This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Change some strncmp(), etc. to strBEGINs()
[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
8d063cd8
LW
469#define strNE(s1,s2) (strcmp(s1,s2))
470#define strEQ(s1,s2) (!strcmp(s1,s2))
471#define strLT(s1,s2) (strcmp(s1,s2) < 0)
472#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
473#define strGT(s1,s2) (strcmp(s1,s2) > 0)
474#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 475
8d063cd8
LW
476#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
477#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
378cc40b 478
c8b388b0 479#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
62946e08 480
e5d7f4e5
AC
481#define memNE(s1,s2,l) (memcmp(s1,s2,l))
482#define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
36477c24 483
085b7534 484/* memEQ and memNE where second comparand is a string constant */
568a785a 485#define memEQs(s1, l, s2) \
777fa2cb 486 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 487#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 488
777fa2cb
YO
489/* memEQ and memNE where second comparand is a string constant
490 * and we can assume the length of s1 is at least that of the string */
491#define _memEQs(s1, s2) \
492 (memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
493#define _memNEs(s1, s2) (memNE((s1),("" s2 ""),(sizeof(s2)-1)))
494
062b6850
KW
495#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
496#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
497#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
498#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
499
bbce6d69
PP
500/*
501 * Character classes.
502 *
503 * Unfortunately, the introduction of locales means that we
504 * can't trust isupper(), etc. to tell the truth. And when
505 * it comes to /\w+/ with tainting enabled, we *must* be able
506 * to trust our character classes.
507 *
508 * Therefore, the default tests in the text of Perl will be
509 * independent of locale. Any code that wants to depend on
510 * the current locale will use the tests that begin with "lc".
511 */
512
2304df62
AD
513#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
514# ifndef CTYPE256
515# define CTYPE256
516# endif
517#endif
518
954c1994 519/*
ccfc67b7 520
dcccc8ff 521=head1 Character classification
243effed
KW
522This section is about functions (really macros) that classify characters
523into types, such as punctuation versus alphabetic, etc. Most of these are
524analogous to regular expression character classes. (See
525L<perlrecharclass/POSIX Character Classes>.) There are several variants for
526each class. (Not all macros have all variants; each item below lists the
527ones valid for it.) None are affected by C<use bytes>, and only the ones
528with C<LC> in the name are affected by the current locale.
529
530The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
531C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
532represented by that octet is (or on non-ASCII platforms, corresponds to) an
533ASCII character in the named class based on platform, Unicode, and Perl rules.
534If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 535
c98722a4 536Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
537with no suffix C<"_A">. This variant is used to emphasize by its name that
538only ASCII-range characters can return TRUE.
4b9734bf 539
d60679e1 540Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
541onto the platform. That is, the code points that are ASCII are unaffected,
542since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
543as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
544true when called with the code point 0xDF, which is a word character in both
4650c663 545ASCII and EBCDIC (though it represents different characters in each).
243effed 546
c98722a4 547Variant C<isI<FOO>_uvchr> is like the C<isI<FOO>_L1> variant, but accepts any UV code
243effed
KW
548point as input. If the code point is larger than 255, Unicode rules are used
549to determine if it is in the character class. For example,
d0da05db 550C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 551WITH MACRON in Unicode, and is a word character.
243effed 552
c98722a4 553Variant C<isI<FOO>_utf8_safe> is like C<isI<FOO>_uvchr>, but is used for UTF-8
da8c1a98
KW
554encoded strings. Each call classifies one character, even if the string
555contains many. This variant takes two parameters. The first, C<p>, is a
556pointer to the first byte of the character to be classified. (Recall that it
557may take more than one byte to represent a character in UTF-8 strings.) The
558second parameter, C<e>, points to anywhere in the string beyond the first
559character, up to one byte past the end of the entire string. The suffix
560C<_safe> in the function's name indicates that it will not attempt to read
561beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
562is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
563character is malformed in some way, the program may croak, or the function may
564return FALSE, at the discretion of the implementation, and subject to change in
565future releases.
