This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Make new macros private
[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
6a5bc5ac
KW
14#ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */
15#define PERL_HANDY_H_
9d745869 16
24792b8d
NC
17#ifndef PERL_CORE
18# define Null(type) ((type)NULL)
954c1994
GS
19
20/*
ccfc67b7 21=head1 Handy Values
954c1994 22
3609ea0d 23=for apidoc AmU||Nullch
72d33970
FC
24Null character pointer. (No longer available when C<PERL_CORE> is
25defined.)
2307c6d0 26
954c1994 27=for apidoc AmU||Nullsv
72d33970 28Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
29
30=cut
31*/
32
24792b8d
NC
33# define Nullch Null(char*)
34# define Nullfp Null(PerlIO*)
35# define Nullsv Null(SV*)
36#endif
8d063cd8 37
641d3f0b
PP
38#ifdef TRUE
39#undef TRUE
40#endif
41#ifdef FALSE
42#undef FALSE
43#endif
44#define TRUE (1)
45#define FALSE (0)
46
cf3f0ffb
DM
47/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
48 * (compiler permitting) that casting away const-ness will give a warning;
49 * e.g.:
50 *
51 * const SV *sv = ...;
52 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
53 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
54 */
55
b1bc3f34 56#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
f065df04 57# define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
b1bc3f34
NC
58#else
59# define MUTABLE_PTR(p) ((void *) (p))
60#endif
61
a062e10d 62#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 63#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 64#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 65#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 66#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 67#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 68
f789f6a4 69#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
70# include <stdbool.h>
71# ifndef HAS_BOOL
72# define HAS_BOOL 1
73# endif
74#endif
75
8e84507e 76/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
77 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
78 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 79 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
80 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
81 g++ can be identified by __GNUG__.
82 Andy Dougherty February 2000
5d94fbed 83*/
3609ea0d 84#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 85# ifndef PERL_BOOL_AS_CHAR
5d94fbed 86# ifndef HAS_BOOL
c1d22f6b 87# define HAS_BOOL 1
5d94fbed 88# endif
f789f6a4 89# endif
5d94fbed
AD
90#endif
91
92#ifndef HAS_BOOL
f789f6a4
FC
93# ifdef bool
94# undef bool
95# endif
70d5cb32 96# define bool char
c1d22f6b 97# define HAS_BOOL 1
a687059c 98#endif
0d3e774c 99
236415e2
KW
100/* cast-to-bool. A simple (bool) cast may not do the right thing: if bool is
101 * defined as char for example, then the cast from int is
102 * implementation-defined (bool)!!(cbool) in a ternary triggers a bug in xlc on
103 * AIX */
18f5643b 104#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 105
46c6c7e2 106/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
107 * XXX Should really be a Configure probe, with HAS__FUNCTION__
108 * and FUNCTION__ as results.
109 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
46c6c7e2
JH
110#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
111# define FUNCTION__ __func__
07798b17
AC
112#elif (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
113 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
114# define FUNCTION__ ""
46c6c7e2 115#else
07798b17 116# define FUNCTION__ __FUNCTION__ /* Common extension. */
46c6c7e2
JH
117#endif
118
27d4fb96
PP
119/* XXX A note on the perl source internal type system. The
120 original intent was that I32 be *exactly* 32 bits.
121
122 Currently, we only guarantee that I32 is *at least* 32 bits.
123 Specifically, if int is 64 bits, then so is I32. (This is the case
124 for the Cray.) This has the advantage of meshing nicely with
125 standard library calls (where we pass an I32 and the library is
126 expecting an int), but the disadvantage that an I32 is not 32 bits.
127 Andy Dougherty August 1996
24fef2a7 128
dc45a647
MB
129 There is no guarantee that there is *any* integral type with
130 exactly 32 bits. It is perfectly legal for a system to have
131 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 132
dc45a647
MB
133 Similarly, there is no guarantee that I16 and U16 have exactly 16
134 bits.
693762b4 135
8e84507e
NIS
136 For dealing with issues that may arise from various 32/64-bit
137 systems, we will ask Configure to check out
8175356b 138
3609ea0d
JH
139 SHORTSIZE == sizeof(short)
140 INTSIZE == sizeof(int)
141 LONGSIZE == sizeof(long)
dc45a647 142 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 143 PTRSIZE == sizeof(void *)
dc45a647
MB
144 DOUBLESIZE == sizeof(double)
145 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 146
27d4fb96
PP
147*/
148
69512466
JH
149#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
150# include <inttypes.h>
dd0eed91
JH
151# ifdef INT32_MIN_BROKEN
152# undef INT32_MIN
153# define INT32_MIN (-2147483647-1)
154# endif
155# ifdef INT64_MIN_BROKEN
156# undef INT64_MIN
157# define INT64_MIN (-9223372036854775807LL-1)
158# endif
69512466
JH
159#endif
160
8175356b
JH
161typedef I8TYPE I8;
162typedef U8TYPE U8;
163typedef I16TYPE I16;
164typedef U16TYPE U16;
165typedef I32TYPE I32;
166typedef U32TYPE U32;
16d89be8
JH
167
168#ifdef HAS_QUAD
8175356b
JH
169typedef I64TYPE I64;
170typedef U64TYPE U64;
16d89be8 171#endif
8175356b 172
9c39b21c
JH
173/* INT64_C/UINT64_C are C99 from <stdint.h> (so they will not be
174 * available in strict C89 mode), but they are nice, so let's define
175 * them if necessary. */
0d3bd0a9 176#if defined(HAS_QUAD)
9c39b21c
JH
177# undef PeRl_INT64_C
178# undef PeRl_UINT64_C
179/* Prefer the native integer types (int and long) over long long
180 * (which is not C89) and Win32-specific __int64. */
181# if QUADKIND == QUAD_IS_INT && INTSIZE == 8
182# define PeRl_INT64_C(c) (c)
183# define PeRl_UINT64_C(c) CAT2(c,U)
184# endif
185# if QUADKIND == QUAD_IS_LONG && LONGSIZE == 8
186# define PeRl_INT64_C(c) CAT2(c,L)
187# define PeRl_UINT64_C(c) CAT2(c,UL)
188# endif
189# if QUADKIND == QUAD_IS_LONG_LONG && defined(HAS_LONG_LONG)
190# define PeRl_INT64_C(c) CAT2(c,LL)
191# define PeRl_UINT64_C(c) CAT2(c,ULL)
192# endif
193# if QUADKIND == QUAD_IS___INT64
194# define PeRl_INT64_C(c) CAT2(c,I64)
195# define PeRl_UINT64_C(c) CAT2(c,UI64)
196# endif
197# ifndef PeRl_INT64_C
702c92eb
JH
198# define PeRl_INT64_C(c) ((I64)(c)) /* last resort */
199# define PeRl_UINT64_C(c) ((U64)(c))
9c39b21c
JH
200# endif
201/* In OS X the INT64_C/UINT64_C are defined with LL/ULL, which will
202 * not fly with C89-pedantic gcc, so let's undefine them first so that
203 * we can redefine them with our native integer preferring versions. */
8eafc6eb 204# if defined(PERL_DARWIN) && defined(PERL_GCC_PEDANTIC)
9c39b21c
JH
205# undef INT64_C
206# undef UINT64_C
207# endif
208# ifndef INT64_C
209# define INT64_C(c) PeRl_INT64_C(c)
210# endif
211# ifndef UINT64_C
212# define UINT64_C(c) PeRl_UINT64_C(c)
213# endif
e8c95190 214#endif
dc750f2e 215
d8668976 216#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 217
5ff3f7a4
GS
218/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
219 Please search CHAR_MAX in perl.h for further details. */
220#define U8_MAX UINT8_MAX
221#define U8_MIN UINT8_MIN
222
5ff3f7a4
GS
223#define I16_MAX INT16_MAX
224#define I16_MIN INT16_MIN
225#define U16_MAX UINT16_MAX
226#define U16_MIN UINT16_MIN
227
5ff3f7a4
GS
228#define I32_MAX INT32_MAX
229#define I32_MIN INT32_MIN
0e983133
GS
230#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
231# define U32_MAX UINT32_MAX
232#else
233# define U32_MAX 4294967295U
234#endif
5ff3f7a4
GS
235#define U32_MIN UINT32_MIN
236
237#else
238
5c9fa16e
KA
239/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
240 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
241#define U8_MAX PERL_UCHAR_MAX
242#define U8_MIN PERL_UCHAR_MIN
79072805 243
27d4fb96
PP
244#define I16_MAX PERL_SHORT_MAX
245#define I16_MIN PERL_SHORT_MIN
246#define U16_MAX PERL_USHORT_MAX
247#define U16_MIN PERL_USHORT_MIN
79072805 248
c4f23d77 249#if LONGSIZE > 4
27d4fb96
PP
250# define I32_MAX PERL_INT_MAX
251# define I32_MIN PERL_INT_MIN
252# define U32_MAX PERL_UINT_MAX
253# define U32_MIN PERL_UINT_MIN
79072805 254#else
27d4fb96
PP
255# define I32_MAX PERL_LONG_MAX
256# define I32_MIN PERL_LONG_MIN
257# define U32_MAX PERL_ULONG_MAX
258# define U32_MIN PERL_ULONG_MIN
79072805
LW
259#endif
260
5ff3f7a4
GS
261#endif
262
58a9a5d5 263/* log(2) is pretty close to 0.30103, just in case anyone is grepping for it */
fc36a67e
PP
264#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
265#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
266#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
267
88794300 268/* Unused by core; should be deprecated */
ff68c719 269#define Ctl(ch) ((ch) & 037)
8d063cd8 270
98fce2a4
KW
271#if defined(PERL_CORE) || defined(PERL_EXT)
272# ifndef MIN
273# define MIN(a,b) ((a) < (b) ? (a) : (b))
274# endif
275# ifndef MAX
276# define MAX(a,b) ((a) > (b) ? (a) : (b))
277# endif
278#endif
279
4650c663
KW
280/* This is a helper macro to avoid preprocessor issues, replaced by nothing
281 * unless under DEBUGGING, where it expands to an assert of its argument,
282 * followed by a comma (hence the comma operator). If we just used a straight
3e94db23
JH
283 * assert(), we would get a comma with nothing before it when not DEBUGGING.
284 *
285 * We also use empty definition under Coverity since the __ASSERT__
286 * checks often check for things that Really Cannot Happen, and Coverity
287 * detects that and gets all excited. */
288
289#if defined(DEBUGGING) && !defined(__COVERITY__)
0f092d08
KW
290# define __ASSERT_(statement) assert(statement),
291#else
292# define __ASSERT_(statement)
293#endif
294
3fe05580
MHM
295/*
296=head1 SV-Body Allocation
297
298=for apidoc Ama|SV*|newSVpvs|const char* s
0c395ea5 299Like C<newSVpvn>, but takes a C<NUL>-terminated literal string instead of a
30a15352 300string/length pair.
3fe05580 301
84bafc02 302=for apidoc Ama|SV*|newSVpvs_flags|const char* s|U32 flags
0c395ea5 303Like C<newSVpvn_flags>, but takes a C<NUL>-terminated literal string instead of
30a15352 304a string/length pair.
84bafc02 305
3fe05580 306=for apidoc Ama|SV*|newSVpvs_share|const char* s
0c395ea5 307Like C<newSVpvn_share>, but takes a C<NUL>-terminated literal string instead of
30a15352 308a string/length pair and omits the hash parameter.
