This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Consistently use NOT_REACHED; /* NOTREACHED */
[perl5.git] / handy.h
CommitLineData
a0d0e21e 1/* handy.h
a687059c 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
da5d8dbb 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
a687059c 5 *
6e21c824
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8d063cd8 8 *
8d063cd8
LW
9 */
10
4650c663
KW
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
9d745869
KW
14#ifndef HANDY_H /* Guard against nested #inclusion */
15#define HANDY_H
16
85e6fe83 17#if !defined(__STDC__)
378cc40b
LW
18#ifdef NULL
19#undef NULL
20#endif
a687059c 21# define NULL 0
85e6fe83
LW
22#endif
23
24792b8d
NC
24#ifndef PERL_CORE
25# define Null(type) ((type)NULL)
954c1994
GS
26
27/*
ccfc67b7 28=head1 Handy Values
954c1994 29
3609ea0d 30=for apidoc AmU||Nullch
72d33970
FC
31Null character pointer. (No longer available when C<PERL_CORE> is
32defined.)
2307c6d0 33
954c1994 34=for apidoc AmU||Nullsv
72d33970 35Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
36
37=cut
38*/
39
24792b8d
NC
40# define Nullch Null(char*)
41# define Nullfp Null(PerlIO*)
42# define Nullsv Null(SV*)
43#endif
8d063cd8 44
641d3f0b
PP
45#ifdef TRUE
46#undef TRUE
47#endif
48#ifdef FALSE
49#undef FALSE
50#endif
51#define TRUE (1)
52#define FALSE (0)
53
cf3f0ffb
DM
54/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
55 * (compiler permitting) that casting away const-ness will give a warning;
56 * e.g.:
57 *
58 * const SV *sv = ...;
59 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
60 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
61 */
62
b1bc3f34 63#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
f065df04 64# define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
b1bc3f34
NC
65#else
66# define MUTABLE_PTR(p) ((void *) (p))
67#endif
68
a062e10d 69#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 70#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 71#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 72#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 73#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 74#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 75
f789f6a4 76#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
77# include <stdbool.h>
78# ifndef HAS_BOOL
79# define HAS_BOOL 1
80# endif
81#endif
82
8e84507e 83/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
84 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
85 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 86 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
87 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
88 g++ can be identified by __GNUG__.
89 Andy Dougherty February 2000
5d94fbed 90*/
3609ea0d 91#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 92# ifndef PERL_BOOL_AS_CHAR
5d94fbed 93# ifndef HAS_BOOL
c1d22f6b 94# define HAS_BOOL 1
5d94fbed 95# endif
f789f6a4 96# endif
5d94fbed
AD
97#endif
98
99#ifndef HAS_BOOL
f789f6a4
FC
100# ifdef bool
101# undef bool
102# endif
70d5cb32 103# define bool char
c1d22f6b 104# define HAS_BOOL 1
a687059c 105#endif
0d3e774c 106
236415e2
KW
107/* cast-to-bool. A simple (bool) cast may not do the right thing: if bool is
108 * defined as char for example, then the cast from int is
109 * implementation-defined (bool)!!(cbool) in a ternary triggers a bug in xlc on
110 * AIX */
18f5643b 111#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 112
46c6c7e2 113/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
114 * XXX Should really be a Configure probe, with HAS__FUNCTION__
115 * and FUNCTION__ as results.
116 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
46c6c7e2
JH
117#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
118# define FUNCTION__ __func__
119#else
9ea40801 120# if (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
be73ebb0 121 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
46c6c7e2
JH
122# define FUNCTION__ ""
123# else
124# define FUNCTION__ __FUNCTION__ /* Common extension. */
125# endif
126#endif
127
27d4fb96
PP
128/* XXX A note on the perl source internal type system. The
129 original intent was that I32 be *exactly* 32 bits.
130
131 Currently, we only guarantee that I32 is *at least* 32 bits.
132 Specifically, if int is 64 bits, then so is I32. (This is the case
133 for the Cray.) This has the advantage of meshing nicely with
134 standard library calls (where we pass an I32 and the library is
135 expecting an int), but the disadvantage that an I32 is not 32 bits.
136 Andy Dougherty August 1996
24fef2a7 137
dc45a647
MB
138 There is no guarantee that there is *any* integral type with
139 exactly 32 bits. It is perfectly legal for a system to have
140 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 141
dc45a647
MB
142 Similarly, there is no guarantee that I16 and U16 have exactly 16
143 bits.
693762b4 144
8e84507e
NIS
145 For dealing with issues that may arise from various 32/64-bit
146 systems, we will ask Configure to check out
8175356b 147
3609ea0d
JH
148 SHORTSIZE == sizeof(short)
149 INTSIZE == sizeof(int)
150 LONGSIZE == sizeof(long)
dc45a647 151 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 152 PTRSIZE == sizeof(void *)
dc45a647
MB
153 DOUBLESIZE == sizeof(double)
154 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 155
27d4fb96
PP
156*/
157
69512466
JH
158#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
159# include <inttypes.h>
dd0eed91
JH
160# ifdef INT32_MIN_BROKEN
161# undef INT32_MIN
162# define INT32_MIN (-2147483647-1)
163# endif
164# ifdef INT64_MIN_BROKEN
165# undef INT64_MIN
166# define INT64_MIN (-9223372036854775807LL-1)
167# endif
69512466
JH
168#endif
169
8175356b
JH
170typedef I8TYPE I8;
171typedef U8TYPE U8;
172typedef I16TYPE I16;
173typedef U16TYPE U16;
174typedef I32TYPE I32;
175typedef U32TYPE U32;
6b8eaf93
JH
176#ifdef PERL_CORE
177# ifdef HAS_QUAD
8175356b
JH
178typedef I64TYPE I64;
179typedef U64TYPE U64;
6b8eaf93
JH
180# endif
181#endif /* PERL_CORE */
8175356b 182
9c39b21c
JH
183/* INT64_C/UINT64_C are C99 from <stdint.h> (so they will not be
184 * available in strict C89 mode), but they are nice, so let's define
185 * them if necessary. */
0d3bd0a9 186#if defined(HAS_QUAD)
9c39b21c
JH
187# undef PeRl_INT64_C
188# undef PeRl_UINT64_C
189/* Prefer the native integer types (int and long) over long long
190 * (which is not C89) and Win32-specific __int64. */
191# if QUADKIND == QUAD_IS_INT && INTSIZE == 8
192# define PeRl_INT64_C(c) (c)
193# define PeRl_UINT64_C(c) CAT2(c,U)
194# endif
195# if QUADKIND == QUAD_IS_LONG && LONGSIZE == 8
196# define PeRl_INT64_C(c) CAT2(c,L)
197# define PeRl_UINT64_C(c) CAT2(c,UL)
198# endif
199# if QUADKIND == QUAD_IS_LONG_LONG && defined(HAS_LONG_LONG)
200# define PeRl_INT64_C(c) CAT2(c,LL)
201# define PeRl_UINT64_C(c) CAT2(c,ULL)
202# endif
203# if QUADKIND == QUAD_IS___INT64
204# define PeRl_INT64_C(c) CAT2(c,I64)
205# define PeRl_UINT64_C(c) CAT2(c,UI64)
206# endif
207# ifndef PeRl_INT64_C
208# define PeRl_INT64_C(c) ((I64TYPE)(c)) /* last resort */
209# define PeRl_UINT64_C(c) ((U64TYPE)(c))
210# endif
211/* In OS X the INT64_C/UINT64_C are defined with LL/ULL, which will
212 * not fly with C89-pedantic gcc, so let's undefine them first so that
213 * we can redefine them with our native integer preferring versions. */
8eafc6eb 214# if defined(PERL_DARWIN) && defined(PERL_GCC_PEDANTIC)
9c39b21c
JH
215# undef INT64_C
216# undef UINT64_C
217# endif
218# ifndef INT64_C
219# define INT64_C(c) PeRl_INT64_C(c)
220# endif
221# ifndef UINT64_C
222# define UINT64_C(c) PeRl_UINT64_C(c)
223# endif
e8c95190 224#endif
dc750f2e 225
d8668976 226#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 227
5ff3f7a4
GS
228/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
229 Please search CHAR_MAX in perl.h for further details. */
230#define U8_MAX UINT8_MAX
231#define U8_MIN UINT8_MIN
232
5ff3f7a4
GS
233#define I16_MAX INT16_MAX
234#define I16_MIN INT16_MIN
235#define U16_MAX UINT16_MAX
236#define U16_MIN UINT16_MIN
237
5ff3f7a4
GS
238#define I32_MAX INT32_MAX
239#define I32_MIN INT32_MIN
0e983133
GS
240#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
241# define U32_MAX UINT32_MAX
242#else
243# define U32_MAX 4294967295U
244#endif
5ff3f7a4
GS
245#define U32_MIN UINT32_MIN
246
247#else
248
5c9fa16e
KA
249/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
250 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
251#define U8_MAX PERL_UCHAR_MAX
252#define U8_MIN PERL_UCHAR_MIN
79072805 253
27d4fb96
PP
254#define I16_MAX PERL_SHORT_MAX
255#define I16_MIN PERL_SHORT_MIN
256#define U16_MAX PERL_USHORT_MAX
257#define U16_MIN PERL_USHORT_MIN
79072805 258
c4f23d77 259#if LONGSIZE > 4
27d4fb96
PP
260# define I32_MAX PERL_INT_MAX
261# define I32_MIN PERL_INT_MIN
262# define U32_MAX PERL_UINT_MAX
263# define U32_MIN PERL_UINT_MIN
79072805 264#else
27d4fb96
PP
265# define I32_MAX PERL_LONG_MAX
266# define I32_MIN PERL_LONG_MIN
267# define U32_MAX PERL_ULONG_MAX
268# define U32_MIN PERL_ULONG_MIN
79072805
LW
269#endif
270
5ff3f7a4
GS
271#endif
272
58a9a5d5 273/* log(2) is pretty close to 0.30103, just in case anyone is grepping for it */
fc36a67e
PP
274#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
275#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
276#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
277
88794300 278/* Unused by core; should be deprecated */
ff68c719 279#define Ctl(ch) ((ch) & 037)
8d063cd8 280
4650c663
KW
281/* This is a helper macro to avoid preprocessor issues, replaced by nothing
282 * unless under DEBUGGING, where it expands to an assert of its argument,
283 * followed by a comma (hence the comma operator). If we just used a straight
284 * assert(), we would get a comma with nothing before it when not DEBUGGING */
0f092d08
KW
285#ifdef DEBUGGING
286# define __ASSERT_(statement) assert(statement),
287#else
288# define __ASSERT_(statement)
289#endif
290
3fe05580
MHM
291/*
292=head1 SV-Body Allocation
293
294=for apidoc Ama|SV*|newSVpvs|const char* s
30a15352
KW
295Like C<newSVpvn>, but takes a literal C<NUL>-terminated string instead of a
296string/length pair.