566
c98722a4 567Variant C<isI<FOO>_utf8> is like C<isI<FOO>_utf8_safe>, but takes just a single
da8c1a98 568parameter, C<p>, which has the same meaning as the corresponding parameter does
c98722a4 569in C<isI<FOO>_utf8_safe>. The function therefore can't check if it is reading
34aeb2e9 570beyond the end of the string. Starting in Perl v5.30, it will take a second
c98722a4 571parameter, becoming a synonym for C<isI<FOO>_utf8_safe>. At that time every
34aeb2e9 572program that uses it will have to be changed to successfully compile. In the
c98722a4 573meantime, the first runtime call to C<isI<FOO>_utf8> from each call point in the
34aeb2e9 574program will raise a deprecation warning, enabled by default. You can convert
c98722a4 575your program now to use C<isI<FOO>_utf8_safe>, and avoid the warnings, and get an
34aeb2e9
KW
576extra measure of protection, or you can wait until v5.30, when you'll be forced
577to add the C<e> parameter.
243effed 578
c98722a4 579Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants, but the
31f05a37
KW
580result is based on the current locale, which is what C<LC> in the name stands
581for. If Perl can determine that the current locale is a UTF-8 locale, it uses
582the published Unicode rules; otherwise, it uses the C library function that
583gives the named classification. For example, C<isDIGIT_LC()> when not in a
584UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
585returned if the input won't fit into an octet. On some platforms where the C
586library function is known to be defective, Perl changes its result to follow
587the POSIX standard's rules.
243effed 588
c98722a4
KW
589Variant C<isI<FOO>_LC_uvchr> is like C<isI<FOO>_LC>, but is defined on any UV. It
590returns the same as C<isI<FOO>_LC> for input code points less than 256, and
243effed
KW
591returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
592
c98722a4 593Variant C<isI<FOO>_LC_utf8_safe> is like C<isI<FOO>_LC_uvchr>, but is used for UTF-8
da8c1a98
KW
594encoded strings. Each call classifies one character, even if the string
595contains many. This variant takes two parameters. The first, C<p>, is a
596pointer to the first byte of the character to be classified. (Recall that it
597may take more than one byte to represent a character in UTF-8 strings.) The
598second parameter, C<e>, points to anywhere in the string beyond the first
599character, up to one byte past the end of the entire string. The suffix
600C<_safe> in the function's name indicates that it will not attempt to read
601beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
602is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
603character is malformed in some way, the program may croak, or the function may
604return FALSE, at the discretion of the implementation, and subject to change in
605future releases.
606
c98722a4 607Variant C<isI<FOO>_LC_utf8> is like C<isI<FOO>_LC_utf8_safe>, but takes just a single
da8c1a98 608parameter, C<p>, which has the same meaning as the corresponding parameter does
c98722a4 609in C<isI<FOO>_LC_utf8_safe>. The function therefore can't check if it is reading
34aeb2e9 610beyond the end of the string. Starting in Perl v5.30, it will take a second
c98722a4 611parameter, becoming a synonym for C<isI<FOO>_LC_utf8_safe>. At that time every
34aeb2e9 612program that uses it will have to be changed to successfully compile. In the
c98722a4 613meantime, the first runtime call to C<isI<FOO>_LC_utf8> from each call point in
34aeb2e9 614the program will raise a deprecation warning, enabled by default. You can
c98722a4 615convert your program now to use C<isI<FOO>_LC_utf8_safe>, and avoid the warnings,
34aeb2e9
KW
616and get an extra measure of protection, or you can wait until v5.30, when
617you'll be forced to add the C<e> parameter.
ccfc67b7 618
954c1994 619=for apidoc Am|bool|isALPHA|char ch
2787a470 620Returns a boolean indicating whether the specified character is an
6aff1f14 621alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
622See the L<top of this section|/Character classification> for an explanation of
623variants
da8c1a98
KW
624C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8_safe>,
625C<isALPHA_LC>, C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8_safe>.
8a58bdcf 626
15861f94
KW
627=for apidoc Am|bool|isALPHANUMERIC|char ch
628Returns a boolean indicating whether the specified character is a either an
6aff1f14 629alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
630See the L<top of this section|/Character classification> for an explanation of
631variants
d0da05db 632C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
da8c1a98
KW
633C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>,
634and C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 635
8a58bdcf
KW
636=for apidoc Am|bool|isASCII|char ch
637Returns a boolean indicating whether the specified character is one of the 128
243effed 638characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 639On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
640character corresponds to an ASCII character. Variants C<isASCII_A()> and
641C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
642See the L<top of this section|/Character classification> for an explanation of
643variants
da8c1a98
KW
644C<isASCII_uvchr>, C<isASCII_utf8_safe>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
645C<isASCII_LC_utf8_safe>. Note, however, that some platforms do not have the C
243effed
KW
646library routine C<isascii()>. In these cases, the variants whose names contain
647C<LC> are the same as the corresponding ones without.