3fe05580 309
9dcc53ea 310=for apidoc Am|void|sv_catpvs_flags|SV* sv|const char* s|I32 flags
0c395ea5 311Like C<sv_catpvn_flags>, but takes a C<NUL>-terminated literal string instead
30a15352 312of a string/length pair.
9dcc53ea
Z
313
314=for apidoc Am|void|sv_catpvs_nomg|SV* sv|const char* s
0c395ea5
KW
315Like C<sv_catpvn_nomg>, but takes a C<NUL>-terminated literal string instead of
316a string/length pair.
9dcc53ea 317
7b4e7469 318=for apidoc Am|void|sv_catpvs|SV* sv|const char* s
0c395ea5
KW
319Like C<sv_catpvn>, but takes a C<NUL>-terminated literal string instead of a
320string/length pair.
3fe05580 321
9dcc53ea 322=for apidoc Am|void|sv_catpvs_mg|SV* sv|const char* s
0c395ea5 323Like C<sv_catpvn_mg>, but takes a C<NUL>-terminated literal string instead of a
9dcc53ea
Z
324string/length pair.
325
7b4e7469 326=for apidoc Am|void|sv_setpvs|SV* sv|const char* s
0c395ea5
KW
327Like C<sv_setpvn>, but takes a C<NUL>-terminated literal string instead of a
328string/length pair.
3fe05580 329
9dcc53ea 330=for apidoc Am|void|sv_setpvs_mg|SV* sv|const char* s
0c395ea5 331Like C<sv_setpvn_mg>, but takes a C<NUL>-terminated literal string instead of a
9dcc53ea
Z
332string/length pair.
333
334=for apidoc Am|SV *|sv_setref_pvs|const char* s
0c395ea5
KW
335Like C<sv_setref_pvn>, but takes a C<NUL>-terminated literal string instead of
336a string/length pair.
9dcc53ea 337
3fe05580
MHM
338=head1 Memory Management
339
340=for apidoc Ama|char*|savepvs|const char* s
0c395ea5 341Like C<savepvn>, but takes a C<NUL>-terminated literal string instead of a
30a15352 342string/length pair.
3fe05580 343
9dcc53ea
Z
344=for apidoc Ama|char*|savesharedpvs|const char* s
345A version of C<savepvs()> which allocates the duplicate string in memory
346which is shared between threads.
347
3fe05580
MHM
348=head1 GV Functions
349
350=for apidoc Am|HV*|gv_stashpvs|const char* name|I32 create
0c395ea5
KW
351Like C<gv_stashpvn>, but takes a C<NUL>-terminated literal string instead of a
352string/length pair.
3fe05580
MHM
353
354=head1 Hash Manipulation Functions
355
356=for apidoc Am|SV**|hv_fetchs|HV* tb|const char* key|I32 lval
0c395ea5
KW
357Like C<hv_fetch>, but takes a C<NUL>-terminated literal string instead of a
358string/length pair.
3fe05580
MHM
359
360=for apidoc Am|SV**|hv_stores|HV* tb|const char* key|NULLOK SV* val
0c395ea5
KW
361Like C<hv_store>, but takes a C<NUL>-terminated literal string instead of a
362string/length pair
3fe05580
MHM
363and omits the hash parameter.
364
510966aa
Z
365=head1 Lexer interface
366
367=for apidoc Amx|void|lex_stuff_pvs|const char *pv|U32 flags
368
0c395ea5
KW
369Like L</lex_stuff_pvn>, but takes a C<NUL>-terminated literal string instead of
370a string/length pair.
510966aa 371
3fe05580
MHM
372=cut
373*/
374
926bb54c
KW
375/* concatenating with "" ensures that only literal strings are accepted as
376 * argument */
26bfd924 377#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
2efa8cc7 378
926bb54c
KW
379/* note that STR_WITH_LEN() can't be used as argument to macros or functions
380 * that under some configurations might be macros, which means that it requires
381 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
ba3a79e7
GA
382 */
383
384/* STR_WITH_LEN() shortcuts */
385#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
386#define newSVpvs_flags(str,flags) \
387 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 388#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
389#define sv_catpvs_flags(sv, str, flags) \
390 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
391#define sv_catpvs_nomg(sv, str) \
392 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
393#define sv_catpvs(sv, str) \
394 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
395#define sv_catpvs_mg(sv, str) \
396 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 397#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
398#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
399#define sv_setref_pvs(rv, classname, str) \
400 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 401#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
402#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
403#define gv_stashpvs(str, create) \
404 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
405#define gv_fetchpvs(namebeg, add, sv_type) \
406 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
407#define gv_fetchpvn(namebeg, len, add, sv_type) \
408 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
409#define sv_catxmlpvs(dsv, str, utf8) \
410 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 411
ba3a79e7 412
510966aa
Z
413#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
414
b96d8cd9
NC
415#define get_cvs(str, flags) \
416 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 417
954c1994 418/*
ccfc67b7
JH
419=head1 Miscellaneous Functions
420
954c1994 421=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
422Test two C<NUL>-terminated strings to see if they are different. Returns true
423or false.
954c1994
GS
424
425=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
426Test two C<NUL>-terminated strings to see if they are equal. Returns true or
427false.
954c1994
GS
428
429=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
430Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
431second, C<s2>. Returns true or false.
954c1994
GS
432
433=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
434Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
435equal to the second, C<s2>. Returns true or false.
954c1994
GS
436
437=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
438Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
439the second, C<s2>. Returns true or false.
954c1994
GS
440
441=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
442Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
443or equal to the second, C<s2>. Returns true or false.
954c1994
GS
444
445=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
446Test two C<NUL>-terminated strings to see if they are different. The C<len>
447parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
448wrapper for C<strncmp>).
449
450=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
451Test two C<NUL>-terminated strings to see if they are equal. The C<len>
452parameter indicates the number of bytes to compare. Returns true or false. (A
453wrapper for C<strncmp>).
954c1994 454
bd18bd40
KW
455=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
456Test two buffers (which may contain embedded C<NUL> characters, to see if they
457are equal. The C<len> parameter indicates the number of bytes to compare.
458Returns zero if equal, or non-zero if non-equal.
459
460=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
461Test two buffers (which may contain embedded C<NUL> characters, to see if they
462are not equal. The C<len> parameter indicates the number of bytes to compare.
463Returns zero if non-equal, or non-zero if equal.
464
954c1994
GS
465=cut
466*/
467
62946e08 468
75400963
KW
469#define strNE(s1,s2) (strcmp(s1,s2) != 0)
470#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
471#define strLT(s1,s2) (strcmp(s1,s2) < 0)
472#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
473#define strGT(s1,s2) (strcmp(s1,s2) > 0)
474#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 475
75400963
KW
476#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
477#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 478
75400963
KW
479#define memNE(s1,s2,l) (memcmp(s1,s2,l) != 0)
480#define memEQ(s1,s2,l) (memcmp(s1,s2,l) == 0)
36477c24 481
085b7534 482/* memEQ and memNE where second comparand is a string constant */
568a785a 483#define memEQs(s1, l, s2) \
777fa2cb 484 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 485#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 486
fdbb9a7c
KW
487/* Keep these private until we decide it was a good idea */
488#if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
489
490#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
491
bdb7e3f0
KW
492#define memBEGINs(s1, l, s2) \
493 ( (l) >= sizeof(s2) - 1 \
494 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158
KW
495#define memBEGINPs(s1, l, s2) \
496 ( (l) > sizeof(s2) - 1 \
497 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0
KW
498#define memENDs(s1, l, s2) \
499 ( (l) >= sizeof(s2) - 1 \
500 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424
KW
501#define memENDPs(s1, l, s2) \
502 ( (l) > sizeof(s2) \
503 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
fdbb9a7c 504#endif /* End of making macros private */
bdb7e3f0 505
062b6850
KW
506#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
507#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
508#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
509#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
510
bbce6d69
PP
511/*
512 * Character classes.
513 *
514 * Unfortunately, the introduction of locales means that we
515 * can't trust isupper(), etc. to tell the truth. And when
516 * it comes to /\w+/ with tainting enabled, we *must* be able
517 * to trust our character classes.
518 *
519 * Therefore, the default tests in the text of Perl will be
520 * independent of locale. Any code that wants to depend on
521 * the current locale will use the tests that begin with "lc".
522 */
523
2304df62
AD
524#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
525# ifndef CTYPE256
526# define CTYPE256
527# endif
528#endif
529
954c1994 530/*
ccfc67b7 531
dcccc8ff 532=head1 Character classification
243effed
KW
533This section is about functions (really macros) that classify characters
534into types, such as punctuation versus alphabetic, etc. Most of these are
535analogous to regular expression character classes. (See
536L<perlrecharclass/POSIX Character Classes>.) There are several variants for
537each class. (Not all macros have all variants; each item below lists the
538ones valid for it.) None are affected by C<use bytes>, and only the ones
539with C<LC> in the name are affected by the current locale.
540
541The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
542C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
543represented by that octet is (or on non-ASCII platforms, corresponds to) an
544ASCII character in the named class based on platform, Unicode, and Perl rules.
545If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 546
c98722a4 547Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
548with no suffix C<"_A">. This variant is used to emphasize by its name that
549only ASCII-range characters can return TRUE.
4b9734bf 550
d60679e1 551Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
552onto the platform. That is, the code points that are ASCII are unaffected,
553since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
554as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
555true when called with the code point 0xDF, which is a word character in both
4650c663 556ASCII and EBCDIC (though it represents different characters in each).
243effed 557
c98722a4 558Variant C<isI<FOO>_uvchr> is like the C<isI<FOO>_L1> variant, but accepts any UV code
243effed
KW
559point as input. If the code point is larger than 255, Unicode rules are used
560to determine if it is in the character class. For example,
d0da05db 561C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 562WITH MACRON in Unicode, and is a word character.
243effed 563
c98722a4 564Variant C<isI<FOO>_utf8_safe> is like C<isI<FOO>_uvchr>, but is used for UTF-8
da8c1a98
KW
565encoded strings. Each call classifies one character, even if the string
566contains many. This variant takes two parameters. The first, C<p>, is a
567pointer to the first byte of the character to be classified. (Recall that it
568may take more than one byte to represent a character in UTF-8 strings.) The
569second parameter, C<e>, points to anywhere in the string beyond the first
570character, up to one byte past the end of the entire string. The suffix
571C<_safe> in the function's name indicates that it will not attempt to read
572beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
573is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
574character is malformed in some way, the program may croak, or the function may
575return FALSE, at the discretion of the implementation, and subject to change in
576future releases.
577
c98722a4 578Variant C<isI<FOO>_utf8> is like C<isI<FOO>_utf8_safe>, but takes just a single
da8c1a98 579parameter, C<p>, which has the same meaning as the corresponding parameter does
c98722a4 580in C<isI<FOO>_utf8_safe>. The function therefore can't check if it is reading
34aeb2e9 581beyond the end of the string. Starting in Perl v5.30, it will take a second
c98722a4 582parameter, becoming a synonym for C<isI<FOO>_utf8_safe>. At that time every
34aeb2e9 583program that uses it will have to be changed to successfully compile. In the
c98722a4 584meantime, the first runtime call to C<isI<FOO>_utf8> from each call point in the
34aeb2e9 585program will raise a deprecation warning, enabled by default. You can convert
c98722a4 586your program now to use C<isI<FOO>_utf8_safe>, and avoid the warnings, and get an
34aeb2e9
KW
587extra measure of protection, or you can wait until v5.30, when you'll be forced
588to add the C<e> parameter.