3fe05580 297
84bafc02 298=for apidoc Ama|SV*|newSVpvs_flags|const char* s|U32 flags
30a15352
KW
299Like C<newSVpvn_flags>, but takes a literal C<NUL>-terminated string instead of
300a string/length pair.
84bafc02 301
3fe05580 302=for apidoc Ama|SV*|newSVpvs_share|const char* s
30a15352
KW
303Like C<newSVpvn_share>, but takes a literal C<NUL>-terminated string instead of
304a string/length pair and omits the hash parameter.
3fe05580 305
9dcc53ea 306=for apidoc Am|void|sv_catpvs_flags|SV* sv|const char* s|I32 flags
30a15352
KW
307Like C<sv_catpvn_flags>, but takes a literal C<NUL>-terminated string instead
308of a string/length pair.
9dcc53ea
Z
309
310=for apidoc Am|void|sv_catpvs_nomg|SV* sv|const char* s
311Like C<sv_catpvn_nomg>, but takes a literal string instead of a
312string/length pair.
313
7b4e7469 314=for apidoc Am|void|sv_catpvs|SV* sv|const char* s
3fe05580
MHM
315Like C<sv_catpvn>, but takes a literal string instead of a string/length pair.
316
9dcc53ea
Z
317=for apidoc Am|void|sv_catpvs_mg|SV* sv|const char* s
318Like C<sv_catpvn_mg>, but takes a literal string instead of a
319string/length pair.
320
7b4e7469 321=for apidoc Am|void|sv_setpvs|SV* sv|const char* s
3fe05580
MHM
322Like C<sv_setpvn>, but takes a literal string instead of a string/length pair.
323
9dcc53ea
Z
324=for apidoc Am|void|sv_setpvs_mg|SV* sv|const char* s
325Like C<sv_setpvn_mg>, but takes a literal string instead of a
326string/length pair.
327
328=for apidoc Am|SV *|sv_setref_pvs|const char* s
329Like C<sv_setref_pvn>, but takes a literal string instead of a
330string/length pair.
331
3fe05580
MHM
332=head1 Memory Management
333
334=for apidoc Ama|char*|savepvs|const char* s
30a15352
KW
335Like C<savepvn>, but takes a literal C<NUL>-terminated string instead of a
336string/length pair.
3fe05580 337
9dcc53ea
Z
338=for apidoc Ama|char*|savesharedpvs|const char* s
339A version of C<savepvs()> which allocates the duplicate string in memory
340which is shared between threads.
341
3fe05580
MHM
342=head1 GV Functions
343
344=for apidoc Am|HV*|gv_stashpvs|const char* name|I32 create
345Like C<gv_stashpvn>, but takes a literal string instead of a string/length pair.
346
347=head1 Hash Manipulation Functions
348
349=for apidoc Am|SV**|hv_fetchs|HV* tb|const char* key|I32 lval
350Like C<hv_fetch>, but takes a literal string instead of a string/length pair.
351
352=for apidoc Am|SV**|hv_stores|HV* tb|const char* key|NULLOK SV* val
353Like C<hv_store>, but takes a literal string instead of a string/length pair
354and omits the hash parameter.
355
510966aa
Z
356=head1 Lexer interface
357
358=for apidoc Amx|void|lex_stuff_pvs|const char *pv|U32 flags
359
360Like L</lex_stuff_pvn>, but takes a literal string instead of a
361string/length pair.
362
3fe05580
MHM
363=cut
364*/
365
926bb54c
KW
366/* concatenating with "" ensures that only literal strings are accepted as
367 * argument */
26bfd924 368#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
2efa8cc7 369
926bb54c
KW
370/* note that STR_WITH_LEN() can't be used as argument to macros or functions
371 * that under some configurations might be macros, which means that it requires
372 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
ba3a79e7
GA
373 */
374
375/* STR_WITH_LEN() shortcuts */
376#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
377#define newSVpvs_flags(str,flags) \
378 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 379#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
380#define sv_catpvs_flags(sv, str, flags) \
381 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
382#define sv_catpvs_nomg(sv, str) \
383 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
384#define sv_catpvs(sv, str) \
385 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
386#define sv_catpvs_mg(sv, str) \
387 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 388#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
389#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
390#define sv_setref_pvs(rv, classname, str) \
391 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 392#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
393#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
394#define gv_stashpvs(str, create) \
395 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
396#define gv_fetchpvs(namebeg, add, sv_type) \
397 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
398#define gv_fetchpvn(namebeg, len, add, sv_type) \
399 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
400#define sv_catxmlpvs(dsv, str, utf8) \
401 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
a038e571
NC
402#define hv_fetchs(hv,key,lval) \
403 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
404 (lval) ? (HV_FETCH_JUST_SV | HV_FETCH_LVALUE) \
405 : HV_FETCH_JUST_SV, NULL, 0))
406
407#define hv_stores(hv,key,val) \
408 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
409 (HV_FETCH_ISSTORE|HV_FETCH_JUST_SV), (val), 0))
ba3a79e7 410
510966aa
Z
411#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
412
b96d8cd9
NC
413#define get_cvs(str, flags) \
414 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 415
954c1994 416/*
ccfc67b7
JH
417=head1 Miscellaneous Functions
418
954c1994
GS
419=for apidoc Am|bool|strNE|char* s1|char* s2
420Test two strings to see if they are different. Returns true or
421false.
422
423=for apidoc Am|bool|strEQ|char* s1|char* s2
424Test two strings to see if they are equal. Returns true or false.
425
426=for apidoc Am|bool|strLT|char* s1|char* s2
427Test two strings to see if the first, C<s1>, is less than the second,
428C<s2>. Returns true or false.
429
430=for apidoc Am|bool|strLE|char* s1|char* s2
431Test two strings to see if the first, C<s1>, is less than or equal to the
432second, C<s2>. Returns true or false.
433
434=for apidoc Am|bool|strGT|char* s1|char* s2
435Test two strings to see if the first, C<s1>, is greater than the second,
436C<s2>. Returns true or false.
437
438=for apidoc Am|bool|strGE|char* s1|char* s2
439Test two strings to see if the first, C<s1>, is greater than or equal to
440the second, C<s2>. Returns true or false.
441
442=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
443Test two strings to see if they are different. The C<len> parameter
72d33970 444indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
445wrapper for C<strncmp>).
446
447=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
448Test two strings to see if they are equal. The C<len> parameter indicates
72d33970 449the number of bytes to compare. Returns true or false. (A wrapper for
954c1994
GS
450C<strncmp>).
451
452=cut
453*/
454
8d063cd8
LW
455#define strNE(s1,s2) (strcmp(s1,s2))
456#define strEQ(s1,s2) (!strcmp(s1,s2))
457#define strLT(s1,s2) (strcmp(s1,s2) < 0)
458#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
459#define strGT(s1,s2) (strcmp(s1,s2) > 0)
460#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
461#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
462#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
378cc40b 463
36477c24
PP
464#ifdef HAS_MEMCMP
465# define memNE(s1,s2,l) (memcmp(s1,s2,l))
466# define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
467#else
468# define memNE(s1,s2,l) (bcmp(s1,s2,l))
469# define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
470#endif
471
568a785a 472#define memEQs(s1, l, s2) \
8e6497b6 473 (sizeof(s2)-1 == l && memEQ(s1, ("" s2 ""), (sizeof(s2)-1)))
568a785a
NC
474#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
475
bbce6d69
PP
476/*
477 * Character classes.
478 *
479 * Unfortunately, the introduction of locales means that we
480 * can't trust isupper(), etc. to tell the truth. And when
481 * it comes to /\w+/ with tainting enabled, we *must* be able
482 * to trust our character classes.
483 *
484 * Therefore, the default tests in the text of Perl will be
485 * independent of locale. Any code that wants to depend on
486 * the current locale will use the tests that begin with "lc".
487 */
488
2304df62
AD
489#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
490# ifndef CTYPE256
491# define CTYPE256
492# endif
493#endif
494
954c1994 495/*
ccfc67b7 496
dcccc8ff 497=head1 Character classification
243effed
KW
498This section is about functions (really macros) that classify characters
499into types, such as punctuation versus alphabetic, etc. Most of these are
500analogous to regular expression character classes. (See
501L<perlrecharclass/POSIX Character Classes>.) There are several variants for
502each class. (Not all macros have all variants; each item below lists the
503ones valid for it.) None are affected by C<use bytes>, and only the ones
504with C<LC> in the name are affected by the current locale.
505
506The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
507C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
508represented by that octet is (or on non-ASCII platforms, corresponds to) an
509ASCII character in the named class based on platform, Unicode, and Perl rules.
510If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 511
6aff1f14
KW
512Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
513with no suffix C<"_A">.
4b9734bf
KW
514
515Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
516onto the platform. That is, the code points that are ASCII are unaffected,
517since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
518as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
519true when called with the code point 0xDF, which is a word character in both
4650c663 520ASCII and EBCDIC (though it represents different characters in each).