648
d98532ea
KW
649Also note, that because all ASCII characters are UTF-8 invariant (meaning they
650have the exact same representation (always a single byte) whether encoded in
651UTF-8 or not), C<isASCII> will give the correct results when called with any
da8c1a98
KW
652byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8_safe>
653will work properly on any string encoded or not in UTF-8.
d98532ea 654
243effed
KW
655=for apidoc Am|bool|isBLANK|char ch
656Returns a boolean indicating whether the specified character is a
6aff1f14 657character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
658See the L<top of this section|/Character classification> for an explanation of
659variants
da8c1a98
KW
660C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8_safe>,
661C<isBLANK_LC>, C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8_safe>. Note,
662however, that some platforms do not have the C library routine
663C<isblank()>. In these cases, the variants whose names contain C<LC> are
664the same as the corresponding ones without.
243effed
KW
665
666=for apidoc Am|bool|isCNTRL|char ch
667Returns a boolean indicating whether the specified character is a
6aff1f14 668control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
669See the L<top of this section|/Character classification> for an explanation of
670variants
da8c1a98
KW
671C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8_safe>,
672C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8_safe> On EBCDIC
673platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 674
954c1994 675=for apidoc Am|bool|isDIGIT|char ch
2787a470 676Returns a boolean indicating whether the specified character is a
6aff1f14 677digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 678Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
679See the L<top of this section|/Character classification> for an explanation of
680variants
da8c1a98
KW
681C<isDIGIT_uvchr>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
682C<isDIGIT_LC_utf8_safe>.
243effed
KW
683
684=for apidoc Am|bool|isGRAPH|char ch
685Returns a boolean indicating whether the specified character is a
6aff1f14 686graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 687See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
688variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8_safe>,
689C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8_safe>.
954c1994 690
0c82b6df 691=for apidoc Am|bool|isLOWER|char ch
2787a470 692Returns a boolean indicating whether the specified character is a
6aff1f14 693lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
694See the L<top of this section|/Character classification> for an explanation of
695variants
da8c1a98
KW
696C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8_safe>,
697C<isLOWER_LC>, C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8_safe>.
0c82b6df 698
c99e91e9 699=for apidoc Am|bool|isOCTAL|char ch
2787a470 700Returns a boolean indicating whether the specified character is an
6aff1f14 701octal digit, [0-7].
243effed
KW
702The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
703C<isOCTAL>.
704
705=for apidoc Am|bool|isPUNCT|char ch
706Returns a boolean indicating whether the specified character is a
6aff1f14
KW
707punctuation character, analogous to C<m/[[:punct:]]/>.
708Note that the definition of what is punctuation isn't as
243effed
KW
709straightforward as one might desire. See L<perlrecharclass/POSIX Character
710Classes> for details.
dcccc8ff 711See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
712variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8_safe>,
713C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8_safe>.
c99e91e9 714
0c82b6df 715=for apidoc Am|bool|isSPACE|char ch
2787a470 716Returns a boolean indicating whether the specified character is a
6aff1f14 717whitespace character. This is analogous
398d098a 718to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 719this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
720locale forms of this macro (the ones with C<LC> in their names) matched
721precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
722in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
723(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
724See the L<top of this section|/Character classification> for an explanation of
725variants
da8c1a98
KW
726C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8_safe>,
727C<isSPACE_LC>, C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8_safe>.
0c82b6df 728
398d098a
KW
729=for apidoc Am|bool|isPSXSPC|char ch
730(short for Posix Space)
779cf272
KW
731Starting in 5.18, this is identical in all its forms to the
732corresponding C<isSPACE()> macros.