243effed 589
c98722a4 590Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants, but the
31f05a37
KW
591result is based on the current locale, which is what C<LC> in the name stands
592for. If Perl can determine that the current locale is a UTF-8 locale, it uses
593the published Unicode rules; otherwise, it uses the C library function that
594gives the named classification. For example, C<isDIGIT_LC()> when not in a
595UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
596returned if the input won't fit into an octet. On some platforms where the C
597library function is known to be defective, Perl changes its result to follow
598the POSIX standard's rules.
243effed 599
c98722a4
KW
600Variant C<isI<FOO>_LC_uvchr> is like C<isI<FOO>_LC>, but is defined on any UV. It
601returns the same as C<isI<FOO>_LC> for input code points less than 256, and
243effed
KW
602returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
603
c98722a4 604Variant C<isI<FOO>_LC_utf8_safe> is like C<isI<FOO>_LC_uvchr>, but is used for UTF-8
da8c1a98
KW
605encoded strings. Each call classifies one character, even if the string
606contains many. This variant takes two parameters. The first, C<p>, is a
607pointer to the first byte of the character to be classified. (Recall that it
608may take more than one byte to represent a character in UTF-8 strings.) The
609second parameter, C<e>, points to anywhere in the string beyond the first
610character, up to one byte past the end of the entire string. The suffix
611C<_safe> in the function's name indicates that it will not attempt to read
612beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
613is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
614character is malformed in some way, the program may croak, or the function may
615return FALSE, at the discretion of the implementation, and subject to change in
616future releases.
617
c98722a4 618Variant C<isI<FOO>_LC_utf8> is like C<isI<FOO>_LC_utf8_safe>, but takes just a single
da8c1a98 619parameter, C<p>, which has the same meaning as the corresponding parameter does
c98722a4 620in C<isI<FOO>_LC_utf8_safe>. The function therefore can't check if it is reading
34aeb2e9 621beyond the end of the string. Starting in Perl v5.30, it will take a second
c98722a4 622parameter, becoming a synonym for C<isI<FOO>_LC_utf8_safe>. At that time every
34aeb2e9 623program that uses it will have to be changed to successfully compile. In the
c98722a4 624meantime, the first runtime call to C<isI<FOO>_LC_utf8> from each call point in
34aeb2e9 625the program will raise a deprecation warning, enabled by default. You can
c98722a4 626convert your program now to use C<isI<FOO>_LC_utf8_safe>, and avoid the warnings,
34aeb2e9
KW
627and get an extra measure of protection, or you can wait until v5.30, when
628you'll be forced to add the C<e> parameter.
ccfc67b7 629
954c1994 630=for apidoc Am|bool|isALPHA|char ch
2787a470 631Returns a boolean indicating whether the specified character is an
6aff1f14 632alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
633See the L<top of this section|/Character classification> for an explanation of
634variants
da8c1a98
KW
635C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8_safe>,
636C<isALPHA_LC>, C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8_safe>.
8a58bdcf 637
15861f94
KW
638=for apidoc Am|bool|isALPHANUMERIC|char ch
639Returns a boolean indicating whether the specified character is a either an
6aff1f14 640alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
641See the L<top of this section|/Character classification> for an explanation of
642variants
d0da05db 643C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
da8c1a98
KW
644C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>,
645and C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 646
8a58bdcf
KW
647=for apidoc Am|bool|isASCII|char ch
648Returns a boolean indicating whether the specified character is one of the 128
243effed 649characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 650On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
651character corresponds to an ASCII character. Variants C<isASCII_A()> and
652C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
653See the L<top of this section|/Character classification> for an explanation of
654variants
da8c1a98
KW
655C<isASCII_uvchr>, C<isASCII_utf8_safe>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
656C<isASCII_LC_utf8_safe>. Note, however, that some platforms do not have the C
243effed
KW
657library routine C<isascii()>. In these cases, the variants whose names contain
658C<LC> are the same as the corresponding ones without.
659
d98532ea
KW
660Also note, that because all ASCII characters are UTF-8 invariant (meaning they
661have the exact same representation (always a single byte) whether encoded in
662UTF-8 or not), C<isASCII> will give the correct results when called with any
da8c1a98
KW
663byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8_safe>
664will work properly on any string encoded or not in UTF-8.
d98532ea 665
243effed
KW
666=for apidoc Am|bool|isBLANK|char ch
667Returns a boolean indicating whether the specified character is a
6aff1f14 668character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
669See the L<top of this section|/Character classification> for an explanation of
670variants
da8c1a98
KW
671C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8_safe>,
672C<isBLANK_LC>, C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8_safe>. Note,
673however, that some platforms do not have the C library routine
674C<isblank()>. In these cases, the variants whose names contain C<LC> are
675the same as the corresponding ones without.
243effed
KW
676
677=for apidoc Am|bool|isCNTRL|char ch
678Returns a boolean indicating whether the specified character is a
6aff1f14 679control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
680See the L<top of this section|/Character classification> for an explanation of
681variants
da8c1a98
KW
682C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8_safe>,
683C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8_safe> On EBCDIC
684platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 685
954c1994 686=for apidoc Am|bool|isDIGIT|char ch
2787a470 687Returns a boolean indicating whether the specified character is a
6aff1f14 688digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 689Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
690See the L<top of this section|/Character classification> for an explanation of
691variants
da8c1a98
KW
692C<isDIGIT_uvchr>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
693C<isDIGIT_LC_utf8_safe>.
243effed
KW
694
695=for apidoc Am|bool|isGRAPH|char ch
696Returns a boolean indicating whether the specified character is a
6aff1f14 697graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 698See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
699variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8_safe>,
700C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8_safe>.
954c1994 701
0c82b6df 702=for apidoc Am|bool|isLOWER|char ch
2787a470 703Returns a boolean indicating whether the specified character is a
6aff1f14 704lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
705See the L<top of this section|/Character classification> for an explanation of
706variants
da8c1a98
KW
707C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8_safe>,
708C<isLOWER_LC>, C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8_safe>.
0c82b6df 709
c99e91e9 710=for apidoc Am|bool|isOCTAL|char ch
2787a470 711Returns a boolean indicating whether the specified character is an
6aff1f14 712octal digit, [0-7].
243effed
KW
713The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
714C<isOCTAL>.
715
716=for apidoc Am|bool|isPUNCT|char ch
717Returns a boolean indicating whether the specified character is a
6aff1f14
KW
718punctuation character, analogous to C<m/[[:punct:]]/>.
719Note that the definition of what is punctuation isn't as
243effed
KW
720straightforward as one might desire. See L<perlrecharclass/POSIX Character
721Classes> for details.
dcccc8ff 722See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
723variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8_safe>,
724C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8_safe>.
c99e91e9 725
0c82b6df 726=for apidoc Am|bool|isSPACE|char ch
2787a470 727Returns a boolean indicating whether the specified character is a
6aff1f14 728whitespace character. This is analogous
398d098a 729to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 730this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
731locale forms of this macro (the ones with C<LC> in their names) matched
732precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
733in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
734(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
735See the L<top of this section|/Character classification> for an explanation of
736variants
da8c1a98
KW
737C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8_safe>,
738C<isSPACE_LC>, C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8_safe>.
0c82b6df 739
398d098a
KW
740=for apidoc Am|bool|isPSXSPC|char ch
741(short for Posix Space)
779cf272
KW
742Starting in 5.18, this is identical in all its forms to the
743corresponding C<isSPACE()> macros.
398d098a
KW
744The locale forms of this macro are identical to their corresponding
745C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
746non-locale forms differ from their C<isSPACE()> forms only in that the
747C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
748Otherwise they are identical. Thus this macro is analogous to what
749C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 750See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
751variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8_safe>,
752C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 753
954c1994 754=for apidoc Am|bool|isUPPER|char ch
2787a470 755Returns a boolean indicating whether the specified character is an
6aff1f14 756uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 757See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
758variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8_safe>,
759C<isUPPER_LC>, C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8_safe>.
954c1994 760
243effed 761=for apidoc Am|bool|isPRINT|char ch
8eea39dd 762Returns a boolean indicating whether the specified character is a
6aff1f14 763printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
764See the L<top of this section|/Character classification> for an explanation of
765variants
da8c1a98
KW
766C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8_safe>,
767C<isPRINT_LC>, C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8_safe>.
243effed
KW
768
769=for apidoc Am|bool|isWORDCHAR|char ch
770Returns a boolean indicating whether the specified character is a character
771that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
772in a regular expression. A word character is an alphabetic character, a
773decimal digit, a connecting punctuation character (such as an underscore), or
774a "mark" character that attaches to one of those (like some sort of accent).
775C<isALNUM()> is a synonym provided for backward compatibility, even though a
776word character includes more than the standard C language meaning of
777alphanumeric.
dcccc8ff 778See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
779variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and
780C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and
781C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
782include the platform's native underscore.
8a58bdcf
KW
783
784=for apidoc Am|bool|isXDIGIT|char ch
785Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
786digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
787and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
788See the L<top of this section|/Character classification> for an explanation of
789variants
da8c1a98
KW
790C<isXDIGIT_uvchr>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>,
791and C<isXDIGIT_LC_utf8_safe>.
243effed 792
3c3ecf18
KW
793=for apidoc Am|bool|isIDFIRST|char ch
794Returns a boolean indicating whether the specified character can be the first
795character of an identifier. This is very close to, but not quite the same as
796the official Unicode property C<XID_Start>. The difference is that this
797returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
798See the L<top of this section|/Character classification> for an explanation of
799variants
da8c1a98
KW
800C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8_safe>,
801C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18
KW
802
803=for apidoc Am|bool|isIDCONT|char ch
804Returns a boolean indicating whether the specified character can be the
805second or succeeding character of an identifier. This is very close to, but
806not quite the same as the official Unicode property C<XID_Continue>. The
807difference is that this returns true only if the input character also matches
dcccc8ff
KW
808L</isWORDCHAR>. See the L<top of this section|/Character classification> for
809an
d0da05db 810explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
da8c1a98
KW
811C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
812C<isIDCONT_LC_utf8_safe>.
3c3ecf18 813
243effed 814=head1 Miscellaneous Functions
8eea39dd 815
95a59cab 816=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 817Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
818Behaviour is only well defined when isXDIGIT(*str) is true.
819
e7c1e6c1 820=head1 Character case changing
21da7284
KW
821Perl uses "full" Unicode case mappings. This means that converting a single
822character to another case may result in a sequence of more than one character.
823For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
824character sequence C<SS>. This presents some complications The lowercase of
825all characters in the range 0..255 is a single character, and thus
826C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
827return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
828an API that does allow every possible legal result to be returned.) Likewise
829no other function that is crippled by not being able to give the correct
830results for the full range of possible inputs has been implemented here.
e7c1e6c1 831
1f607577
KW
832=for apidoc Am|U8|toUPPER|U8 ch
833Converts the specified character to uppercase. If the input is anything but an
834ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 835C<toUPPER_A> is equivalent.
954c1994 836
d0da05db
KW
837=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
838Converts the code point C<cp> to its uppercase version, and
839stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
840point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
841that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
842bytes since the uppercase version may be longer than the original character.
843
844The first code point of the uppercased version is returned
21da7284
KW
845(but note, as explained at L<the top of this section|/Character case
846changing>, that there may be more.)
1f607577 847
a239b1e2
KW
848=for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
849Converts the first UTF-8 encoded character in the sequence starting at C<p> and
850extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
851stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
852that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
853bytes since the uppercase version may be longer than the original character.