243effed
KW
521
522Variant C<isFOO_uni> is like the C<isFOO_L1> variant, but accepts any UV code
523point as input. If the code point is larger than 255, Unicode rules are used
524to determine if it is in the character class. For example,
6aff1f14
KW
525C<isWORDCHAR_uni(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
526WITH MACRON in Unicode, and is a word character.
243effed
KW
527
528Variant C<isFOO_utf8> is like C<isFOO_uni>, but the input is a pointer to a
529(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
6aff1f14
KW
530classification of just the first (possibly multi-byte) character in the string
531is tested.
243effed 532
31f05a37
KW
533Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
534result is based on the current locale, which is what C<LC> in the name stands
535for. If Perl can determine that the current locale is a UTF-8 locale, it uses
536the published Unicode rules; otherwise, it uses the C library function that
537gives the named classification. For example, C<isDIGIT_LC()> when not in a
538UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
539returned if the input won't fit into an octet. On some platforms where the C
540library function is known to be defective, Perl changes its result to follow
541the POSIX standard's rules.
243effed
KW
542
543Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
544returns the same as C<isFOO_LC> for input code points less than 256, and
545returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
546
547Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer to a
548(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
6aff1f14
KW
549classification of just the first (possibly multi-byte) character in the string
550is tested.
ccfc67b7 551
954c1994 552=for apidoc Am|bool|isALPHA|char ch
2787a470 553Returns a boolean indicating whether the specified character is an
6aff1f14 554alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
555See the L<top of this section|/Character classification> for an explanation of
556variants
e5ad6aba 557C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uni>, C<isALPHA_utf8>, C<isALPHA_LC>,
243effed 558C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
8a58bdcf 559
15861f94
KW
560=for apidoc Am|bool|isALPHANUMERIC|char ch
561Returns a boolean indicating whether the specified character is a either an
6aff1f14 562alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
563See the L<top of this section|/Character classification> for an explanation of
564variants
15861f94 565C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uni>,
e5ad6aba 566C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
15861f94
KW
567C<isALPHANUMERIC_LC_utf8>.
568
8a58bdcf
KW
569=for apidoc Am|bool|isASCII|char ch
570Returns a boolean indicating whether the specified character is one of the 128
243effed 571characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 572On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
573character corresponds to an ASCII character. Variants C<isASCII_A()> and
574C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
575See the L<top of this section|/Character classification> for an explanation of
576variants
243effed
KW
577C<isASCII_uni>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
578C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
579library routine C<isascii()>. In these cases, the variants whose names contain
580C<LC> are the same as the corresponding ones without.
581
d98532ea
KW
582Also note, that because all ASCII characters are UTF-8 invariant (meaning they
583have the exact same representation (always a single byte) whether encoded in
584UTF-8 or not), C<isASCII> will give the correct results when called with any
585byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
586work properly on any string encoded or not in UTF-8.
587
243effed
KW
588=for apidoc Am|bool|isBLANK|char ch
589Returns a boolean indicating whether the specified character is a
6aff1f14 590character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
591See the L<top of this section|/Character classification> for an explanation of
592variants
e5ad6aba 593C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uni>, C<isBLANK_utf8>, C<isBLANK_LC>,
243effed
KW
594C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
595platforms do not have the C library routine C<isblank()>. In these cases, the
596variants whose names contain C<LC> are the same as the corresponding ones
597without.
598
599=for apidoc Am|bool|isCNTRL|char ch
600Returns a boolean indicating whether the specified character is a
6aff1f14 601control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
602See the L<top of this section|/Character classification> for an explanation of
603variants
e5ad6aba 604C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uni>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
6aff1f14
KW
605C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
606On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 607
954c1994 608=for apidoc Am|bool|isDIGIT|char ch
2787a470 609Returns a boolean indicating whether the specified character is a
6aff1f14 610digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 611Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
612See the L<top of this section|/Character classification> for an explanation of
613variants
e5ad6aba 614C<isDIGIT_uni>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
243effed
KW
615C<isDIGIT_LC_utf8>.
616
617=for apidoc Am|bool|isGRAPH|char ch
618Returns a boolean indicating whether the specified character is a
6aff1f14 619graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff
KW
620See the L<top of this section|/Character classification> for an explanation of
621variants
e5ad6aba 622C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uni>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
243effed 623C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
954c1994 624
0c82b6df 625=for apidoc Am|bool|isLOWER|char ch
2787a470 626Returns a boolean indicating whether the specified character is a
6aff1f14 627lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
628See the L<top of this section|/Character classification> for an explanation of
629variants
e5ad6aba 630C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uni>, C<isLOWER_utf8>, C<isLOWER_LC>,
243effed 631C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
0c82b6df 632
c99e91e9 633=for apidoc Am|bool|isOCTAL|char ch
2787a470 634Returns a boolean indicating whether the specified character is an
6aff1f14 635octal digit, [0-7].
243effed
KW
636The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
637C<isOCTAL>.
638
639=for apidoc Am|bool|isPUNCT|char ch
640Returns a boolean indicating whether the specified character is a
6aff1f14
KW
641punctuation character, analogous to C<m/[[:punct:]]/>.
642Note that the definition of what is punctuation isn't as
243effed
KW
643straightforward as one might desire. See L<perlrecharclass/POSIX Character
644Classes> for details.
dcccc8ff
KW
645See the L<top of this section|/Character classification> for an explanation of
646variants
e5ad6aba 647C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uni>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
243effed 648C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
c99e91e9 649
0c82b6df 650=for apidoc Am|bool|isSPACE|char ch
2787a470 651Returns a boolean indicating whether the specified character is a
6aff1f14 652whitespace character. This is analogous
398d098a 653to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 654this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
655locale forms of this macro (the ones with C<LC> in their names) matched
656precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
657in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
658(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
659See the L<top of this section|/Character classification> for an explanation of
660variants
e5ad6aba 661C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uni>, C<isSPACE_utf8>, C<isSPACE_LC>,
243effed 662C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
0c82b6df 663
398d098a
KW
664=for apidoc Am|bool|isPSXSPC|char ch
665(short for Posix Space)
779cf272
KW
666Starting in 5.18, this is identical in all its forms to the
667corresponding C<isSPACE()> macros.
398d098a
KW
668The locale forms of this macro are identical to their corresponding
669C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
670non-locale forms differ from their C<isSPACE()> forms only in that the
671C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
672Otherwise they are identical. Thus this macro is analogous to what
673C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff
KW
674See the L<top of this section|/Character classification> for an explanation of
675variants
e5ad6aba 676C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uni>, C<isPSXSPC_utf8>, C<isPSXSPC_LC>,
398d098a
KW
677C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
678
954c1994 679=for apidoc Am|bool|isUPPER|char ch
2787a470 680Returns a boolean indicating whether the specified character is an
6aff1f14 681uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff
KW
682See the L<top of this section|/Character classification> for an explanation of
683variants
e5ad6aba 684C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uni>, C<isUPPER_utf8>, C<isUPPER_LC>,
243effed 685C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
954c1994 686
243effed 687=for apidoc Am|bool|isPRINT|char ch
8eea39dd 688Returns a boolean indicating whether the specified character is a
6aff1f14 689printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
690See the L<top of this section|/Character classification> for an explanation of
691variants
e5ad6aba 692C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uni>, C<isPRINT_utf8>, C<isPRINT_LC>,
243effed
KW
693C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
694
695=for apidoc Am|bool|isWORDCHAR|char ch
696Returns a boolean indicating whether the specified character is a character
697that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
698in a regular expression. A word character is an alphabetic character, a
699decimal digit, a connecting punctuation character (such as an underscore), or
700a "mark" character that attaches to one of those (like some sort of accent).
701C<isALNUM()> is a synonym provided for backward compatibility, even though a
702word character includes more than the standard C language meaning of
703alphanumeric.
dcccc8ff
KW
704See the L<top of this section|/Character classification> for an explanation of
705variants
1a83413c
KW
706C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uni>, and C<isWORDCHAR_utf8>.
707C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
708described there, but additionally include the platform's native underscore.
8a58bdcf
KW
709
710=for apidoc Am|bool|isXDIGIT|char ch
711Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
712digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
713and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
714See the L<top of this section|/Character classification> for an explanation of
715variants
243effed
KW
716C<isXDIGIT_uni>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
717C<isXDIGIT_LC_utf8>.
718
3c3ecf18
KW
719=for apidoc Am|bool|isIDFIRST|char ch
720Returns a boolean indicating whether the specified character can be the first
721character of an identifier. This is very close to, but not quite the same as
722the official Unicode property C<XID_Start>. The difference is that this
723returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
724See the L<top of this section|/Character classification> for an explanation of
725variants
3c3ecf18
KW
726C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uni>, C<isIDFIRST_utf8>,
727C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
728
729=for apidoc Am|bool|isIDCONT|char ch
730Returns a boolean indicating whether the specified character can be the
731second or succeeding character of an identifier. This is very close to, but
732not quite the same as the official Unicode property C<XID_Continue>. The
733difference is that this returns true only if the input character also matches
dcccc8ff
KW
734L</isWORDCHAR>. See the L<top of this section|/Character classification> for
735an
3c3ecf18
KW
736explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uni>,
737C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
738C<isIDCONT_LC_utf8>.
739
243effed 740=head1 Miscellaneous Functions
8eea39dd 741
95a59cab 742=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 743Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
744Behaviour is only well defined when isXDIGIT(*str) is true.
745
e7c1e6c1
KW
746=head1 Character case changing
747
1f607577
KW
748=for apidoc Am|U8|toUPPER|U8 ch
749Converts the specified character to uppercase. If the input is anything but an
750ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 751C<toUPPER_A> is equivalent.
954c1994 752
1f607577
KW
753=for apidoc Am|UV|toUPPER_uni|UV cp|U8* s|STRLEN* lenp
754Converts the Unicode code point C<cp> to its uppercase version, and
755stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
756that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
757bytes since the uppercase version may be longer than the original character.
758
759The first code point of the uppercased version is returned
760(but note, as explained just above, that there may be more.)
761
762=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
763Converts the UTF-8 encoded character at C<p> to its uppercase version, and
764stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
765that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
766bytes since the uppercase version may be longer than the original character.
