MEM_LOG_REALLOC
2051#define MEM_LOG_REALLOC(n,t,v,a) (a)
2052#endif
2053#ifndef MEM_LOG_FREE
2054#define MEM_LOG_FREE(a) (a)
2055#endif
2056
d1401ee9
MHM
2057#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2058#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2059#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2060
2061#ifndef PERL_CORE
a02a5408
JC
2062/* pre 5.9.x compatibility */
2063#define New(x,v,n,t) Newx(v,n,t)
2064#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2065#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2066#endif
a02a5408 2067
ff68c719 2068#define Renew(v,n,t) \
d1401ee9 2069 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2070#define Renewc(v,n,t,c) \
d1401ee9 2071 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2072
2073#ifdef PERL_POISON
2074#define Safefree(d) \
06c0cc96 2075 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2076#else
fe4f188c 2077#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2078#endif
55497cff 2079
8b44ba4c
NC
2080#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2081#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2082#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2083
8b44ba4c
NC
2084#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2085#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2086#ifdef HAS_MEMSET
8b44ba4c 2087#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2088#else
8b44ba4c
NC
2089/* Using bzero(), which returns void. */
2090#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2091#endif
2092
7e337ee0
JH
2093#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2094#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2095#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2096#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2097
caa674f3
DD
2098#ifdef PERL_POISON
2099# define PERL_POISON_EXPR(x) x
2100#else
2101# define PERL_POISON_EXPR(x)
2102#endif
2103
2304df62 2104#ifdef USE_STRUCT_COPY
ff68c719 2105#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2106#else
2107#define StructCopy(s,d,t) Copy(s,d,1,t)
2108#endif
2cc61e15 2109
c3caa5c3
JH
2110/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2111 * want your zero-based indices to be less than but not equal to).
2112 *
2113 * C_ARRAY_END is one past the last: half-open/half-closed range,
2114 * not last-inclusive range. */
622913ab 2115#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2116#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2117
2cc61e15
DD
2118#ifdef NEED_VA_COPY
2119# ifdef va_copy
2120# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2121# else
a1866d1b
JH
2122# if defined(__va_copy)
2123# define Perl_va_copy(s, d) __va_copy(d, s)
2124# else
2125# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2126# endif
2cc61e15
DD
2127# endif
2128#endif
2129
472d47bc
SB
2130/* convenience debug macros */
2131#ifdef USE_ITHREADS
2132#define pTHX_FORMAT "Perl interpreter: 0x%p"
2133#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2134#define pTHX_VALUE_ (void *)my_perl,
2135#define pTHX_VALUE (void *)my_perl
2136#define pTHX__VALUE_ ,(void *)my_perl,
2137#define pTHX__VALUE ,(void *)my_perl
472d47bc 2138#else
3609ea0d 2139#define pTHX_FORMAT
472d47bc 2140#define pTHX__FORMAT
3609ea0d 2141#define pTHX_VALUE_
472d47bc 2142#define pTHX_VALUE
3609ea0d 2143#define pTHX__VALUE_
472d47bc
SB
2144#define pTHX__VALUE
2145#endif /* USE_ITHREADS */
3609ea0d 2146
2acdbac1
NC
2147/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2148 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2149 CPAN::Unpack show any users outside the core. */
2150#ifdef PERL_CORE
b7d90381
KW
2151# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2152 "Use of " s " is deprecated")
2acdbac1
NC
2153#endif
2154
dfff4baf
BF
2155/* Internal macros to deal with gids and uids */
2156#ifdef PERL_CORE
2157
2158# if Uid_t_size > IVSIZE
2159# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2160# define SvUID(sv) SvNV(sv)
2161# else
2162# if Uid_t_sign <= 0
2163# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2164# define SvUID(sv) SvIV(sv)
2165# else
2166# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2167# define SvUID(sv) SvUV(sv)
2168# endif
2169# endif /* Uid_t_size */
2170
2171# if Gid_t_size > IVSIZE
2172# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2173# define SvGID(sv) SvNV(sv)
2174# else
2175# if Gid_t_sign <= 0
2176# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2177# define SvGID(sv) SvIV(sv)
2178# else
2179# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2180# define SvGID(sv) SvUV(sv)
2181# endif
2182# endif /* Gid_t_size */
2183
2184#endif
2185
9d745869
KW
2186#endif /* HANDY_H */
2187
e9a8c099 2188/*
14d04a33 2189 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2190 */