854
855The first code point of the uppercased version is returned
21da7284
KW
856(but note, as explained at L<the top of this section|/Character case
857changing>, that there may be more).
1f607577 858
a239b1e2
KW
859The suffix C<_safe> in the function's name indicates that it will not attempt
860to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
861true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
862input character is malformed in some way, the program may croak, or the
863function may return the REPLACEMENT CHARACTER, at the discretion of the
864implementation, and subject to change in future releases.
865
866=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
867This is like C<L</toUPPER_utf8_safe>>, but doesn't have the C<e>
868parameter The function therefore can't check if it is reading
607313a1
KW
869beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
870parameter, becoming a synonym for C<toUPPER_utf8_safe>. At that time every
871program that uses it will have to be changed to successfully compile. In the
872meantime, the first runtime call to C<toUPPER_utf8> from each call point in the
873program will raise a deprecation warning, enabled by default. You can convert
874your program now to use C<toUPPER_utf8_safe>, and avoid the warnings, and get an
875extra measure of protection, or you can wait until v5.30, when you'll be forced
876to add the C<e> parameter.
1f607577 877
25200305
KW
878=for apidoc Am|U8|toFOLD|U8 ch
879Converts the specified character to foldcase. If the input is anything but an
880ASCII uppercase character, that input character itself is returned. Variant
881C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 882Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 883
d0da05db
KW
884=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
885Converts the code point C<cp> to its foldcase version, and
886stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
887point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
888that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
889bytes since the foldcase version may be longer than the original character.
890
891The first code point of the foldcased version is returned
21da7284
KW
892(but note, as explained at L<the top of this section|/Character case
893changing>, that there may be more).
1f607577 894
a239b1e2
KW
895=for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
896Converts the first UTF-8 encoded character in the sequence starting at C<p> and
897extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
898stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
899that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
900bytes since the foldcase version may be longer than the original character.
901
902The first code point of the foldcased version is returned
21da7284
KW
903(but note, as explained at L<the top of this section|/Character case
904changing>, that there may be more).
1f607577 905
a239b1e2
KW
906The suffix C<_safe> in the function's name indicates that it will not attempt
907to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
908true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
909input character is malformed in some way, the program may croak, or the
910function may return the REPLACEMENT CHARACTER, at the discretion of the
911implementation, and subject to change in future releases.
912
913=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
914This is like C<L</toFOLD_utf8_safe>>, but doesn't have the C<e>
915parameter The function therefore can't check if it is reading
607313a1
KW
916beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
917parameter, becoming a synonym for C<toFOLD_utf8_safe>. At that time every
918program that uses it will have to be changed to successfully compile. In the
919meantime, the first runtime call to C<toFOLD_utf8> from each call point in the
920program will raise a deprecation warning, enabled by default. You can convert
921your program now to use C<toFOLD_utf8_safe>, and avoid the warnings, and get an
922extra measure of protection, or you can wait until v5.30, when you'll be forced
923to add the C<e> parameter.
1f607577
KW
924
925=for apidoc Am|U8|toLOWER|U8 ch
926Converts the specified character to lowercase. If the input is anything but an
927ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 928C<toLOWER_A> is equivalent.
954c1994 929
1f607577 930=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
931Converts the specified Latin1 character to lowercase. The results are
932undefined if the input doesn't fit in a byte.
1f607577
KW
933
934=for apidoc Am|U8|toLOWER_LC|U8 ch
935Converts the specified character to lowercase using the current locale's rules,
936if possible; otherwise returns the input character itself.
937
d0da05db
KW
938=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
939Converts the code point C<cp> to its lowercase version, and
940stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
941point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
942that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
943bytes since the lowercase version may be longer than the original character.
944
945The first code point of the lowercased version is returned
21da7284
KW
946(but note, as explained at L<the top of this section|/Character case
947changing>, that there may be more).
1f607577 948
a239b1e2
KW
949
950=for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
951Converts the first UTF-8 encoded character in the sequence starting at C<p> and
952extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
953stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
954that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
955bytes since the lowercase version may be longer than the original character.
956
957The first code point of the lowercased version is returned
21da7284
KW
958(but note, as explained at L<the top of this section|/Character case
959changing>, that there may be more).
1f607577 960
a239b1e2
KW
961The suffix C<_safe> in the function's name indicates that it will not attempt
962to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
963true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
964input character is malformed in some way, the program may croak, or the
965function may return the REPLACEMENT CHARACTER, at the discretion of the
966implementation, and subject to change in future releases.
967
968=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
969This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
970parameter The function therefore can't check if it is reading
607313a1
KW
971beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
972parameter, becoming a synonym for C<toLOWER_utf8_safe>. At that time every
973program that uses it will have to be changed to successfully compile. In the
974meantime, the first runtime call to C<toLOWER_utf8> from each call point in the
975program will raise a deprecation warning, enabled by default. You can convert
976your program now to use C<toLOWER_utf8_safe>, and avoid the warnings, and get an
977extra measure of protection, or you can wait until v5.30, when you'll be forced
978to add the C<e> parameter.
1f607577 979
25200305
KW
980=for apidoc Am|U8|toTITLE|U8 ch
981Converts the specified character to titlecase. If the input is anything but an
982ASCII lowercase character, that input character itself is returned. Variant
b7d90381 983C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 984range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 985not a concept used in locale handling, so there is no functionality for that.)
25200305 986
d0da05db
KW
987=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
988Converts the code point C<cp> to its titlecase version, and
989stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
990point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
991that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
992bytes since the titlecase version may be longer than the original character.
993
994The first code point of the titlecased version is returned
21da7284
KW
995(but note, as explained at L<the top of this section|/Character case
996changing>, that there may be more).
1f607577 997
a239b1e2
KW
998=for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
999Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1000extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
1001stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1002that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1003bytes since the titlecase version may be longer than the original character.
1004
1005The first code point of the titlecased version is returned
21da7284
KW
1006(but note, as explained at L<the top of this section|/Character case
1007changing>, that there may be more).
1f607577 1008
a239b1e2
KW
1009The suffix C<_safe> in the function's name indicates that it will not attempt
1010to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1011true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1012input character is malformed in some way, the program may croak, or the
1013function may return the REPLACEMENT CHARACTER, at the discretion of the
1014implementation, and subject to change in future releases.
1015
1016=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
1017This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1018parameter The function therefore can't check if it is reading
607313a1
KW
1019beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
1020parameter, becoming a synonym for C<toTITLE_utf8_safe>. At that time every
1021program that uses it will have to be changed to successfully compile. In the
1022meantime, the first runtime call to C<toTITLE_utf8> from each call point in the
1023program will raise a deprecation warning, enabled by default. You can convert
1024your program now to use C<toTITLE_utf8_safe>, and avoid the warnings, and get an
1025extra measure of protection, or you can wait until v5.30, when you'll be forced
1026to add the C<e> parameter.
1f607577 1027
954c1994 1028=cut
353c9b6f 1029
d0da05db 1030XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1031really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1032and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1033
8a58bdcf 1034Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1035patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1036
954c1994
GS
1037*/
1038
7c062697
KW
1039/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
1040 * is known only in the perl core, and this macro can be called from outside
1041 * that */
1042#ifdef HAS_QUAD
1043# define WIDEST_UTYPE U64TYPE
1044#else
1045# define WIDEST_UTYPE U32
1046#endif
1047
3912bc88
KW
1048/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1049 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1050 * bits of information are lost even on a 64-bit machine, but to get the
1051 * compiler to optimize it out if possible. This is because Configure makes
1052 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1053 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1054 *
1055 * For Coverity, be always true, because otherwise Coverity thinks
1056 * it finds several expressions that are always true, independent
1057 * of operands. Well, they are, but that is kind of the point.
220c71bf 1058 */
7e75d1a1 1059#ifndef __COVERITY__
6c5b02ac
KW
1060 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1061 * pointer) */
1062#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1063 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1064#else
1065#define FITS_IN_8_BITS(c) (1)
1066#endif
cf301eb7 1067
41f43cc2 1068#ifdef EBCDIC
b6340bd0 1069# ifndef _ALL_SOURCE
0852beac
KW
1070 /* The native libc isascii() et.al. functions return the wrong results
1071 * on at least z/OS unless this is defined. */
b6340bd0
KW
1072# error _ALL_SOURCE should probably be defined
1073# endif
41f43cc2 1074#else
0852beac
KW
1075 /* There is a simple definition of ASCII for ASCII platforms. But the
1076 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1077 * other macros below.
3f3c579d
KW
1078 *
1079 * The cast here is used instead of '(c) >= 0', because some compilers emit
1080 * a warning that that test is always true when the parameter is an
1081 * unsigned type. khw supposes that it could be written as
1082 * && ((c) == '\0' || (c) > 0)
1083 * to avoid the message, but the cast will likely avoid extra branches even
1084 * with stupid compilers.
1085 *
1086 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1087 * a pointer) */
9c903d59 1088# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1089#endif
1090
38694112
KW
1091/* Take the eight possible bit patterns of the lower 3 bits and you get the
1092 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1093 * can be ignored. If the rest match '0', we have an octal */
1094#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1095
9fb1bf9d 1096#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1097
a500dc72
KW
1098/* Character class numbers. For internal core Perl use only. The ones less
1099 * than 32 are used in PL_charclass[] and the ones up through the one that
1100 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1101 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1102 * their actual definitions are here. If that file has a name not used here,
1103 * it won't compile.
1709d539
KW
1104 *
1105 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
1106 * frequency of their use. This gives a slight edge to exiting a loop earlier
1107 * (in reginclass() in regexec.c) */
1709d539
KW
1108# define _CC_WORDCHAR 0 /* \w and [:word:] */
1109# define _CC_DIGIT 1 /* \d and [:digit:] */
1110# define _CC_ALPHA 2 /* [:alpha:] */
1111# define _CC_LOWER 3 /* [:lower:] */
1112# define _CC_UPPER 4 /* [:upper:] */
1113# define _CC_PUNCT 5 /* [:punct:] */
1114# define _CC_PRINT 6 /* [:print:] */
15861f94 1115# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1116# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1117# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1709d539 1118
b0d691b2 1119#define _FIRST_NON_SWASH_CC 10
1709d539
KW
1120/* The character classes above are implemented with swashes. The second group
1121 * (just below) contains the ones implemented without. These are also sorted
1122 * in rough order of the frequency of their use, except that \v should be last,
1123 * as it isn't a real Posix character class, and some (small) inefficiencies in
1124 * regular expression handling would be introduced by putting it in the middle
1125 * of those that are. Also, cntrl and ascii come after the others as it may be
1126 * useful to group these which have no members that match above Latin1, (or
1127 * above ASCII in the latter case) */
1128
779cf272 1129# define _CC_SPACE 10 /* \s, [:space:] */
34aeb2e9
KW
1130# define _CC_PSXSPC _CC_SPACE /* XXX Temporary, can be removed
1131 when the deprecated isFOO_utf8()
1132 functions are removed */
b0d691b2
KW
1133# define _CC_BLANK 11 /* [:blank:] */
1134# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1135# define _CC_CNTRL 13 /* [:cntrl:] */
1136# define _CC_ASCII 14 /* [:ascii:] */
1137# define _CC_VERTSPACE 15 /* \v */
1709d539 1138
a0947d7b
KW
1139# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1140
1709d539 1141/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1142 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1143# define _CC_IDFIRST 16
1144# define _CC_CHARNAME_CONT 17
1145# define _CC_NONLATIN1_FOLD 18
1146# define _CC_NONLATIN1_SIMPLE_FOLD 19
1147# define _CC_QUOTEMETA 20
1148# define _CC_NON_FINAL_FOLD 21
1149# define _CC_IS_IN_SOME_FOLD 22
1150# define _CC_MNEMONIC_CNTRL 23
073c22b3 1151
34aeb2e9
KW
1152# define _CC_IDCONT 24 /* XXX Temporary, can be removed when the deprecated
1153 isFOO_utf8() functions are removed */
1154
073c22b3
KW
1155/* This next group is only used on EBCDIC platforms, so theoretically could be
1156 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
1157# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1158# define _CC_UTF8_IS_START 29
1159# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1160# define _CC_UTF8_IS_CONTINUATION 31
1161/* Unused: 24-27
f4cdb42c
KW
1162 * If more bits are needed, one could add a second word for non-64bit
1163 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1164 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1165 * is used only for optimization (as of this writing), and differs in the
1166 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1167 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1168 * would just cause /i regexes which match them to run less efficiently.