767
768The first code point of the uppercased version is returned
769(but note, as explained just above, that there may be more.)
770
771The input character at C<p> is assumed to be well-formed.
772
25200305
KW
773=for apidoc Am|U8|toFOLD|U8 ch
774Converts the specified character to foldcase. If the input is anything but an
775ASCII uppercase character, that input character itself is returned. Variant
776C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
777Latin1 range, as the full generality of L</toFOLD_uni> is needed there.)
778
1f607577
KW
779=for apidoc Am|UV|toFOLD_uni|UV cp|U8* s|STRLEN* lenp
780Converts the Unicode code point C<cp> to its foldcase version, and
781stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
782that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
783bytes since the foldcase version may be longer than the original character.
784
785The first code point of the foldcased version is returned
786(but note, as explained just above, that there may be more.)
787
788=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
789Converts the UTF-8 encoded character at C<p> to its foldcase version, and
790stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
791that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
792bytes since the foldcase version may be longer than the original character.
793
794The first code point of the foldcased version is returned
795(but note, as explained just above, that there may be more.)
796
797The input character at C<p> is assumed to be well-formed.
798
799=for apidoc Am|U8|toLOWER|U8 ch
800Converts the specified character to lowercase. If the input is anything but an
801ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 802C<toLOWER_A> is equivalent.
954c1994 803
1f607577
KW
804=for apidoc Am|U8|toLOWER_L1|U8 ch
805Converts the specified Latin1 character to lowercase. The results are undefined if
806the input doesn't fit in a byte.
807
808=for apidoc Am|U8|toLOWER_LC|U8 ch
809Converts the specified character to lowercase using the current locale's rules,
810if possible; otherwise returns the input character itself.
811
812=for apidoc Am|UV|toLOWER_uni|UV cp|U8* s|STRLEN* lenp
813Converts the Unicode code point C<cp> to its lowercase version, and
814stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
815that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
816bytes since the lowercase version may be longer than the original character.
817
818The first code point of the lowercased version is returned
819(but note, as explained just above, that there may be more.)
820
821=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
822Converts the UTF-8 encoded character at C<p> to its lowercase version, and
823stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
824that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
825bytes since the lowercase version may be longer than the original character.
826
827The first code point of the lowercased version is returned
828(but note, as explained just above, that there may be more.)
829
830The input character at C<p> is assumed to be well-formed.
831
25200305
KW
832=for apidoc Am|U8|toTITLE|U8 ch
833Converts the specified character to titlecase. If the input is anything but an
834ASCII lowercase character, that input character itself is returned. Variant
835C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1 range,
836as the full generality of L</toTITLE_uni> is needed there. Titlecase is not a
837concept used in locale handling, so there is no functionality for that.)
838
1f607577
KW
839=for apidoc Am|UV|toTITLE_uni|UV cp|U8* s|STRLEN* lenp
840Converts the Unicode code point C<cp> to its titlecase version, and
841stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
842that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
843bytes since the titlecase version may be longer than the original character.
844
845The first code point of the titlecased version is returned
846(but note, as explained just above, that there may be more.)
847
848=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
849Converts the UTF-8 encoded character at C<p> to its titlecase version, and
850stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
851that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
852bytes since the titlecase version may be longer than the original character.
853
854The first code point of the titlecased version is returned
855(but note, as explained just above, that there may be more.)
856
857The input character at C<p> is assumed to be well-formed.
858
954c1994 859=cut
353c9b6f 860
1f607577 861XXX Still undocumented isVERTWS_uni and _utf8; it's unclear what their names
25200305 862really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 863
8a58bdcf 864Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
865patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
866
954c1994
GS
867*/
868
7c062697
KW
869/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
870 * is known only in the perl core, and this macro can be called from outside
871 * that */
872#ifdef HAS_QUAD
873# define WIDEST_UTYPE U64TYPE
874#else
875# define WIDEST_UTYPE U32
876#endif
877
3912bc88
KW
878/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
879 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
880 * bits of information are lost even on a 64-bit machine, but to get the
881 * compiler to optimize it out if possible. This is because Configure makes
882 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
883 * sizeof() guarantees that this evaluates to a constant true at compile time.
220c71bf
KW
884 */
885#define FITS_IN_8_BITS(c) ((sizeof(c) == 1) || !(((WIDEST_UTYPE)(c)) & ~0xFF))
cf301eb7 886
41f43cc2 887#ifdef EBCDIC
b6340bd0 888# ifndef _ALL_SOURCE
0852beac
KW
889 /* The native libc isascii() et.al. functions return the wrong results
890 * on at least z/OS unless this is defined. */
b6340bd0
KW
891# error _ALL_SOURCE should probably be defined
892# endif
41f43cc2 893#else
0852beac
KW
894 /* There is a simple definition of ASCII for ASCII platforms. But the
895 * EBCDIC one isn't so simple, so is defined using table look-up like the
896 * other macros below */
41f43cc2
KW
897# define isASCII(c) ((WIDEST_UTYPE)(c) < 128)
898#endif
899
c2da0b36
KW
900/* The lower 3 bits in both the ASCII and EBCDIC representations of '0' are 0,
901 * and the 8 possible permutations of those bits exactly comprise the 8 octal
902 * digits */
3d33b3e3 903#define isOCTAL_A(c) cBOOL(FITS_IN_8_BITS(c) && (0xF8 & (c)) == '0')
c2da0b36 904
9fb1bf9d 905#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 906
a500dc72
KW
907/* Character class numbers. For internal core Perl use only. The ones less
908 * than 32 are used in PL_charclass[] and the ones up through the one that
909 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
910 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
911 * their actual definitions are here. If that file has a name not used here,
912 * it won't compile.
1709d539
KW
913 *
914 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
915 * frequency of their use. This gives a slight edge to exiting a loop earlier
916 * (in reginclass() in regexec.c) */
1709d539
KW
917# define _CC_WORDCHAR 0 /* \w and [:word:] */
918# define _CC_DIGIT 1 /* \d and [:digit:] */
919# define _CC_ALPHA 2 /* [:alpha:] */
920# define _CC_LOWER 3 /* [:lower:] */
921# define _CC_UPPER 4 /* [:upper:] */
922# define _CC_PUNCT 5 /* [:punct:] */
923# define _CC_PRINT 6 /* [:print:] */
15861f94 924# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 925# define _CC_GRAPH 8 /* [:graph:] */
b0d691b2 926# define _CC_CASED 9 /* [:lower:] and [:upper:] under /i */
1709d539 927
b0d691b2 928#define _FIRST_NON_SWASH_CC 10
1709d539
KW
929/* The character classes above are implemented with swashes. The second group
930 * (just below) contains the ones implemented without. These are also sorted
931 * in rough order of the frequency of their use, except that \v should be last,
932 * as it isn't a real Posix character class, and some (small) inefficiencies in
933 * regular expression handling would be introduced by putting it in the middle
934 * of those that are. Also, cntrl and ascii come after the others as it may be
935 * useful to group these which have no members that match above Latin1, (or
936 * above ASCII in the latter case) */
937
779cf272 938# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
939# define _CC_BLANK 11 /* [:blank:] */
940# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
941# define _CC_CNTRL 13 /* [:cntrl:] */
942# define _CC_ASCII 14 /* [:ascii:] */
943# define _CC_VERTSPACE 15 /* \v */
1709d539 944
a0947d7b
KW
945# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
946
1709d539 947/* The members of the third group below do not need to be coordinated with data
3ffc8c70 948 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
949# define _CC_IDFIRST 16
950# define _CC_CHARNAME_CONT 17
951# define _CC_NONLATIN1_FOLD 18
952# define _CC_NONLATIN1_SIMPLE_FOLD 19
953# define _CC_QUOTEMETA 20
954# define _CC_NON_FINAL_FOLD 21
955# define _CC_IS_IN_SOME_FOLD 22
956# define _CC_MNEMONIC_CNTRL 23
957/* Unused: 24-31
f4cdb42c
KW
958 * If more bits are needed, one could add a second word for non-64bit
959 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
960 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
961 * is used only for optimization (as of this writing), and differs in the
962 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72
KW
963 * characters: the masculine and feminine ordinal indicators, so removing it
964 * would just cause /i regexes which match them to run less efficiently */
96ac0975 965
3a371f2f
KW
966#if defined(PERL_CORE) || defined(PERL_EXT)
967/* An enum version of the character class numbers, to help compilers
968 * optimize */
969typedef enum {
3a371f2f 970 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
971 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
972 _CC_ENUM_ASCII = _CC_ASCII,
973 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 974 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 975 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
976 _CC_ENUM_DIGIT = _CC_DIGIT,
977 _CC_ENUM_GRAPH = _CC_GRAPH,
978 _CC_ENUM_LOWER = _CC_LOWER,
979 _CC_ENUM_PRINT = _CC_PRINT,
980 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 981 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 982 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 983 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 984 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 985 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
986} _char_class_number;
987#endif
988
63c61c3f 989#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 990#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 991
3018b823 992#if defined(PERL_IN_UTF8_C) || defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_REGEXEC_C)
922e8cb4 993# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
994 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
995 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
996 #error Need to adjust order of swash_property_names[]
997# endif
998
b44daeeb
KW
999/* This is declared static in each of the few files that this is #defined for
1000 * to keep them from being publicly accessible. Hence there is a small amount
1001 * of wasted space */
1002
922e8cb4
KW
1003static const char* const swash_property_names[] = {
1004 "XPosixWord",
1005 "XPosixDigit",
1006 "XPosixAlpha",
1007 "XPosixLower",
1008 "XPosixUpper",
1009 "XPosixPunct",
1010 "XPosixPrint",
1011 "XPosixAlnum",
b0d691b2
KW
1012 "XPosixGraph",
1013 "Cased"
922e8cb4
KW
1014};
1015#endif
1016
6635f04f 1017START_EXTERN_C
96ac0975
NC
1018# ifdef DOINIT
1019EXTCONST U32 PL_charclass[] = {
1020# include "l1_char_class_tab.h"
1021};
1022
1023# else /* ! DOINIT */
1024EXTCONST U32 PL_charclass[];
1025# endif
6635f04f 1026END_EXTERN_C
96ac0975 1027
265c1f46 1028 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1029# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1030
1031 /* For internal core Perl use only: the base macro for defining macros like