398d098a
KW
733The locale forms of this macro are identical to their corresponding
734C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
735non-locale forms differ from their C<isSPACE()> forms only in that the
736C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
737Otherwise they are identical. Thus this macro is analogous to what
738C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 739See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
740variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8_safe>,
741C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 742
954c1994 743=for apidoc Am|bool|isUPPER|char ch
2787a470 744Returns a boolean indicating whether the specified character is an
6aff1f14 745uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 746See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
747variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8_safe>,
748C<isUPPER_LC>, C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8_safe>.
954c1994 749
243effed 750=for apidoc Am|bool|isPRINT|char ch
8eea39dd 751Returns a boolean indicating whether the specified character is a
6aff1f14 752printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
753See the L<top of this section|/Character classification> for an explanation of
754variants
da8c1a98
KW
755C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8_safe>,
756C<isPRINT_LC>, C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8_safe>.
243effed
KW
757
758=for apidoc Am|bool|isWORDCHAR|char ch
759Returns a boolean indicating whether the specified character is a character
760that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
761in a regular expression. A word character is an alphabetic character, a
762decimal digit, a connecting punctuation character (such as an underscore), or
763a "mark" character that attaches to one of those (like some sort of accent).
764C<isALNUM()> is a synonym provided for backward compatibility, even though a
765word character includes more than the standard C language meaning of
766alphanumeric.
dcccc8ff 767See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
768variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and
769C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and
770C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
771include the platform's native underscore.
8a58bdcf
KW
772
773=for apidoc Am|bool|isXDIGIT|char ch
774Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
775digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
776and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
777See the L<top of this section|/Character classification> for an explanation of
778variants
da8c1a98
KW
779C<isXDIGIT_uvchr>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>,
780and C<isXDIGIT_LC_utf8_safe>.
243effed 781
3c3ecf18
KW
782=for apidoc Am|bool|isIDFIRST|char ch
783Returns a boolean indicating whether the specified character can be the first
784character of an identifier. This is very close to, but not quite the same as
785the official Unicode property C<XID_Start>. The difference is that this
786returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
787See the L<top of this section|/Character classification> for an explanation of
788variants
da8c1a98
KW
789C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8_safe>,
790C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18
KW
791
792=for apidoc Am|bool|isIDCONT|char ch
793Returns a boolean indicating whether the specified character can be the
794second or succeeding character of an identifier. This is very close to, but
795not quite the same as the official Unicode property C<XID_Continue>. The
796difference is that this returns true only if the input character also matches
dcccc8ff
KW
797L</isWORDCHAR>. See the L<top of this section|/Character classification> for
798an
d0da05db 799explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
da8c1a98
KW
800C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
801C<isIDCONT_LC_utf8_safe>.
3c3ecf18 802
243effed 803=head1 Miscellaneous Functions
8eea39dd 804
95a59cab 805=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 806Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
807Behaviour is only well defined when isXDIGIT(*str) is true.
808
e7c1e6c1 809=head1 Character case changing
21da7284
KW
810Perl uses "full" Unicode case mappings. This means that converting a single
811character to another case may result in a sequence of more than one character.
812For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
813character sequence C<SS>. This presents some complications The lowercase of
814all characters in the range 0..255 is a single character, and thus
815C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
816return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
817an API that does allow every possible legal result to be returned.) Likewise
818no other function that is crippled by not being able to give the correct
819results for the full range of possible inputs has been implemented here.
e7c1e6c1 820
1f607577
KW
821=for apidoc Am|U8|toUPPER|U8 ch
822Converts the specified character to uppercase. If the input is anything but an
823ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 824C<toUPPER_A> is equivalent.
954c1994 825
d0da05db
KW
826=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
827Converts the code point C<cp> to its uppercase version, and
828stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
829point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
830that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
831bytes since the uppercase version may be longer than the original character.
832
833The first code point of the uppercased version is returned
21da7284
KW
834(but note, as explained at L<the top of this section|/Character case
835changing>, that there may be more.)
1f607577 836
a239b1e2
KW
837=for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
838Converts the first UTF-8 encoded character in the sequence starting at C<p> and
839extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
840stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
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
848The suffix C<_safe> in the function's name indicates that it will not attempt
849to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
850true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
851input character is malformed in some way, the program may croak, or the
852function may return the REPLACEMENT CHARACTER, at the discretion of the
853implementation, and subject to change in future releases.