1169 * Similarly the EBCDIC-only bits are used just for speed, and could be
1170 * replaced by other means */
96ac0975 1171
3a371f2f
KW
1172#if defined(PERL_CORE) || defined(PERL_EXT)
1173/* An enum version of the character class numbers, to help compilers
1174 * optimize */
1175typedef enum {
3a371f2f 1176 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1177 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1178 _CC_ENUM_ASCII = _CC_ASCII,
1179 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1180 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1181 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1182 _CC_ENUM_DIGIT = _CC_DIGIT,
1183 _CC_ENUM_GRAPH = _CC_GRAPH,
1184 _CC_ENUM_LOWER = _CC_LOWER,
1185 _CC_ENUM_PRINT = _CC_PRINT,
1186 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1187 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1188 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1189 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1190 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1191 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1192} _char_class_number;
1193#endif
1194
63c61c3f 1195#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1196#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1197
b7d90381
KW
1198#if defined(PERL_IN_UTF8_C) \
1199 || defined(PERL_IN_REGCOMP_C) \
1200 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1201# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1202 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1203 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1204 #error Need to adjust order of swash_property_names[]
1205# endif
1206
b44daeeb
KW
1207/* This is declared static in each of the few files that this is #defined for
1208 * to keep them from being publicly accessible. Hence there is a small amount
1209 * of wasted space */
1210
922e8cb4
KW
1211static const char* const swash_property_names[] = {
1212 "XPosixWord",
1213 "XPosixDigit",
1214 "XPosixAlpha",
1215 "XPosixLower",
1216 "XPosixUpper",
1217 "XPosixPunct",
1218 "XPosixPrint",
1219 "XPosixAlnum",
b0d691b2
KW
1220 "XPosixGraph",
1221 "Cased"
922e8cb4
KW
1222};
1223#endif
1224
6635f04f 1225START_EXTERN_C
96ac0975
NC
1226# ifdef DOINIT
1227EXTCONST U32 PL_charclass[] = {
1228# include "l1_char_class_tab.h"
1229};
1230
1231# else /* ! DOINIT */
1232EXTCONST U32 PL_charclass[];
1233# endif
6635f04f 1234END_EXTERN_C
96ac0975 1235
265c1f46 1236 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1237# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1238
1239 /* For internal core Perl use only: the base macro for defining macros like
1240 * isALPHA */
ff7ecfc3 1241# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1242 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1243
f4cdb42c
KW
1244 /* The mask for the _A versions of the macros; it just adds in the bit for
1245 * ASCII. */
1246# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1247
4650c663
KW
1248 /* For internal core Perl use only: the base macro for defining macros like
1249 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1250 * the ASCII bit are present */
b7d90381
KW
1251# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1252 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1253 == _CC_mask_A(classnum)))
f4cdb42c 1254
f4cdb42c 1255# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1256# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1257# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1258# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1259# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1260# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1261# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1262# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1263# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1264# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1265# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1266# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1267# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1268 */
d95f8b6a 1269# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1270# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1271# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1272# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1273
1274 /* continuation character for legal NAME in \N{NAME} */
1275# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1276
1277# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1278# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1279# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1280# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1281# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1282# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1283# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1284# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1285# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1286# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1287
0852beac
KW
1288# ifdef EBCDIC
1289# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1290# endif
1291
f12c0118
KW
1292 /* Participates in a single-character fold with a character above 255 */
1293# 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)))
1294
1295 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1296# 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 1297
4eeeb416 1298# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1299# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1300 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1301# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1302 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1303# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1304 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1305#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1306
1307 /* If we don't have perl.h, we are compiling a utility program. Below we
1308 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1309 * to it. Most are coded based on first principles. These are written to
74665a89 1310 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
0852beac
KW
1311# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1312# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1313# define isSPACE_A(c) (isBLANK_A(c) \
1314 || (c) == '\n' \
1315 || (c) == '\r' \
1316 || (c) == '\v' \
0852beac 1317 || (c) == '\f')
74665a89
KW
1318 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1319 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1320 * performance (if optimization isn't on), and allow the same code to be
1321 * used for both platform types */
1322# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1323 && ( (c) <= 'i' \
1324 || ((c) >= 'j' && (c) <= 'r') \
1325 || (c) >= 's'))
1326# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1327 && ( (c) <= 'I' \
1328 || ((c) >= 'J' && (c) <= 'R') \
1329 || (c) >= 'S'))
a4d7a999
KW
1330# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1331# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1332# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1333# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
74665a89
KW
1334# define isXDIGIT_A(c) (isDIGIT_A(c) \
1335 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1336 || ((c) <= 'F' && (c) >= 'A'))
74665a89
KW
1337# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1338 || (c) == '#' || (c) == '$' || (c) == '%' \
1339 || (c) == '&' || (c) == '\'' || (c) == '(' \
1340 || (c) == ')' || (c) == '*' || (c) == '+' \
1341 || (c) == ',' || (c) == '.' || (c) == '/' \
1342 || (c) == ':' || (c) == ';' || (c) == '<' \
1343 || (c) == '=' || (c) == '>' || (c) == '?' \
1344 || (c) == '@' || (c) == '[' || (c) == '\\' \
1345 || (c) == ']' || (c) == '^' || (c) == '_' \
1346 || (c) == '`' || (c) == '{' || (c) == '|' \
1347 || (c) == '}' || (c) == '~')
1348# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1349# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1350
0852beac 1351# ifdef EBCDIC
74665a89
KW
1352 /* The below is accurate for the 3 EBCDIC code pages traditionally
1353 * supported by perl. The only difference between them in the controls
1354 * is the position of \n, and that is represented symbolically below */
1355# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1356 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1357 || (c) == '\t' || (c) == '\v' \
1358 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1359 || (c) == 7 /* U+7F DEL */ \
1360 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1361 /* DLE, DC[1-3] */ \
1362 || (c) == 0x18 /* U+18 CAN */ \
1363 || (c) == 0x19 /* U+19 EOM */ \
1364 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1365 || (c) == 0x26 /* U+17 ETB */ \
1366 || (c) == 0x27 /* U+1B ESC */ \
1367 || (c) == 0x2D /* U+05 ENQ */ \
1368 || (c) == 0x2E /* U+06 ACK */ \
1369 || (c) == 0x32 /* U+16 SYN */ \
1370 || (c) == 0x37 /* U+04 EOT */ \
1371 || (c) == 0x3C /* U+14 DC4 */ \
1372 || (c) == 0x3D /* U+15 NAK */ \
1373 || (c) == 0x3F)/* U+1A SUB */
0852beac 1374# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1375# else /* isASCII is already defined for ASCII platforms, so can use that to
1376 define isCNTRL */
1377# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1378# endif
1379
3ffc8c70 1380 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1381 * during debugging, and it seems best to keep them. We may be called
1382 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1383 * do anything anyway, so make it not a problem */
1384# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1385# define NATIVE_TO_LATIN1(ch) (ch)
1386# endif
3ded5eb0
KW
1387# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1388# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1389# define isBLANK_L1(c) (isBLANK_A(c) \
1390 || (FITS_IN_8_BITS(c) \
1391 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1392# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1393# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1394# define isLOWER_L1(c) (isLOWER_A(c) \
1395 || (FITS_IN_8_BITS(c) \
ae683a5f 1396 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1397 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1398 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1399 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1400 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1401# define isPRINT_L1(c) (isPRINT_A(c) \
1402 || (FITS_IN_8_BITS(c) \
1403 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1404# define isPUNCT_L1(c) (isPUNCT_A(c) \
1405 || (FITS_IN_8_BITS(c) \
ae683a5f 1406 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1407 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1408 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1409 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1410 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1411 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1412 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1413# define isSPACE_L1(c) (isSPACE_A(c) \
1414 || (FITS_IN_8_BITS(c) \
ae683a5f 1415 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1416 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1417# define isUPPER_L1(c) (isUPPER_A(c) \
1418 || (FITS_IN_8_BITS(c) \
ae683a5f 1419 && ( NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
3ded5eb0
KW
1420 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1421 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1422# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1423# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1424# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1425 || isBLANK_L1(c) \
1426 || (c) == '-' \
1427 || (c) == '(' \
1428 || (c) == ')')
1429 /* The following are not fully accurate in the above-ASCII range. I (khw)
1430 * don't think it's necessary to be so for the purposes where this gets
1431 * compiled */
1432# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1433# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1434
1435 /* And these aren't accurate at all. They are useful only for above
1436 * Latin1, which utilities and bootstrapping don't deal with */
1437# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1438# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1439# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1440
1441 /* Many of the macros later in this file are defined in terms of these. By
1442 * implementing them with a function, which converts the class number into
1443 * a call to the desired macro, all of the later ones work. However, that
1444 * function won't be actually defined when building a utility program (no
1445 * perl.h), and so a compiler error will be generated if one is attempted
1446 * to be used. And the above-Latin1 code points require Unicode tables to
1447 * be present, something unlikely to be the case when bootstrapping */
1448# define _generic_isCC(c, classnum) \
1449 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1450# define _generic_isCC_A(c, classnum) \
1451 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1452#endif /* End of no perl.h H_PERL */
8a58bdcf 1453
e66b99e9
KW
1454#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1455#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1456#define isASCII_A(c) isASCII(c)
1457#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1458#define isBLANK(c) isBLANK_A(c)
1459#define isCNTRL(c) isCNTRL_A(c)
1460#define isDIGIT(c) isDIGIT_A(c)
1461#define isGRAPH(c) isGRAPH_A(c)
1462#define isIDFIRST(c) isIDFIRST_A(c)
1463#define isLOWER(c) isLOWER_A(c)
1464#define isPRINT(c) isPRINT_A(c)
779cf272 1465#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1466#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1467#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1468#define isPUNCT(c) isPUNCT_A(c)
1469#define isSPACE(c) isSPACE_A(c)
1470#define isUPPER(c) isUPPER_A(c)
1471#define isWORDCHAR(c) isWORDCHAR_A(c)
1472#define isXDIGIT(c) isXDIGIT_A(c)
1473
1474/* ASCII casing. These could also be written as
1475 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1476 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1477 which uses table lookup and mask instead of subtraction. (This would
1478 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1479#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1480#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1481
25200305
KW
1482/* In the ASCII range, these are equivalent to what they're here defined to be.