1032 * isALPHA */
ff7ecfc3 1033# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1034 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1035
f4cdb42c
KW
1036 /* The mask for the _A versions of the macros; it just adds in the bit for
1037 * ASCII. */
1038# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1039
4650c663
KW
1040 /* For internal core Perl use only: the base macro for defining macros like
1041 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1042 * the ASCII bit are present */
3ded5eb0 1043# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
f4cd282c 1044 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
f4cdb42c
KW
1045 == _CC_mask_A(classnum)))
1046
f4cdb42c 1047# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1048# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1049# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1050# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1051# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1052# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1053# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1054# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1055# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1056# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1057# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1058# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
96ca48da 1059# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits */
d95f8b6a 1060# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1061# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1062# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1063# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1064
1065 /* continuation character for legal NAME in \N{NAME} */
1066# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1067
1068# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1069# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1070# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1071# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
779cf272 1072# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1073# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1074# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1075# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1076# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1077# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1078
0852beac
KW
1079# ifdef EBCDIC
1080# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1081# endif
1082
f12c0118
KW
1083 /* Participates in a single-character fold with a character above 255 */
1084# 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)))
1085
1086 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1087# 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 1088
4eeeb416 1089# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd
KW
1090# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1091 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926
KW
1092# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1093 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1094# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1095 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1096#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1097
1098 /* If we don't have perl.h, we are compiling a utility program. Below we
1099 * hard-code various macro definitions that wouldn't otherwise be available
0852beac
KW
1100 * to it. Most are coded based on first principals. First some ones common
1101 * to both ASCII and EBCDIC */
1102# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1103# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1104# define isSPACE_A(c) (isBLANK_A(c) \
1105 || (c) == '\n' \
1106 || (c) == '\r' \
1107 || (c) == '\v' \
1108 || (c) == '\f')
1109# ifdef EBCDIC /* There are gaps between 'i' and 'j'; 'r' and 's'. Same
1110 for uppercase. This is ordered to exclude most things
1111 early */
1112# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1113 && ((c) <= 'i' \
1114 || ((c) >= 'j' && (c) <= 'r') \
1115 || (c) >= 's'))
1116# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1117 && ((c) <= 'I' \
1118 || ((c) >= 'J' && (c) <= 'R') \
1119 || (c) >= 'S'))
1120# else /* ASCII platform. */
9fb1bf9d 1121# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z')
2181da29 1122# define isUPPER_A(c) ((c) <= 'Z' && (c) >= 'A')
0852beac
KW
1123# endif
1124
1125 /* Some more ASCII, non-ASCII common definitions */
a4d7a999
KW
1126# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1127# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1128# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac
KW
1129# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1130# define isXDIGIT_A(c) (isDIGIT_A(c) \
1131 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1132 || ((c) <= 'F' && (c) >= 'A'))
3ded5eb0 1133
0852beac
KW
1134# ifdef EBCDIC
1135# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1136 || (c) == '#' || (c) == '$' || (c) == '%' \
1137 || (c) == '&' || (c) == '\'' || (c) == '(' \
1138 || (c) == ')' || (c) == '*' || (c) == '+' \
1139 || (c) == ',' || (c) == '.' || (c) == '/' \
1140 || (c) == ':' || (c) == ';' || (c) == '<' \
1141 || (c) == '=' || (c) == '>' || (c) == '?' \
1142 || (c) == '@' || (c) == '[' || (c) == '\\' \
1143 || (c) == ']' || (c) == '^' || (c) == '_' \
1144 || (c) == '`' || (c) == '{' || (c) == '|' \
1145 || (c) == '}' || (c) == '~')
1146# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1147# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1148
1149# ifdef QUESTION_MARK_CTRL
1150# define _isQMC(c) ((c) == QUESTION_MARK_CTRL)
1151# else
1152# define _isQMC(c) 0
1153# endif
1154
1155 /* I (khw) can't think of a way to define all the ASCII controls
1156 * without resorting to a libc (locale-sensitive) call. But we know
1157 * that all controls but the question-mark one are in the range 0-0x3f.
1158 * This makes sure that all the controls that have names are included,
1159 * and all controls that are also considered ASCII in the locale. This
1160 * may include more or fewer than what it actually should, but the
1161 * wrong ones are less-important controls, so likely won't impact
1162 * things (keep in mind that this is compiled only if perl.h isn't
1163 * available). The question mark control is included if available */
1164# define isCNTRL_A(c) (((c) < 0x40 && isascii(c)) \
1165 || (c) == '\0' || (c) == '\a' || (c) == '\b' \
1166 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1167 || (c) == '\t' || (c) == '\v' || _isQMC(c))
1168
1169# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1170# else /* ASCII platform; things are simpler, and isASCII has already
1171 been defined */
1172# define isGRAPH_A(c) (((c) > ' ' && (c) < 127))
1173# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1174# define isPUNCT_A(c) (isGRAPH_A(c) && (! isALPHANUMERIC_A(c)))
1175# define isCNTRL_A(c) (isASCII(c) && (! isPRINT_A(c)))
1176# endif
1177
3ffc8c70 1178 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1179 * during debugging, and it seems best to keep them. We may be called
1180 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1181 * do anything anyway, so make it not a problem */
1182# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1183# define NATIVE_TO_LATIN1(ch) (ch)
1184# endif
3ded5eb0
KW
1185# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1186# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1187# define isBLANK_L1(c) (isBLANK_A(c) \
1188 || (FITS_IN_8_BITS(c) \
1189 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1190# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1191# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1192# define isLOWER_L1(c) (isLOWER_A(c) \
1193 || (FITS_IN_8_BITS(c) \
1194 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1195 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1196 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1197 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1198 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1199# define isPRINT_L1(c) (isPRINT_A(c) \
1200 || (FITS_IN_8_BITS(c) \
1201 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1202# define isPUNCT_L1(c) (isPUNCT_A(c) \
1203 || (FITS_IN_8_BITS(c) \
1204 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1205 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1206 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1207 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1208 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1209 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1210 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1211# define isSPACE_L1(c) (isSPACE_A(c) \
1212 || (FITS_IN_8_BITS(c) \
1213 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1214 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1215# define isUPPER_L1(c) (isUPPER_A(c) \
1216 || (FITS_IN_8_BITS(c) \
1217 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1218 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1219 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1220# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1221# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1222# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1223 || isBLANK_L1(c) \
1224 || (c) == '-' \
1225 || (c) == '(' \
1226 || (c) == ')')
1227 /* The following are not fully accurate in the above-ASCII range. I (khw)
1228 * don't think it's necessary to be so for the purposes where this gets
1229 * compiled */
1230# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1231# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1232
1233 /* And these aren't accurate at all. They are useful only for above
1234 * Latin1, which utilities and bootstrapping don't deal with */
1235# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1236# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1237# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1238
1239 /* Many of the macros later in this file are defined in terms of these. By
1240 * implementing them with a function, which converts the class number into
1241 * a call to the desired macro, all of the later ones work. However, that
1242 * function won't be actually defined when building a utility program (no
1243 * perl.h), and so a compiler error will be generated if one is attempted
1244 * to be used. And the above-Latin1 code points require Unicode tables to
1245 * be present, something unlikely to be the case when bootstrapping */
1246# define _generic_isCC(c, classnum) \
1247 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1248# define _generic_isCC_A(c, classnum) \
1249 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1250#endif /* End of no perl.h H_PERL */
8a58bdcf 1251
e66b99e9
KW
1252#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1253#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1254#define isASCII_A(c) isASCII(c)
1255#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1256#define isBLANK(c) isBLANK_A(c)
1257#define isCNTRL(c) isCNTRL_A(c)
1258#define isDIGIT(c) isDIGIT_A(c)
1259#define isGRAPH(c) isGRAPH_A(c)
1260#define isIDFIRST(c) isIDFIRST_A(c)
1261#define isLOWER(c) isLOWER_A(c)
1262#define isPRINT(c) isPRINT_A(c)
779cf272 1263#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1264#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1265#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1266#define isPUNCT(c) isPUNCT_A(c)
1267#define isSPACE(c) isSPACE_A(c)
1268#define isUPPER(c) isUPPER_A(c)
1269#define isWORDCHAR(c) isWORDCHAR_A(c)
1270#define isXDIGIT(c) isXDIGIT_A(c)
1271
1272/* ASCII casing. These could also be written as
1273 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1274 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1275 which uses table lookup and mask instead of subtraction. (This would
1276 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1277#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1278#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1279
25200305
KW
1280/* In the ASCII range, these are equivalent to what they're here defined to be.