854
855=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
856This is like C<L</toUPPER_utf8_safe>>, but doesn't have the C<e>
857parameter The function therefore can't check if it is reading
607313a1
KW
858beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
859parameter, becoming a synonym for C<toUPPER_utf8_safe>. At that time every
860program that uses it will have to be changed to successfully compile. In the
861meantime, the first runtime call to C<toUPPER_utf8> from each call point in the
862program will raise a deprecation warning, enabled by default. You can convert
863your program now to use C<toUPPER_utf8_safe>, and avoid the warnings, and get an
864extra measure of protection, or you can wait until v5.30, when you'll be forced
865to add the C<e> parameter.
1f607577 866
25200305
KW
867=for apidoc Am|U8|toFOLD|U8 ch
868Converts the specified character to foldcase. If the input is anything but an
869ASCII uppercase character, that input character itself is returned. Variant
870C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 871Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 872
d0da05db
KW
873=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
874Converts the code point C<cp> to its foldcase version, and
875stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
876point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
877that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
878bytes since the foldcase version may be longer than the original character.
879
880The first code point of the foldcased version is returned
21da7284
KW
881(but note, as explained at L<the top of this section|/Character case
882changing>, that there may be more).
1f607577 883
a239b1e2
KW
884=for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
885Converts the first UTF-8 encoded character in the sequence starting at C<p> and
886extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
887stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
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
895The suffix C<_safe> in the function's name indicates that it will not attempt
896to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
897true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
898input character is malformed in some way, the program may croak, or the
899function may return the REPLACEMENT CHARACTER, at the discretion of the
900implementation, and subject to change in future releases.
901
902=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
903This is like C<L</toFOLD_utf8_safe>>, but doesn't have the C<e>
904parameter The function therefore can't check if it is reading
607313a1
KW
905beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
906parameter, becoming a synonym for C<toFOLD_utf8_safe>. At that time every
907program that uses it will have to be changed to successfully compile. In the
908meantime, the first runtime call to C<toFOLD_utf8> from each call point in the
909program will raise a deprecation warning, enabled by default. You can convert
910your program now to use C<toFOLD_utf8_safe>, and avoid the warnings, and get an
911extra measure of protection, or you can wait until v5.30, when you'll be forced
912to add the C<e> parameter.
1f607577
KW
913
914=for apidoc Am|U8|toLOWER|U8 ch
915Converts the specified character to lowercase. If the input is anything but an
916ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 917C<toLOWER_A> is equivalent.
954c1994 918
1f607577 919=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
920Converts the specified Latin1 character to lowercase. The results are
921undefined if the input doesn't fit in a byte.
1f607577
KW
922
923=for apidoc Am|U8|toLOWER_LC|U8 ch
924Converts the specified character to lowercase using the current locale's rules,
925if possible; otherwise returns the input character itself.
926
d0da05db
KW
927=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
928Converts the code point C<cp> to its lowercase version, and
929stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
930point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
931that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
932bytes since the lowercase version may be longer than the original character.
933
934The first code point of the lowercased version is returned
21da7284
KW
935(but note, as explained at L<the top of this section|/Character case
936changing>, that there may be more).
1f607577 937
a239b1e2
KW
938
939=for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
940Converts the first UTF-8 encoded character in the sequence starting at C<p> and
941extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
942stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
943that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
944bytes since the lowercase version may be longer than the original character.
945
946The first code point of the lowercased version is returned
21da7284
KW
947(but note, as explained at L<the top of this section|/Character case
948changing>, that there may be more).
1f607577 949
a239b1e2
KW
950The suffix C<_safe> in the function's name indicates that it will not attempt
951to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
952true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
953input character is malformed in some way, the program may croak, or the
954function may return the REPLACEMENT CHARACTER, at the discretion of the
955implementation, and subject to change in future releases.
956
957=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
958This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
959parameter The function therefore can't check if it is reading
607313a1
KW
960beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
961parameter, becoming a synonym for C<toLOWER_utf8_safe>. At that time every
962program that uses it will have to be changed to successfully compile. In the
963meantime, the first runtime call to C<toLOWER_utf8> from each call point in the
964program will raise a deprecation warning, enabled by default. You can convert
965your program now to use C<toLOWER_utf8_safe>, and avoid the warnings, and get an
966extra measure of protection, or you can wait until v5.30, when you'll be forced
967to add the C<e> parameter.