1483 * But by creating these definitions, other code doesn't have to be aware of
1484 * this detail */
1485#define toFOLD(c) toLOWER(c)
25200305
KW
1486#define toTITLE(c) toUPPER(c)
1487
c753c8d3
KW
1488#define toLOWER_A(c) toLOWER(c)
1489#define toUPPER_A(c) toUPPER(c)
25200305
KW
1490#define toFOLD_A(c) toFOLD(c)
1491#define toTITLE_A(c) toTITLE(c)
1a0901db 1492
4650c663 1493/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1494#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1495 ? (c) \
f4cd282c 1496 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1497#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1498
1a0901db 1499/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1500 * all mapped to one of them, and these need special handling; returns the
1501 * input itself if is out-of-range */
b2bf251f 1502#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1503 ? (c) \
f4cd282c 1504 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1505#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1506
beab9ebe
KW
1507/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1508
1509/* For internal core Perl use only: the base macro for defining macros like
1510 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1511 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1512 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1513 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1514 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1515 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1516 * code points under locale rules, use the macros based on _generic_LC_uvchr
1517 * instead of this. */
beab9ebe
KW
1518#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1519 (! FITS_IN_8_BITS(c) \
1520 ? 0 \
31f05a37
KW
1521 : IN_UTF8_CTYPE_LOCALE \
1522 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1523 : cBOOL(non_utf8))
1524
1525/* For internal core Perl use only: a helper macro for defining macros like
1526 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1527 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1528 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1529 * _generic_LC_base for more info */
1530#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1531 _generic_LC_base(c,utf8_locale_classnum, \
1532 non_utf8_func( (_LC_CAST) (c)))
1533
1534/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1535 * 'c' is the platform's native underscore character */
1536#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1537 _generic_LC_base(c, utf8_locale_classnum, \
1538 (non_utf8_func( (_LC_CAST) (c)) \
1539 || (char)(c) == '_'))
1540
1541/* These next three are also for internal core Perl use only: case-change
1542 * helper macros */
1543#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1544 ? (c) \
31f05a37
KW
1545 : (IN_UTF8_CTYPE_LOCALE) \
1546 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1547 : (cast)function((cast)(c)))
beab9ebe 1548
31f05a37
KW
1549/* Note that the result can be larger than a byte in a UTF-8 locale. It
1550 * returns a single value, so can't adequately return the upper case of LATIN
1551 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1552 * values "SS"); instead it asserts against that under DEBUGGING, and
1553 * otherwise returns its input */
beab9ebe
KW
1554#define _generic_toUPPER_LC(c, function, cast) \
1555 (! FITS_IN_8_BITS(c) \
1556 ? (c) \
31f05a37 1557 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1558 ? (cast)function((cast)(c)) \
31f05a37
KW
1559 : ((((U8)(c)) == MICRO_SIGN) \
1560 ? GREEK_CAPITAL_LETTER_MU \
1561 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1562 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1563 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1564 ? (__ASSERT_(0) (c)) \
1565 : PL_mod_latin1_uc[ (U8) (c) ])))))
1566
1567/* Note that the result can be larger than a byte in a UTF-8 locale. It
1568 * returns a single value, so can't adequately return the fold case of LATIN
1569 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1570 * values "ss"); instead it asserts against that under DEBUGGING, and
1571 * otherwise returns its input */
beab9ebe 1572#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1573 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1574 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1575 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1576 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1577 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1578
84061b6a 1579/* Use the libc versions for these if available. */
f05550c0 1580#if defined(HAS_ISASCII)
84061b6a
KW
1581# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1582#else
1583# define isASCII_LC(c) isASCII(c)
1584#endif
1585
f05550c0 1586#if defined(HAS_ISBLANK)
84061b6a 1587# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1588#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1589# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1590#endif
1591
f05550c0 1592#define _LC_CAST U8
bbce6d69 1593
f05550c0 1594#ifdef WIN32
375f5f06
KW
1595 /* The Windows functions don't bother to follow the POSIX standard, which
1596 * for example says that something can't both be a printable and a control.
1597 * But Windows treats the \t control as a printable, and does such things
1598 * as making superscripts into both digits and punctuation. This tames
1599 * these flaws by assuming that the definitions of both controls and space
1600 * are correct, and then making sure that other definitions don't have
1601 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1602 * Not all possible weirdnesses are checked for, just the ones that were
1603 * detected on actual Microsoft code pages */
1604
b7d90381
KW
1605# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1606# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1607
1608# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1609 && isALPHANUMERIC_LC(c))
1610# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1611 ! isPUNCT_LC(c))
1612# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1613 isALPHANUMERIC_LC(c))
1614# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1615# define isIDFIRST_LC(c) (((c) == '_') \
1616 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1617# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1618# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1619# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1620# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1621# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1622# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1623 && isALPHANUMERIC_LC(c))
f05550c0
BF
1624
1625# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1626# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1627# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1628
1629#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1630 /* For most other platforms */
beab9ebe 1631
f05550c0
BF
1632# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1633# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1634# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1635# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1636# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1637# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1638# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1639# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1640# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1641# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1642# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1643# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1644# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1645
1646
1647# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1648# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1649# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1650
1651#else /* The final fallback position */
1652
b7d90381
KW
1653# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1654# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1655# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1656# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1657# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1658# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1659# define isLOWER_LC(c) (isascii(c) && islower(c))
1660# define isPRINT_LC(c) (isascii(c) && isprint(c))
1661# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1662# define isSPACE_LC(c) (isascii(c) && isspace(c))
1663# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1664# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1665# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1666
1667# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1668# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1669# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1670
f05550c0 1671#endif
55204971 1672
eba68aa0
KW
1673#define isIDCONT(c) isWORDCHAR(c)
1674#define isIDCONT_A(c) isWORDCHAR_A(c)
1675#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1676#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1677#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1678
4650c663 1679/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1680 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1681 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1682 * classes where there is a macro or function 'above_latin1' that takes the
1683 * single argument 'c' and returns the desired value. These exist for those
1684 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1685 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1686 * for classes where that overhead is faster than a direct lookup.
d0da05db 1687 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1688 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1689 * have duplicate checks here, so could create versions of the macros that
1690 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1691
1692/* Note that all ignore 'use bytes' */
1e222e4f
KW
1693#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1694 ? _generic_isCC(c, classnum) \
cd500f2f 1695 : above_latin1(c))
1e222e4f
KW
1696#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
1697 ? _generic_isCC(c, classnum) \
922e8cb4 1698 : _is_uni_FOO(classnum, c))
d0da05db
KW
1699#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1700#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1701#define isASCII_uvchr(c) isASCII(c)
1702#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1703#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1704#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1705#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1706#define isIDCONT_uvchr(c) \
1707 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1708#define isIDFIRST_uvchr(c) \
1709 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
d0da05db
KW
1710#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1711#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1712
1713#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1714#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1715#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1716
1717#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1718#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1719#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1720#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1721
1722#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1723#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1724#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1725#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1726
1727/* For backwards compatibility, even though '_uni' should mean official Unicode
1728 * code points, in Perl it means native for those below 256 */
1729#define isALPHA_uni(c) isALPHA_uvchr(c)
1730#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1731#define isASCII_uni(c) isASCII_uvchr(c)
1732#define isBLANK_uni(c) isBLANK_uvchr(c)
1733#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1734#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1735#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1736#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1737#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1738#define isLOWER_uni(c) isLOWER_uvchr(c)
1739#define isPRINT_uni(c) isPRINT_uvchr(c)
1740#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1741#define isSPACE_uni(c) isSPACE_uvchr(c)
1742#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1743#define isUPPER_uni(c) isUPPER_uvchr(c)
1744#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1745#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1746#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1747#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1748#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1749#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1750#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1751
4650c663
KW
1752/* For internal core Perl use only: the base macros for defining macros like
1753 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1754 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1755 * simple class definitions; the other for more complex. These are like
d0da05db 1756 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1757#define _generic_LC_uvchr(latin1, above_latin1, c) \
1758 (c < 256 ? latin1(c) : above_latin1(c))
1759#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1760 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1761
1762#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1763#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1764 _CC_ALPHANUMERIC, c)
b7d90381
KW
1765#define isASCII_LC_uvchr(c) isASCII_LC(c)
1766#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1767 is_HORIZWS_cp_high, c)
feeab5a9 1768#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1769#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1770#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1771#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1772 _is_uni_perl_idcont, c)
b7d90381 1773#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1774 _is_uni_perl_idstart, c)
1775#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1776#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1777#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1778#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1779#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1780 is_XPERLSPACE_cp_high, c)
cd500f2f 1781#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1782#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1783 _CC_WORDCHAR, c)
b7d90381
KW
1784#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1785 is_XDIGIT_cp_high, c)
e712593e 1786
b7d90381 1787#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1788
4650c663
KW
1789/* For internal core Perl use only: the base macros for defining macros like
1790 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1791 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1792 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1793 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1794 * same representation whether utf8 or not. Note that it assumes that the utf8
1795 * has been validated, and ignores 'use bytes' */
34aeb2e9
KW
1796#define _base_generic_utf8(enum_name, name, p, use_locale ) \
1797 _is_utf8_FOO(CAT2(_CC_, enum_name), \
1798 (const U8 *) p, \
1799 "is" STRINGIFY(name) "_utf8", \
1800 "is" STRINGIFY(name) "_utf8_safe", \
1801 1, use_locale, __FILE__,__LINE__)
1802
1803#define _generic_utf8(name, p) _base_generic_utf8(name, name, p, 0)
da8c1a98
KW
1804
1805/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
1806 * they don't otherwise go out of their way to look for malformed UTF-8. If
1807 * they can return accurate results without knowing if the input is otherwise
1808 * malformed, they do so. For example isASCII is accurate in spite of any
1809 * non-length malformations because it looks only at a single byte. Likewise
1810 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
1811 * variant ones return FALSE. But, if the input has to be well-formed in order
1812 * for the results to be accurate, the macros will test and if malformed will
1813 * call a routine to die
1814 *
1815 * Except for toke.c, the macros do assume that e > p, asserting that on
1816 * DEBUGGING builds. Much code that calls these depends on this being true,
1817 * for other reasons. toke.c is treated specially as using the regular
1818 * assertion breaks it in many ways. All strings that these operate on there
1819 * are supposed to have an extra NUL character at the end, so that *e = \0. A
1820 * bunch of code in toke.c assumes that this is true, so the assertion allows
1821 * for that */
1822#ifdef PERL_IN_TOKE_C
1823# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
1824#else
1825# define _utf8_safe_assert(p,e) ((e) > (p))
1826#endif
1827
1828#define _generic_utf8_safe(classnum, p, e, above_latin1) \
1829 (__ASSERT_(_utf8_safe_assert(p, e)) \
1830 (UTF8_IS_INVARIANT(*(p))) \
1831 ? _generic_isCC(*(p), classnum) \
1832 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1833 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1834 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
1835 classnum) \
1836 : (_force_out_malformed_utf8_message( \
1837 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1838 : above_latin1))
b7d90381
KW
1839/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1840 * 'above_latin1' can be a macro */
da8c1a98
KW
1841#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
1842 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
1843#define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1844 _generic_utf8_safe(classnum, p, e, \
1845 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1846 ? (_force_out_malformed_utf8_message( \
1847 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1848 : above_latin1(p)))
4650c663 1849/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1850 * 'above_latin1' parameter */
da8c1a98
KW
1851#define _generic_swash_utf8_safe(classnum, p, e) \
1852_generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e))
922e8cb4 1853
cc8ab7c0 1854/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1855 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1856 * class is TRUE for. Hence it can skip the tests for this range.