1281 * But by creating these definitions, other code doesn't have to be aware of
1282 * this detail */
1283#define toFOLD(c) toLOWER(c)
25200305
KW
1284#define toTITLE(c) toUPPER(c)
1285
c753c8d3
KW
1286#define toLOWER_A(c) toLOWER(c)
1287#define toUPPER_A(c) toUPPER(c)
25200305
KW
1288#define toFOLD_A(c) toFOLD(c)
1289#define toTITLE_A(c) toTITLE(c)
1a0901db 1290
4650c663 1291/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1292#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1293 ? (c) \
f4cd282c 1294 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1295#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1296
1a0901db 1297/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1298 * all mapped to one of them, and these need special handling; returns the
1299 * input itself if is out-of-range */
b2bf251f 1300#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1301 ? (c) \
f4cd282c 1302 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1303#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1304
beab9ebe
KW
1305/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1306
1307/* For internal core Perl use only: the base macro for defining macros like
1308 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1309 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1310 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1311 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1312 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1313 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1314 * code points under locale rules, use the macros based on _generic_LC_uvchr
1315 * instead of this. */
beab9ebe
KW
1316#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1317 (! FITS_IN_8_BITS(c) \
1318 ? 0 \
31f05a37
KW
1319 : IN_UTF8_CTYPE_LOCALE \
1320 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1321 : cBOOL(non_utf8))
1322
1323/* For internal core Perl use only: a helper macro for defining macros like
1324 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1325 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1326 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1327 * _generic_LC_base for more info */
1328#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1329 _generic_LC_base(c,utf8_locale_classnum, \
1330 non_utf8_func( (_LC_CAST) (c)))
1331
1332/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1333 * 'c' is the platform's native underscore character */
1334#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1335 _generic_LC_base(c, utf8_locale_classnum, \
1336 (non_utf8_func( (_LC_CAST) (c)) \
1337 || (char)(c) == '_'))
1338
1339/* These next three are also for internal core Perl use only: case-change
1340 * helper macros */
1341#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1342 ? (c) \
31f05a37
KW
1343 : (IN_UTF8_CTYPE_LOCALE) \
1344 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1345 : (cast)function((cast)(c)))
beab9ebe 1346
31f05a37
KW
1347/* Note that the result can be larger than a byte in a UTF-8 locale. It
1348 * returns a single value, so can't adequately return the upper case of LATIN
1349 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1350 * values "SS"); instead it asserts against that under DEBUGGING, and
1351 * otherwise returns its input */
beab9ebe
KW
1352#define _generic_toUPPER_LC(c, function, cast) \
1353 (! FITS_IN_8_BITS(c) \
1354 ? (c) \
31f05a37 1355 : ((! IN_UTF8_CTYPE_LOCALE) \
d2a8cf7f 1356 ? (cast)function((cast)(c)) \
31f05a37
KW
1357 : ((((U8)(c)) == MICRO_SIGN) \
1358 ? GREEK_CAPITAL_LETTER_MU \
1359 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1360 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1361 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1362 ? (__ASSERT_(0) (c)) \
1363 : PL_mod_latin1_uc[ (U8) (c) ])))))
1364
1365/* Note that the result can be larger than a byte in a UTF-8 locale. It
1366 * returns a single value, so can't adequately return the fold case of LATIN
1367 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1368 * values "ss"); instead it asserts against that under DEBUGGING, and
1369 * otherwise returns its input */
beab9ebe 1370#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1371 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1372 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1373 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1374 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1375 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1376
84061b6a 1377/* Use the libc versions for these if available. */
f05550c0 1378#if defined(HAS_ISASCII)
84061b6a
KW
1379# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1380#else
1381# define isASCII_LC(c) isASCII(c)
1382#endif
1383
f05550c0 1384#if defined(HAS_ISBLANK)
84061b6a 1385# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1386#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1387# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1388#endif
1389
f05550c0 1390#define _LC_CAST U8
bbce6d69 1391
f05550c0 1392#ifdef WIN32
375f5f06
KW
1393 /* The Windows functions don't bother to follow the POSIX standard, which
1394 * for example says that something can't both be a printable and a control.
1395 * But Windows treats the \t control as a printable, and does such things
1396 * as making superscripts into both digits and punctuation. This tames
1397 * these flaws by assuming that the definitions of both controls and space
1398 * are correct, and then making sure that other definitions don't have
1399 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1400 * Not all possible weirdnesses are checked for, just the ones that were
1401 * detected on actual Microsoft code pages */
1402
f05550c0
BF
1403# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1404# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1405
1406# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) && isALPHANUMERIC_LC(c))
1407# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && ! isPUNCT_LC(c))
1408# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && isALPHANUMERIC_LC(c))
1409# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1410# define isIDFIRST_LC(c) (((c) == '_') || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1411# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1412# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1413# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1414# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1415# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1416# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) && isALPHANUMERIC_LC(c))
1417
1418# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1419# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1420# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1421
1422#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1423 /* For most other platforms */
beab9ebe 1424
f05550c0
BF
1425# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1426# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1427# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1428# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1429# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1430# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1431# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1432# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1433# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1434# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1435# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1436# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1437# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1438
1439
1440# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1441# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1442# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1443
1444#else /* The final fallback position */
1445
1446# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1447# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1448# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1449# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1450# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1451# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1452# define isLOWER_LC(c) (isascii(c) && islower(c))
1453# define isPRINT_LC(c) (isascii(c) && isprint(c))
1454# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1455# define isSPACE_LC(c) (isascii(c) && isspace(c))
1456# define isUPPER_LC(c) (isascii(c) && isupper(c))
1457# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1458# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1459
1460# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1461# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1462# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1463
f05550c0 1464#endif
55204971 1465
eba68aa0
KW
1466#define isIDCONT(c) isWORDCHAR(c)
1467#define isIDCONT_A(c) isWORDCHAR_A(c)
1468#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1469#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1470#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1471
4650c663
KW
1472/* For internal core Perl use only: the base macros for defining macros like
1473 * isALPHA_uni. 'c' is the code point to check. 'classnum' is the POSIX class
1474 * number defined earlier in this file. _generic_uni() is used for POSIX
1475 * classes where there is a macro or function 'above_latin1' that takes the
1476 * single argument 'c' and returns the desired value. These exist for those
1477 * classes which have simple definitions, avoiding the overhead of a hash
1478 * lookup or inversion list binary search. _generic_swash_uni() can be used
1479 * for classes where that overhead is faster than a direct lookup.
1480 * _generic_uni() won't compile if 'c' isn't unsigned, as it won't match the
1481 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1482 * have duplicate checks here, so could create versions of the macros that
1483 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1484
1485/* Note that all ignore 'use bytes' */
cd500f2f
KW
1486#define _generic_uni(classnum, above_latin1, c) ((c) < 256 \
1487 ? _generic_isCC(c, classnum) \
1488 : above_latin1(c))
1489#define _generic_swash_uni(classnum, c) ((c) < 256 \
1490 ? _generic_isCC(c, classnum) \
922e8cb4 1491 : _is_uni_FOO(classnum, c))
cd500f2f
KW
1492#define isALPHA_uni(c) _generic_swash_uni(_CC_ALPHA, c)
1493#define isALPHANUMERIC_uni(c) _generic_swash_uni(_CC_ALPHANUMERIC, c)
509fb054
KW
1494#define isASCII_uni(c) isASCII(c)
1495#define isBLANK_uni(c) _generic_uni(_CC_BLANK, is_HORIZWS_cp_high, c)
1496#define isCNTRL_uni(c) isCNTRL_L1(c) /* All controls are in Latin1 */
cd500f2f
KW
1497#define isDIGIT_uni(c) _generic_swash_uni(_CC_DIGIT, c)
1498#define isGRAPH_uni(c) _generic_swash_uni(_CC_GRAPH, c)
eba68aa0 1499#define isIDCONT_uni(c) _generic_uni(_CC_WORDCHAR, _is_uni_perl_idcont, c)
509fb054 1500#define isIDFIRST_uni(c) _generic_uni(_CC_IDFIRST, _is_uni_perl_idstart, c)
cd500f2f
KW
1501#define isLOWER_uni(c) _generic_swash_uni(_CC_LOWER, c)
1502#define isPRINT_uni(c) _generic_swash_uni(_CC_PRINT, c)
aea0c08c 1503
cd500f2f 1504#define isPUNCT_uni(c) _generic_swash_uni(_CC_PUNCT, c)
509fb054 1505#define isSPACE_uni(c) _generic_uni(_CC_SPACE, is_XPERLSPACE_cp_high, c)
779cf272
KW
1506#define isPSXSPC_uni(c) isSPACE_uni(c)
1507
cd500f2f 1508#define isUPPER_uni(c) _generic_swash_uni(_CC_UPPER, c)
509fb054 1509#define isVERTWS_uni(c) _generic_uni(_CC_VERTSPACE, is_VERTWS_cp_high, c)
cd500f2f 1510#define isWORDCHAR_uni(c) _generic_swash_uni(_CC_WORDCHAR, c)
509fb054 1511#define isXDIGIT_uni(c) _generic_uni(_CC_XDIGIT, is_XDIGIT_cp_high, c)
b18192e9 1512
b8d68ded 1513#define toFOLD_uni(c,s,l) to_uni_fold(c,s,l)
509fb054
KW
1514#define toLOWER_uni(c,s,l) to_uni_lower(c,s,l)
1515#define toTITLE_uni(c,s,l) to_uni_title(c,s,l)
1516#define toUPPER_uni(c,s,l) to_uni_upper(c,s,l)
a0ed51b3 1517
4650c663
KW
1518/* For internal core Perl use only: the base macros for defining macros like
1519 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1520 * point, not just 0-255. Like _generic_uni, there are two versions, one for
1521 * simple class definitions; the other for more complex. These are like
1522 * _generic_uni, so see it for more info. */
cd500f2f
KW
1523#define _generic_LC_uvchr(latin1, above_latin1, c) \
1524 (c < 256 ? latin1(c) : above_latin1(c))
1525#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1526 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1527
1528#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1529#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1530 _CC_ALPHANUMERIC, c)
509fb054 1531#define isASCII_LC_uvchr(c) isASCII_LC(c)
cd500f2f 1532#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, is_HORIZWS_cp_high, c)
feeab5a9 1533#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1534#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1535#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
eba68aa0
KW
1536#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
1537 _is_uni_perl_idcont, c)
cd500f2f
KW
1538#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
1539 _is_uni_perl_idstart, c)
1540#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1541#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
779cf272 1542#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f
KW
1543#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
1544#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1545 is_XPERLSPACE_cp_high, c)
cd500f2f
KW
1546#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
1547#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
1548 _CC_WORDCHAR, c)
1549#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, is_XDIGIT_cp_high, c)
e712593e 1550
36b00f93 1551#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1552
4650c663
KW
1553/* For internal core Perl use only: the base macros for defining macros like
1554 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1555 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1556 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1557 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1558 * same representation whether utf8 or not. Note that it assumes that the utf8
1559 * has been validated, and ignores 'use bytes' */
cd500f2f 1560#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1561 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1562 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1563 ? _generic_isCC( \
94bb8c36 1564 TWO_BYTE_UTF8_TO_NATIVE(*(p), \
aea0c08c 1565 *((p)+1 )), \
94bb8c36 1566 classnum) \
252810d0 1567 : utf8)
cd500f2f
KW
1568/* Like the above, but calls 'above_latin1(p)' to get the utf8 value. 'above_latin1'
1569 * can be a macro */
1570#define _generic_func_utf8(classnum, above_latin1, p) \
1571 _generic_utf8(classnum, p, above_latin1(p))
4650c663 1572/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1573 * 'above_latin1' parameter */
922e8cb4 1574#define _generic_swash_utf8(classnum, p) \
cd500f2f 1575 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
922e8cb4 1576
cc8ab7c0 1577/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1578 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1579 * class is TRUE for. Hence it can skip the tests for this range.