1f607577 968
25200305
KW
969=for apidoc Am|U8|toTITLE|U8 ch
970Converts the specified character to titlecase. If the input is anything but an
971ASCII lowercase character, that input character itself is returned. Variant
b7d90381 972C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 973range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 974not a concept used in locale handling, so there is no functionality for that.)
25200305 975
d0da05db
KW
976=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
977Converts the code point C<cp> to its titlecase version, and
978stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
979point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
980that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
981bytes since the titlecase version may be longer than the original character.
982
983The first code point of the titlecased version is returned
21da7284
KW
984(but note, as explained at L<the top of this section|/Character case
985changing>, that there may be more).
1f607577 986
a239b1e2
KW
987=for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
988Converts the first UTF-8 encoded character in the sequence starting at C<p> and
989extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
990stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
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
998The suffix C<_safe> in the function's name indicates that it will not attempt
999to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1000true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1001input character is malformed in some way, the program may croak, or the
1002function may return the REPLACEMENT CHARACTER, at the discretion of the
1003implementation, and subject to change in future releases.
1004
1005=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
1006This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1007parameter The function therefore can't check if it is reading
607313a1
KW
1008beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
1009parameter, becoming a synonym for C<toTITLE_utf8_safe>. At that time every
1010program that uses it will have to be changed to successfully compile. In the
1011meantime, the first runtime call to C<toTITLE_utf8> from each call point in the
1012program will raise a deprecation warning, enabled by default. You can convert
1013your program now to use C<toTITLE_utf8_safe>, and avoid the warnings, and get an
1014extra measure of protection, or you can wait until v5.30, when you'll be forced
1015to add the C<e> parameter.
1f607577 1016
954c1994 1017=cut
353c9b6f 1018
d0da05db 1019XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1020really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1021and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1022
8a58bdcf 1023Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1024patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1025
954c1994
GS
1026*/
1027
7c062697
KW
1028/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
1029 * is known only in the perl core, and this macro can be called from outside
1030 * that */
1031#ifdef HAS_QUAD
1032# define WIDEST_UTYPE U64TYPE
1033#else
1034# define WIDEST_UTYPE U32
1035#endif
1036
3912bc88
KW
1037/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1038 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1039 * bits of information are lost even on a 64-bit machine, but to get the
1040 * compiler to optimize it out if possible. This is because Configure makes
1041 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1042 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1043 *
1044 * For Coverity, be always true, because otherwise Coverity thinks
1045 * it finds several expressions that are always true, independent
1046 * of operands. Well, they are, but that is kind of the point.
220c71bf 1047 */
7e75d1a1 1048#ifndef __COVERITY__
6c5b02ac
KW
1049 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1050 * pointer) */
1051#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1052 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1053#else
1054#define FITS_IN_8_BITS(c) (1)
1055#endif
cf301eb7 1056
41f43cc2 1057#ifdef EBCDIC
b6340bd0 1058# ifndef _ALL_SOURCE
0852beac
KW
1059 /* The native libc isascii() et.al. functions return the wrong results
1060 * on at least z/OS unless this is defined. */
b6340bd0
KW
1061# error _ALL_SOURCE should probably be defined
1062# endif
41f43cc2 1063#else
0852beac
KW
1064 /* There is a simple definition of ASCII for ASCII platforms. But the
1065 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1066 * other macros below.
3f3c579d
KW
1067 *
1068 * The cast here is used instead of '(c) >= 0', because some compilers emit
1069 * a warning that that test is always true when the parameter is an
1070 * unsigned type. khw supposes that it could be written as
1071 * && ((c) == '\0' || (c) > 0)
1072 * to avoid the message, but the cast will likely avoid extra branches even
1073 * with stupid compilers.
1074 *
1075 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1076 * a pointer) */
9c903d59 1077# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1078#endif
1079
38694112
KW
1080/* Take the eight possible bit patterns of the lower 3 bits and you get the
1081 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1082 * can be ignored. If the rest match '0', we have an octal */
1083#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1084
9fb1bf9d 1085#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1086
a500dc72
KW
1087/* Character class numbers. For internal core Perl use only. The ones less
1088 * than 32 are used in PL_charclass[] and the ones up through the one that
1089 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1090 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1091 * their actual definitions are here. If that file has a name not used here,
1092 * it won't compile.