1857 * 'above_latin1' should include its arguments */
da8c1a98
KW
1858#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
1859 (__ASSERT_(_utf8_safe_assert(p, e)) \
1860 (UTF8_IS_INVARIANT(*(p))) \
1861 ? _generic_isCC(*(p), classnum) \
1862 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1863 ? 0 /* Note that doesn't check validity for latin1 */ \
1864 : above_latin1)
1865
84238efa
KW
1866/* NOTE that some of these macros have very similar ones in regcharclass.h.
1867 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1868 * there, differing in name only by an underscore from the one here
1869 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1870 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1871 * points; the regcharclass.h ones are implemented as a series of
1872 * "if-else-if-else ..." */
1873
34aeb2e9
KW
1874#define isALPHA_utf8(p) _generic_utf8(ALPHA, p)
1875#define isALPHANUMERIC_utf8(p) _generic_utf8(ALPHANUMERIC, p)
1876#define isASCII_utf8(p) _generic_utf8(ASCII, p)
1877#define isBLANK_utf8(p) _generic_utf8(BLANK, p)
1878#define isCNTRL_utf8(p) _generic_utf8(CNTRL, p)
1879#define isDIGIT_utf8(p) _generic_utf8(DIGIT, p)
1880#define isGRAPH_utf8(p) _generic_utf8(GRAPH, p)
1881#define isIDCONT_utf8(p) _generic_utf8(IDCONT, p)
1882#define isIDFIRST_utf8(p) _generic_utf8(IDFIRST, p)
1883#define isLOWER_utf8(p) _generic_utf8(LOWER, p)
1884#define isPRINT_utf8(p) _generic_utf8(PRINT, p)
1885#define isPSXSPC_utf8(p) _generic_utf8(PSXSPC, p)
1886#define isPUNCT_utf8(p) _generic_utf8(PUNCT, p)
1887#define isSPACE_utf8(p) _generic_utf8(SPACE, p)
1888#define isUPPER_utf8(p) _generic_utf8(UPPER, p)
1889#define isVERTWS_utf8(p) _generic_utf8(VERTSPACE, p)
1890#define isWORDCHAR_utf8(p) _generic_utf8(WORDCHAR, p)
1891#define isXDIGIT_utf8(p) _generic_utf8(XDIGIT, p)
e8fa43e2 1892
da8c1a98
KW
1893#define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e)
1894#define isALPHANUMERIC_utf8_safe(p, e) \
1895 _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e)
1896#define isASCII_utf8_safe(p, e) \
1897 /* Because ASCII is invariant under utf8, the non-utf8 macro \
1898 * works */ \
1899 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
1900#define isBLANK_utf8_safe(p, e) \
1901 _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
1902
e8fa43e2
KW
1903#ifdef EBCDIC
1904 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1905 * more efficient macro instead of the more general one */
da8c1a98 1906# define isCNTRL_utf8_safe(p, e) \
56d02b8c 1907 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 1908#else
da8c1a98 1909# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
1910#endif
1911
da8c1a98
KW
1912#define isDIGIT_utf8_safe(p, e) \
1913 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
1914 _is_utf8_FOO_with_len(_CC_DIGIT, p, e))
1915#define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e)
1916#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
1917 _is_utf8_perl_idcont_with_len, p, e)
e5dcd934 1918
c11ff943
KW
1919/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1920 * IDFIRST is an alnum. See
c8362b00 1921 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1922 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1923 * This used to be not the XID version, but we decided to go with the more
1924 * modern Unicode definition */
da8c1a98
KW
1925#define isIDFIRST_utf8_safe(p, e) \
1926 _generic_func_utf8_safe(_CC_IDFIRST, \
1927 _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e))
1928
1929#define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e)
1930#define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e)
1931#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
1932#define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e)
1933#define isSPACE_utf8_safe(p, e) \
1934 _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
1935#define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e)
1936#define isVERTWS_utf8_safe(p, e) \
1937 _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
1938#define isWORDCHAR_utf8_safe(p, e) \
1939 _generic_swash_utf8_safe(_CC_WORDCHAR, p, e)
1940#define isXDIGIT_utf8_safe(p, e) \
1941 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
1942 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1943 ? (_force_out_malformed_utf8_message( \
1944 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1945 : is_XDIGIT_high(p)))
a0ed51b3 1946
7d2b61dd 1947#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1948#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1949#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1950#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1951
567b353c 1952/* For internal core use only, subject to change */
607313a1
KW
1953#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f, "", 0)
1954#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f, "", 0)
1955#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f, "", 0)
1956#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f, "", 0)
a1a5ec35
KW
1957
1958#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
1959#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
1960#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
1961#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 1962
4650c663
KW
1963/* For internal core Perl use only: the base macros for defining macros like
1964 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1965 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1966 * 'macro' parameter */
34aeb2e9
KW
1967#define _generic_LC_utf8(name, p) _base_generic_utf8(name, name, p, 1)
1968
1969#define isALPHA_LC_utf8(p) _generic_LC_utf8(ALPHA, p)
1970#define isALPHANUMERIC_LC_utf8(p) _generic_LC_utf8(ALPHANUMERIC, p)
1971#define isASCII_LC_utf8(p) _generic_LC_utf8(ASCII, p)
1972#define isBLANK_LC_utf8(p) _generic_LC_utf8(BLANK, p)
1973#define isCNTRL_LC_utf8(p) _generic_LC_utf8(CNTRL, p)
1974#define isDIGIT_LC_utf8(p) _generic_LC_utf8(DIGIT, p)
1975#define isGRAPH_LC_utf8(p) _generic_LC_utf8(GRAPH, p)
1976#define isIDCONT_LC_utf8(p) _generic_LC_utf8(IDCONT, p)
1977#define isIDFIRST_LC_utf8(p) _generic_LC_utf8(IDFIRST, p)
1978#define isLOWER_LC_utf8(p) _generic_LC_utf8(LOWER, p)
1979#define isPRINT_LC_utf8(p) _generic_LC_utf8(PRINT, p)
1980#define isPSXSPC_LC_utf8(p) _generic_LC_utf8(PSXSPC, p)
1981#define isPUNCT_LC_utf8(p) _generic_LC_utf8(PUNCT, p)
1982#define isSPACE_LC_utf8(p) _generic_LC_utf8(SPACE, p)
1983#define isUPPER_LC_utf8(p) _generic_LC_utf8(UPPER, p)
1984#define isWORDCHAR_LC_utf8(p) _generic_LC_utf8(WORDCHAR, p)
1985#define isXDIGIT_LC_utf8(p) _generic_LC_utf8(XDIGIT, p)
1986
da8c1a98
KW
1987/* For internal core Perl use only: the base macros for defining macros like
1988 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
1989 * point in 'p' is within the 0-255 range, it uses locale rules from the
1990 * passed-in 'macro' parameter */
1991#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
1992 (__ASSERT_(_utf8_safe_assert(p, e)) \
1993 (UTF8_IS_INVARIANT(*(p))) \
1994 ? macro(*(p)) \
1995 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1996 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1997 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
1998 : (_force_out_malformed_utf8_message( \
1999 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2000 : above_latin1))
2001
2002#define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \
2003 _generic_LC_utf8_safe(macro, p, e, \
2004 _is_utf8_FOO_with_len(classnum, p, e))
2005
2006#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2007 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2008
2009#define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \
2010 _generic_LC_utf8_safe(classnum, p, e, \
2011 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2012 ? (_force_out_malformed_utf8_message( \
2013 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2014 : above_latin1(p)))
2015
2016#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2017 _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \
2018 _CC_ALPHANUMERIC, p, e)
2019#define isALPHA_LC_utf8_safe(p, e) \
2020 _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2021#define isASCII_LC_utf8_safe(p, e) \
2022 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2023#define isBLANK_LC_utf8_safe(p, e) \
2024 _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2025#define isCNTRL_LC_utf8_safe(p, e) \
2026 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2027#define isDIGIT_LC_utf8_safe(p, e) \
2028 _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2029#define isGRAPH_LC_utf8_safe(p, e) \
2030 _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2031#define isIDCONT_LC_utf8_safe(p, e) \
2032 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2033 _is_utf8_perl_idcont_with_len, p, e)
2034#define isIDFIRST_LC_utf8_safe(p, e) \
2035 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2036 _is_utf8_perl_idstart_with_len, p, e)
2037#define isLOWER_LC_utf8_safe(p, e) \
2038 _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2039#define isPRINT_LC_utf8_safe(p, e) \
2040 _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2041#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2042#define isPUNCT_LC_utf8_safe(p, e) \
2043 _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2044#define isSPACE_LC_utf8_safe(p, e) \
2045 _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2046#define isUPPER_LC_utf8_safe(p, e) \
2047 _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2048#define isWORDCHAR_LC_utf8_safe(p, e) \
2049 _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2050#define isXDIGIT_LC_utf8_safe(p, e) \
2051 _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2052
fbc19f27
KW
2053/* Macros for backwards compatibility and for completeness when the ASCII and
2054 * Latin1 values are identical */
b7d90381
KW
2055#define isALPHAU(c) isALPHA_L1(c)
2056#define isDIGIT_L1(c) isDIGIT_A(c)
2057#define isOCTAL(c) isOCTAL_A(c)
2058#define isOCTAL_L1(c) isOCTAL_A(c)
2059#define isXDIGIT_L1(c) isXDIGIT_A(c)
2060#define isALNUM(c) isWORDCHAR(c)
2061#define isALNUMU(c) isWORDCHAR_L1(c)
2062#define isALNUM_LC(c) isWORDCHAR_LC(c)
2063#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2064#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
2065#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
2066#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
2067#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2068#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2069#define isALNUMC(c) isALPHANUMERIC(c)
2070#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2071#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2072#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 2073#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 2074#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 2075
2bd1cbf6
KW
2076/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2077 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2078 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2079 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2080 * the outlier from the block that contains the other controls, just like
2081 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2082 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2083 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2084 * etc. */
2bd1cbf6 2085#ifndef EBCDIC
75763b3a 2086# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2087#else
75763b3a
KW
2088# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2089 ((isPRINT_A(c)) \
2090 ? (UNLIKELY((c) == '?') \
2091 ? QUESTION_MARK_CTRL \
2092 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2093 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2094 ? '?' \
2095 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2096#endif
bbce6d69 2097
dea28490
JJ
2098/* Line numbers are unsigned, 32 bits. */
2099typedef U32 line_t;
e5dcd934 2100#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2101
91152fc1
DG
2102/* Helpful alias for version prescan */
2103#define is_LAX_VERSION(a,b) \
2104 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2105
2106#define is_STRICT_VERSION(a,b) \
2107 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2108
2109#define BADVERSION(a,b,c) \
2110 if (b) { \
2111 *b = c; \
2112 } \
2113 return a;
8c52afec 2114
cb27eebd
KW
2115/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2116 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2117 * which is advanced. The input is validated only by an assert() in DEBUGGING
2118 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2119 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2120#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2121 ? (c) \
2122 : ((c) + 9))))
902008b8
KW
2123#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2124 ? (*(s)++) \
2125 : (*(s)++ + 9))))
95a59cab 2126
cb27eebd
KW
2127/* Converts a character known to represent an octal digit (0-7) to its numeric
2128 * value. The input is validated only by an assert() in DEBUGGING builds. In
2129 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2130#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2131
305b8651
KW
2132/* Efficiently returns a boolean as to if two native characters are equivalent
2133 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2134 * the ALPHA in the name is to remind you of that. This is asserted() in
2135 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2136 * works (on valid input) for both non- and UTF-8-encoded bytes.
2137 *
2138 * When one of the inputs is a compile-time constant and gets folded by the
2139 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2140 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2141 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2142 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2143 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2144 * just a single 0, in the bit position where the upper- and lowercase differ.