1580 * 'above_latin1' should include its arguments */
3ded5eb0 1581#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1582 (UTF8_IS_INVARIANT(*(p)) \
1583 ? _generic_isCC(*(p), classnum) \
1584 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1585 ? above_latin1 \
cc8ab7c0
KW
1586 : 0)
1587
84238efa
KW
1588/* NOTE that some of these macros have very similar ones in regcharclass.h.
1589 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1590 * there, differing in name only by an underscore from the one here
1591 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1592 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1593 * points; the regcharclass.h ones are implemented as a series of
1594 * "if-else-if-else ..." */
1595
922e8cb4
KW
1596#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1597#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
509fb054
KW
1598#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
1599 utf8, the non-utf8 macro works
1600 */
cd500f2f 1601#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2
KW
1602
1603#ifdef EBCDIC
1604 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1605 * more efficient macro instead of the more general one */
88794300 1606# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
e8fa43e2
KW
1607#else
1608# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1609#endif
1610
cd500f2f
KW
1611#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1612 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1613#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1614#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1615 _is_utf8_perl_idcont, p)
e5dcd934 1616
c11ff943
KW
1617/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1618 * IDFIRST is an alnum. See
c8362b00 1619 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1620 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1621 * This used to be not the XID version, but we decided to go with the more
1622 * modern Unicode definition */
cd500f2f 1623#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1624 _is_utf8_perl_idstart, p)
e5dcd934 1625
922e8cb4
KW
1626#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1627#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
779cf272 1628#define isPSXSPC_utf8(p) isSPACE_utf8(p)
922e8cb4 1629#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
cd500f2f 1630#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
922e8cb4 1631#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
cd500f2f 1632#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
922e8cb4 1633#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
cd500f2f
KW
1634#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
1635 is_XDIGIT_high(p))
a0ed51b3 1636
7d2b61dd 1637#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1638#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1639#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1640#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1641
4650c663
KW
1642/* For internal core Perl use only: the base macros for defining macros like
1643 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1644 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1645 * 'macro' parameter */
94bb8c36
KW
1646#define _generic_LC_utf8(macro, p, utf8) \
1647 (UTF8_IS_INVARIANT(*(p)) \
1648 ? macro(*(p)) \
1649 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1650 ? macro(TWO_BYTE_UTF8_TO_NATIVE(*(p), *((p)+1))) \
9201a572
KW
1651 : utf8)
1652
cd500f2f
KW
1653#define _generic_LC_swash_utf8(macro, classnum, p) \
1654 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1655#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1656 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1657
cd500f2f
KW
1658#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
1659 _CC_ALPHANUMERIC, p)
1660#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
509fb054 1661#define isASCII_LC_utf8(p) isASCII_LC(*p)
cd500f2f
KW
1662#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, is_HORIZWS_high, p)
1663#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1664#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1665#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
eba68aa0 1666#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, _is_utf8_perl_idcont, p)
cd500f2f
KW
1667#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, _is_utf8_perl_idstart, p)
1668#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1669#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
779cf272 1670#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
cd500f2f
KW
1671#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1672#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, is_XPERLSPACE_high, p)
1673#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
1674#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1675 _CC_WORDCHAR, p)
1676#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, is_XDIGIT_high, p)
aaa51d5e 1677
fbc19f27
KW
1678/* Macros for backwards compatibility and for completeness when the ASCII and
1679 * Latin1 values are identical */
fbc19f27
KW
1680#define isALPHAU(c) isALPHA_L1(c)
1681#define isDIGIT_L1(c) isDIGIT_A(c)
1682#define isOCTAL(c) isOCTAL_A(c)
1683#define isOCTAL_L1(c) isOCTAL_A(c)
1684#define isXDIGIT_L1(c) isXDIGIT_A(c)
2e28f0b9
KW
1685#define isALNUM(c) isWORDCHAR(c)
1686#define isALNUMU(c) isWORDCHAR_L1(c)
1687#define isALNUM_LC(c) isWORDCHAR_LC(c)
1688#define isALNUM_uni(c) isWORDCHAR_uni(c)
1689#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
1690#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1691#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
15861f94
KW
1692#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1693#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1694#define isALNUMC(c) isALPHANUMERIC(c)
1695#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1696#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
1697#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
1698#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
1699#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1700
2bd1cbf6
KW
1701/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1702 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1703 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1704 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1705 * the outlier from the block that contains the other controls, just like
1706 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1707 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
1708 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1709 * etc. */
2bd1cbf6 1710#ifndef EBCDIC
75763b3a 1711# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 1712#else
75763b3a
KW
1713# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1714 ((isPRINT_A(c)) \
1715 ? (UNLIKELY((c) == '?') \
1716 ? QUESTION_MARK_CTRL \
1717 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1718 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1719 ? '?' \
1720 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 1721#endif
bbce6d69 1722
dea28490
JJ
1723/* Line numbers are unsigned, 32 bits. */
1724typedef U32 line_t;
e5dcd934 1725#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 1726
91152fc1
DG
1727/* Helpful alias for version prescan */
1728#define is_LAX_VERSION(a,b) \
1729 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1730
1731#define is_STRICT_VERSION(a,b) \
1732 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1733
1734#define BADVERSION(a,b,c) \
1735 if (b) { \
1736 *b = c; \
1737 } \
1738 return a;
8c52afec 1739
cb27eebd
KW
1740/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1741 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1742 * which is advanced. The input is validated only by an assert() in DEBUGGING
1743 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1744 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1745#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1746 ? (c) \
1747 : ((c) + 9))))
902008b8
KW
1748#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1749 ? (*(s)++) \
1750 : (*(s)++ + 9))))
95a59cab 1751
cb27eebd
KW
1752/* Converts a character known to represent an octal digit (0-7) to its numeric
1753 * value. The input is validated only by an assert() in DEBUGGING builds. In
1754 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1755#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1756
305b8651
KW
1757/* Efficiently returns a boolean as to if two native characters are equivalent
1758 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1759 * the ALPHA in the name is to remind you of that. This is asserted() in
1760 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1761 * works (on valid input) for both non- and UTF-8-encoded bytes.
1762 *
1763 * When one of the inputs is a compile-time constant and gets folded by the
1764 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1765 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1766 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
1767 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1768 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1769 * just a single 0, in the bit position where the upper- and lowercase differ.
1770 * */
305b8651
KW
1771#define isALPHA_FOLD_EQ(c1, c2) \
1772 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1773 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1774#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1775
8e84507e 1776/*
ccfc67b7
JH
1777=head1 Memory Management
1778
a02a5408 1779=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
1780The XSUB-writer's interface to the C C<malloc> function.
1781
0d7b2759
MH
1782Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1783
c5008215
JC
1784In 5.9.3, Newx() and friends replace the older New() API, and drops
1785the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 1786themselves. This aid has been superseded by a new build option,
d10b4965 1787PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
1788there for use in XS modules supporting older perls.
1789
a02a5408 1790=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 1791The XSUB-writer's interface to the C C<malloc> function, with
c5008215 1792cast. See also C<Newx>.
954c1994 1793
0d7b2759
MH
1794Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1795
a02a5408 1796=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 1797The XSUB-writer's interface to the C C<malloc> function. The allocated
c5008215 1798memory is zeroed with C<memzero>. See also C<Newx>.
a02a5408 1799
0d7b2759
MH
1800Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1801
954c1994
GS
1802=for apidoc Am|void|Renew|void* ptr|int nitems|type
1803The XSUB-writer's interface to the C C<realloc> function.
1804
0d7b2759
MH
1805Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1806
954c1994
GS
1807=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1808The XSUB-writer's interface to the C C<realloc> function, with
1809cast.
1810
0d7b2759
MH
1811Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1812
49b8b560 1813=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
1814The XSUB-writer's interface to the C C<free> function.
1815
0d7b2759
MH
1816This should B<ONLY> be used on memory obtained using L<"Newx"> and friends.
1817
954c1994
GS
1818=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1819The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c
KW
1820source, C<dest> is the destination, C<nitems> is the number of items, and
1821C<type> is the type. Can do overlapping moves. See also C<Copy>.
954c1994 1822
e90e2364 1823=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
72d33970
FC
1824Like C<Move> but returns dest. Useful
1825for encouraging compilers to tail-call
e90e2364
NC
1826optimise.
1827
954c1994
GS
1828=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1829The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c
KW
1830source, C<dest> is the destination, C<nitems> is the number of items, and
1831C<type> is the type. May fail on overlapping copies. See also C<Move>.
954c1994 1832
e90e2364
NC
1833=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1834
72d33970
FC
1835Like C<Copy> but returns dest. Useful
1836for encouraging compilers to tail-call
e90e2364
NC
1837optimise.
1838
954c1994
GS
1839=for apidoc Am|void|Zero|void* dest|int nitems|type
1840
1841The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1842destination, C<nitems> is the number of items, and C<type> is the type.
1843
e90e2364
NC
1844=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1845
72d33970
FC
1846Like C<Zero> but returns dest. Useful
1847for encouraging compilers to tail-call
e90e2364
NC
1848optimise.
1849
da5d8dbb 1850=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 1851This is an architecture-independent macro to copy one structure to another.