2145 * */
305b8651
KW
2146#define isALPHA_FOLD_EQ(c1, c2) \
2147 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2148 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2149#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2150
8e84507e 2151/*
ccfc67b7
JH
2152=head1 Memory Management
2153
a02a5408 2154=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2155The XSUB-writer's interface to the C C<malloc> function.
2156
596f7718 2157Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2158
c5008215
JC
2159In 5.9.3, Newx() and friends replace the older New() API, and drops
2160the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2161themselves. This aid has been superseded by a new build option,
d10b4965 2162PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2163there for use in XS modules supporting older perls.
2164
a02a5408 2165=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2166The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2167cast. See also C<L</Newx>>.
954c1994 2168
596f7718 2169Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2170
a02a5408 2171=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2172The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2173memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2174
596f7718 2175Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2176
954c1994
GS
2177=for apidoc Am|void|Renew|void* ptr|int nitems|type
2178The XSUB-writer's interface to the C C<realloc> function.
2179
596f7718 2180Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2181
954c1994
GS
2182=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2183The XSUB-writer's interface to the C C<realloc> function, with
2184cast.
2185
596f7718 2186Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2187
49b8b560 2188=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2189The XSUB-writer's interface to the C C<free> function.
2190
596f7718 2191This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2192
954c1994
GS
2193=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2194The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2195source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2196C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2197
e90e2364 2198=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2199Like C<Move> but returns C<dest>. Useful
72d33970 2200for encouraging compilers to tail-call
e90e2364
NC
2201optimise.
2202
954c1994
GS
2203=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2204The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2205source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2206C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2207
e90e2364
NC
2208=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2209
796b6530 2210Like C<Copy> but returns C<dest>. Useful
72d33970 2211for encouraging compilers to tail-call
e90e2364
NC
2212optimise.
2213
954c1994
GS
2214=for apidoc Am|void|Zero|void* dest|int nitems|type
2215
2216The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2217destination, C<nitems> is the number of items, and C<type> is the type.
2218
e90e2364
NC
2219=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2220
72d33970
FC
2221Like C<Zero> but returns dest. Useful
2222for encouraging compilers to tail-call
e90e2364
NC
2223optimise.
2224
da5d8dbb 2225=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2226This is an architecture-independent macro to copy one structure to another.
954c1994 2227
7e337ee0
JH
2228=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2229
2230Fill up memory with a byte pattern (a byte repeated over and over
2231again) that hopefully catches attempts to access uninitialized memory.
2232
2233=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2234
2235PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2236
1c12ffb4 2237=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2238
2239PoisonWith(0xEF) for catching access to freed memory.
2240
9965345d
JH
2241=for apidoc Am|void|Poison|void* dest|int nitems|type
2242
7e337ee0 2243PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2244
2245=cut */
954c1994 2246
561b68a9
SH
2247/* Maintained for backwards-compatibility only. Use newSV() instead. */
2248#ifndef PERL_CORE
ff06c60c 2249#define NEWSV(x,len) newSV(len)
561b68a9 2250#endif
ff06c60c 2251
19a94d75
JH
2252#define MEM_SIZE_MAX ((MEM_SIZE)~0)
2253
e6bdf523 2254
27d5b266 2255#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2256
2257/* This expression will be constant-folded at compile time. It checks
2258 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2259 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2260 * As well as avoiding the need for a run-time check in some cases, it's
2261 * designed to avoid compiler warnings like:
2262 * comparison is always false due to limited range of data type
73e8ff00
DM
2263 * It's mathematically equivalent to
2264 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2265 */
2266
2267# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
004073ba 2268 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
e6bdf523 2269
88f9f128 2270/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2271 * compiler warnings. We *want* to write the expression as
2272 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2273 * (for some compile-time constant C), but even when the LHS
2274 * constant-folds to false at compile-time, g++ insists on emitting
2275 * warnings about the RHS (e.g. "comparison is always false"), so instead
2276 * we write it as
e6bdf523 2277 *
d98e5cde 2278 * (cond ? n : X) > C
88f9f128 2279 *
d98e5cde
DM
2280 * where X is a constant with X > C always false. Choosing a value for X
2281 * is tricky. If 0, some compilers will complain about 0 > C always being
2282 * false; if 1, Coverity complains when n happens to be the constant value
2283 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2284 * for X and hope that nothing else whines.
e6bdf523
DM
2285 */
2286
2287# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2288 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2289 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2290
2291# define MEM_WRAP_CHECK(n,t) \
2292 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2293 && (croak_memory_wrap(),0))
2294
2295# define MEM_WRAP_CHECK_1(n,t,a) \
2296 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2297 && (Perl_croak_nocontext("%s",(a)),0))
2298
8b44ba4c 2299#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2300
d1decf2b 2301#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
2302#else
2303
410319be
NC
2304#define MEM_WRAP_CHECK(n,t)
2305#define MEM_WRAP_CHECK_1(n,t,a)
2306#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
2307#define MEM_WRAP_CHECK_(n,t)
2308
6fff79ce 2309#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 2310
1936d2a7 2311#endif
8b44ba4c 2312
fe4f188c 2313#ifdef PERL_MEM_LOG
46c6c7e2 2314/*
9f653bb5 2315 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2316 * go through functions, which are handy for debugging breakpoints, but
2317 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2318 * line number, and C function name if available) passed in. This info can
2319 * then be used for logging the calls, for which one gets a sample
73d1d973 2320 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2321 *
46c6c7e2 2322 * Known problems:
94e892a6 2323 * - not all memory allocs get logged, only those
46c6c7e2 2324 * that go through Newx() and derivatives (while all
94e892a6 2325 * Safefrees do get logged)
46c6c7e2
JH
2326 * - __FILE__ and __LINE__ do not work everywhere
2327 * - __func__ or __FUNCTION__ even less so
2328 * - I think more goes on after the perlio frees but
2329 * the thing is that STDERR gets closed (as do all
2330 * the file descriptors)
2331 * - no deeper calling stack than the caller of the Newx()
2332 * or the kind, but do I look like a C reflection/introspection
2333 * utility to you?
2334 * - the function prototypes for the logging functions
2335 * probably should maybe be somewhere else than handy.h
2336 * - one could consider inlining (macrofying) the logging
2337 * for speed, but I am too lazy
2338 * - one could imagine recording the allocations in a hash,
2339 * (keyed by the allocation address?), and maintain that
2340 * through reallocs and frees, but how to do that without
2341 * any News() happening...?
73d1d973 2342 * - lots of -Ddefines to get useful/controllable output
b953482e 2343 * - lots of ENV reads
46c6c7e2
JH
2344 */
2345
0b0ab801 2346# ifdef PERL_CORE
73d1d973 2347# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2348enum mem_log_type {
2349 MLT_ALLOC,
2350 MLT_REALLOC,
d7a2c63c
MHM
2351 MLT_FREE,
2352 MLT_NEW_SV,
2353 MLT_DEL_SV
0b0ab801
MHM
2354};
2355# endif
12754f92 2356# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2357void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2358void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2359# endif
0b0ab801
MHM
2360# endif
2361
fe4f188c
JH
2362#endif
2363
2364#ifdef PERL_MEM_LOG
d1401ee9
MHM
2365#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2366#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2367#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2368#endif
2369
2370#ifndef MEM_LOG_ALLOC
2371#define MEM_LOG_ALLOC(n,t,a) (a)
2372#endif
2373#ifndef MEM_LOG_REALLOC
2374#define MEM_LOG_REALLOC(n,t,v,a) (a)
2375#endif
2376#ifndef MEM_LOG_FREE
2377#define MEM_LOG_FREE(a) (a)
2378#endif
2379
d1401ee9
MHM
2380#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2381#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2382#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2383
2384#ifndef PERL_CORE
a02a5408
JC
2385/* pre 5.9.x compatibility */
2386#define New(x,v,n,t) Newx(v,n,t)
2387#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2388#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2389#endif
a02a5408 2390
ff68c719 2391#define Renew(v,n,t) \
d1401ee9 2392 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2393#define Renewc(v,n,t,c) \
d1401ee9 2394 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2395
2396#ifdef PERL_POISON
2397#define Safefree(d) \
06c0cc96 2398 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2399#else
fe4f188c 2400#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2401#endif
55497cff 2402
45908e4d 2403#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2404
45908e4d
YO
2405
2406#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2407#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2408#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2409
2410#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2411#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
45908e4d 2412#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2413
7e337ee0
JH
2414#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2415#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2416#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2417#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2418
caa674f3
DD
2419#ifdef PERL_POISON
2420# define PERL_POISON_EXPR(x) x
2421#else
2422# define PERL_POISON_EXPR(x)
2423#endif
2424
ff68c719 2425#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2426
c3caa5c3
JH
2427/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2428 * want your zero-based indices to be less than but not equal to).
2429 *
2430 * C_ARRAY_END is one past the last: half-open/half-closed range,
2431 * not last-inclusive range. */
622913ab 2432#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2433#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2434
2cc61e15
DD
2435#ifdef NEED_VA_COPY
2436# ifdef va_copy
2437# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2438# elif defined(__va_copy)
2439# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2440# else
07798b17 2441# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2442# endif
2443#endif
2444
472d47bc
SB
2445/* convenience debug macros */
2446#ifdef USE_ITHREADS
2447#define pTHX_FORMAT "Perl interpreter: 0x%p"
2448#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2449#define pTHX_VALUE_ (void *)my_perl,
2450#define pTHX_VALUE (void *)my_perl
2451#define pTHX__VALUE_ ,(void *)my_perl,
2452#define pTHX__VALUE ,(void *)my_perl
472d47bc 2453#else
3609ea0d 2454#define pTHX_FORMAT
472d47bc 2455#define pTHX__FORMAT
3609ea0d 2456#define pTHX_VALUE_
472d47bc 2457#define pTHX_VALUE
3609ea0d 2458#define pTHX__VALUE_
472d47bc
SB
2459#define pTHX__VALUE
2460#endif /* USE_ITHREADS */
3609ea0d 2461
2acdbac1
NC
2462/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2463 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2464 CPAN::Unpack show any users outside the core. */
2465#ifdef PERL_CORE
dc6e8de0
A
2466# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2467 "Use of " s " is deprecated")
c9680906
A
2468# define deprecate_disappears_in(when,message) \
2469 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2470 message ", and will disappear in Perl " when)
ac641426
A
2471# define deprecate_fatal_in(when,message) \
2472 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2473 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2474#endif
2475
dfff4baf
BF
2476/* Internal macros to deal with gids and uids */
2477#ifdef PERL_CORE
2478
2479# if Uid_t_size > IVSIZE
2480# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2481# define SvUID(sv) SvNV(sv)
07798b17
AC
2482# elif Uid_t_sign <= 0
2483# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2484# define SvUID(sv) SvIV(sv)
dfff4baf 2485# else
07798b17
AC
2486# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2487# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2488# endif /* Uid_t_size */
2489
2490# if Gid_t_size > IVSIZE
2491# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2492# define SvGID(sv) SvNV(sv)
07798b17
AC
2493# elif Gid_t_sign <= 0
2494# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2495# define SvGID(sv) SvIV(sv)
dfff4baf 2496# else
07798b17
AC
2497# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2498# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2499# endif /* Gid_t_size */
2500
2501#endif
2502
6a5bc5ac 2503#endif /* PERL_HANDY_H_ */
9d745869 2504
e9a8c099 2505/*
14d04a33 2506 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2507 */