954c1994 1852
7e337ee0
JH
1853=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1854
1855Fill up memory with a byte pattern (a byte repeated over and over
1856again) that hopefully catches attempts to access uninitialized memory.
1857
1858=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1859
1860PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1861
1c12ffb4 1862=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
1863
1864PoisonWith(0xEF) for catching access to freed memory.
1865
9965345d
JH
1866=for apidoc Am|void|Poison|void* dest|int nitems|type
1867
7e337ee0 1868PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
1869
1870=cut */
954c1994 1871
561b68a9
SH
1872/* Maintained for backwards-compatibility only. Use newSV() instead. */
1873#ifndef PERL_CORE
ff06c60c 1874#define NEWSV(x,len) newSV(len)
561b68a9 1875#endif
ff06c60c 1876
19a94d75
JH
1877#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1878
e6bdf523 1879
27d5b266 1880#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
1881
1882/* This expression will be constant-folded at compile time. It checks
1883 * whether or not the type of the count n is so small (e.g. U8 or U16, or
1884 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
1885 * As well as avoiding the need for a run-time check in some cases, it's
1886 * designed to avoid compiler warnings like:
1887 * comparison is always false due to limited range of data type
1888 */
1889
1890# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
1891 (sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
1892
1893/* this is written in a slightly odd way because for an expression like
1894 * cond && (n > expr)
1895 * even when cond constant-folds to false at compile-time, g++ insists
1896 * on emitting warnings about expr (e.g. "comparison is always false").
1897 * So rewrite it as
1898 * (cond ? n : 1) > expr
1899 *
1900 * We happen to know that (1 > expr) will always be false (unless someone
1901 * is doing something with a struct whose sizeof > MEM_SIZE_MAX/2), so
1902 * this is safe.
1903 */
1904
1905# define _MEM_WRAP_WILL_WRAP(n,t) \
1906 ((MEM_SIZE)(_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? n : 1) \
1907 > MEM_SIZE_MAX/sizeof(t))
1908
1909# define MEM_WRAP_CHECK(n,t) \
1910 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1911 && (croak_memory_wrap(),0))
1912
1913# define MEM_WRAP_CHECK_1(n,t,a) \
1914 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1915 && (Perl_croak_nocontext("%s",(a)),0))
1916
8b44ba4c 1917#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 1918
d1decf2b 1919#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
1920#else
1921
410319be
NC
1922#define MEM_WRAP_CHECK(n,t)
1923#define MEM_WRAP_CHECK_1(n,t,a)
1924#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
1925#define MEM_WRAP_CHECK_(n,t)
1926
6fff79ce 1927#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 1928
1936d2a7 1929#endif
8b44ba4c 1930
fe4f188c 1931#ifdef PERL_MEM_LOG
46c6c7e2 1932/*
9f653bb5 1933 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
1934 * go through functions, which are handy for debugging breakpoints, but
1935 * which more importantly get the immediate calling environment (file and
e352bcff
JH
1936 * line number, and C function name if available) passed in. This info can
1937 * then be used for logging the calls, for which one gets a sample
73d1d973 1938 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 1939 *
46c6c7e2 1940 * Known problems:
94e892a6 1941 * - not all memory allocs get logged, only those
46c6c7e2 1942 * that go through Newx() and derivatives (while all
94e892a6 1943 * Safefrees do get logged)
46c6c7e2
JH
1944 * - __FILE__ and __LINE__ do not work everywhere
1945 * - __func__ or __FUNCTION__ even less so
1946 * - I think more goes on after the perlio frees but
1947 * the thing is that STDERR gets closed (as do all
1948 * the file descriptors)
1949 * - no deeper calling stack than the caller of the Newx()
1950 * or the kind, but do I look like a C reflection/introspection
1951 * utility to you?
1952 * - the function prototypes for the logging functions
1953 * probably should maybe be somewhere else than handy.h
1954 * - one could consider inlining (macrofying) the logging
1955 * for speed, but I am too lazy
1956 * - one could imagine recording the allocations in a hash,
1957 * (keyed by the allocation address?), and maintain that
1958 * through reallocs and frees, but how to do that without
1959 * any News() happening...?
73d1d973 1960 * - lots of -Ddefines to get useful/controllable output
b953482e 1961 * - lots of ENV reads
46c6c7e2
JH
1962 */
1963
12754f92 1964PERL_EXPORT_C Malloc_t Perl_mem_log_alloc(const UV n, const UV typesize, const char *type_name, Malloc_t newalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 1965
12754f92 1966PERL_EXPORT_C Malloc_t Perl_mem_log_realloc(const UV n, const UV typesize, const char *type_name, Malloc_t oldalloc, Malloc_t newalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 1967
12754f92 1968PERL_EXPORT_C Malloc_t Perl_mem_log_free(Malloc_t oldalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 1969
0b0ab801 1970# ifdef PERL_CORE
73d1d973 1971# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
1972enum mem_log_type {
1973 MLT_ALLOC,
1974 MLT_REALLOC,
d7a2c63c
MHM
1975 MLT_FREE,
1976 MLT_NEW_SV,
1977 MLT_DEL_SV
0b0ab801
MHM
1978};
1979# endif
12754f92 1980# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
1981void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
1982void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 1983# endif
0b0ab801
MHM
1984# endif
1985
fe4f188c
JH
1986#endif
1987
1988#ifdef PERL_MEM_LOG
d1401ee9
MHM
1989#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
1990#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 1991#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
1992#endif
1993
1994#ifndef MEM_LOG_ALLOC
1995#define MEM_LOG_ALLOC(n,t,a) (a)
1996#endif
1997#ifndef MEM_LOG_REALLOC
1998#define MEM_LOG_REALLOC(n,t,v,a) (a)
1999#endif
2000#ifndef MEM_LOG_FREE
2001#define MEM_LOG_FREE(a) (a)
2002#endif
2003
d1401ee9
MHM
2004#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2005#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2006#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2007
2008#ifndef PERL_CORE
a02a5408
JC
2009/* pre 5.9.x compatibility */
2010#define New(x,v,n,t) Newx(v,n,t)
2011#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2012#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2013#endif
a02a5408 2014
ff68c719 2015#define Renew(v,n,t) \
d1401ee9 2016 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2017#define Renewc(v,n,t,c) \
d1401ee9 2018 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2019
2020#ifdef PERL_POISON
2021#define Safefree(d) \
06c0cc96 2022 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2023#else
fe4f188c 2024#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2025#endif
55497cff 2026
8b44ba4c
NC
2027#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2028#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2029#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2030
8b44ba4c
NC
2031#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2032#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2033#ifdef HAS_MEMSET
8b44ba4c 2034#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2035#else
8b44ba4c
NC
2036/* Using bzero(), which returns void. */
2037#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2038#endif
2039
7e337ee0
JH
2040#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2041#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2042#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2043#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2044
caa674f3
DD
2045#ifdef PERL_POISON
2046# define PERL_POISON_EXPR(x) x
2047#else
2048# define PERL_POISON_EXPR(x)
2049#endif
2050
2304df62 2051#ifdef USE_STRUCT_COPY
ff68c719 2052#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2053#else
2054#define StructCopy(s,d,t) Copy(s,d,1,t)
2055#endif
2cc61e15 2056
c3caa5c3
JH
2057/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2058 * want your zero-based indices to be less than but not equal to).
2059 *
2060 * C_ARRAY_END is one past the last: half-open/half-closed range,
2061 * not last-inclusive range. */
622913ab 2062#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2063#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2064
2cc61e15
DD
2065#ifdef NEED_VA_COPY
2066# ifdef va_copy
2067# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2068# else
a1866d1b
JH
2069# if defined(__va_copy)
2070# define Perl_va_copy(s, d) __va_copy(d, s)
2071# else
2072# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2073# endif
2cc61e15
DD
2074# endif
2075#endif
2076
472d47bc
SB
2077/* convenience debug macros */
2078#ifdef USE_ITHREADS
2079#define pTHX_FORMAT "Perl interpreter: 0x%p"
2080#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2081#define pTHX_VALUE_ (void *)my_perl,
2082#define pTHX_VALUE (void *)my_perl
2083#define pTHX__VALUE_ ,(void *)my_perl,
2084#define pTHX__VALUE ,(void *)my_perl
472d47bc 2085#else
3609ea0d 2086#define pTHX_FORMAT
472d47bc 2087#define pTHX__FORMAT
3609ea0d 2088#define pTHX_VALUE_
472d47bc 2089#define pTHX_VALUE
3609ea0d 2090#define pTHX__VALUE_
472d47bc
SB
2091#define pTHX__VALUE
2092#endif /* USE_ITHREADS */
3609ea0d 2093
2acdbac1
NC
2094/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2095 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2096 CPAN::Unpack show any users outside the core. */
2097#ifdef PERL_CORE
d1d15184 2098# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), "Use of " s " is deprecated")
2acdbac1
NC
2099#endif
2100
dfff4baf
BF
2101/* Internal macros to deal with gids and uids */
2102#ifdef PERL_CORE
2103
2104# if Uid_t_size > IVSIZE
2105# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2106# define SvUID(sv) SvNV(sv)
2107# else
2108# if Uid_t_sign <= 0
2109# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2110# define SvUID(sv) SvIV(sv)
2111# else
2112# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2113# define SvUID(sv) SvUV(sv)
2114# endif
2115# endif /* Uid_t_size */
2116
2117# if Gid_t_size > IVSIZE
2118# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2119# define SvGID(sv) SvNV(sv)
2120# else
2121# if Gid_t_sign <= 0
2122# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2123# define SvGID(sv) SvIV(sv)
2124# else
2125# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2126# define SvGID(sv) SvUV(sv)
2127# endif
2128# endif /* Gid_t_size */
2129
2130#endif
2131
9d745869
KW
2132#endif /* HANDY_H */
2133
e9a8c099
MHM
2134/*
2135 * Local variables:
2136 * c-indentation-style: bsd
2137 * c-basic-offset: 4
14d04a33 2138 * indent-tabs-mode: nil
e9a8c099
MHM
2139 * End:
2140 *
14d04a33 2141 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2142 */