1709d539
KW
1093 *
1094 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
1095 * frequency of their use. This gives a slight edge to exiting a loop earlier
1096 * (in reginclass() in regexec.c) */
1709d539
KW
1097# define _CC_WORDCHAR 0 /* \w and [:word:] */
1098# define _CC_DIGIT 1 /* \d and [:digit:] */
1099# define _CC_ALPHA 2 /* [:alpha:] */
1100# define _CC_LOWER 3 /* [:lower:] */
1101# define _CC_UPPER 4 /* [:upper:] */
1102# define _CC_PUNCT 5 /* [:punct:] */
1103# define _CC_PRINT 6 /* [:print:] */
15861f94 1104# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1105# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1106# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1709d539 1107
b0d691b2 1108#define _FIRST_NON_SWASH_CC 10
1709d539
KW
1109/* The character classes above are implemented with swashes. The second group
1110 * (just below) contains the ones implemented without. These are also sorted
1111 * in rough order of the frequency of their use, except that \v should be last,
1112 * as it isn't a real Posix character class, and some (small) inefficiencies in
1113 * regular expression handling would be introduced by putting it in the middle
1114 * of those that are. Also, cntrl and ascii come after the others as it may be
1115 * useful to group these which have no members that match above Latin1, (or
1116 * above ASCII in the latter case) */
1117
779cf272 1118# define _CC_SPACE 10 /* \s, [:space:] */
34aeb2e9
KW
1119# define _CC_PSXSPC _CC_SPACE /* XXX Temporary, can be removed
1120 when the deprecated isFOO_utf8()
1121 functions are removed */
b0d691b2
KW
1122# define _CC_BLANK 11 /* [:blank:] */
1123# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1124# define _CC_CNTRL 13 /* [:cntrl:] */
1125# define _CC_ASCII 14 /* [:ascii:] */
1126# define _CC_VERTSPACE 15 /* \v */
1709d539 1127
a0947d7b
KW
1128# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1129
1709d539 1130/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1131 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1132# define _CC_IDFIRST 16
1133# define _CC_CHARNAME_CONT 17
1134# define _CC_NONLATIN1_FOLD 18
1135# define _CC_NONLATIN1_SIMPLE_FOLD 19
1136# define _CC_QUOTEMETA 20
1137# define _CC_NON_FINAL_FOLD 21
1138# define _CC_IS_IN_SOME_FOLD 22
1139# define _CC_MNEMONIC_CNTRL 23
073c22b3 1140
34aeb2e9
KW
1141# define _CC_IDCONT 24 /* XXX Temporary, can be removed when the deprecated
1142 isFOO_utf8() functions are removed */
1143
073c22b3
KW
1144/* This next group is only used on EBCDIC platforms, so theoretically could be
1145 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
1146# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1147# define _CC_UTF8_IS_START 29
1148# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1149# define _CC_UTF8_IS_CONTINUATION 31
1150/* Unused: 24-27
f4cdb42c
KW
1151 * If more bits are needed, one could add a second word for non-64bit
1152 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1153 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1154 * is used only for optimization (as of this writing), and differs in the
1155 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1156 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1157 * would just cause /i regexes which match them to run less efficiently.
1158 * Similarly the EBCDIC-only bits are used just for speed, and could be
1159 * replaced by other means */
96ac0975 1160
3a371f2f
KW
1161#if defined(PERL_CORE) || defined(PERL_EXT)
1162/* An enum version of the character class numbers, to help compilers
1163 * optimize */
1164typedef enum {
3a371f2f 1165 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1166 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1167 _CC_ENUM_ASCII = _CC_ASCII,
1168 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1169 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1170 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1171 _CC_ENUM_DIGIT = _CC_DIGIT,
1172 _CC_ENUM_GRAPH = _CC_GRAPH,
1173 _CC_ENUM_LOWER = _CC_LOWER,
1174 _CC_ENUM_PRINT = _CC_PRINT,
1175 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1176 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1177 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1178 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1179 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1180 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1181} _char_class_number;
1182#endif
1183
63c61c3f 1184#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1185#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1186
b7d90381
KW
1187#if defined(PERL_IN_UTF8_C) \
1188 || defined(PERL_IN_REGCOMP_C) \
1189 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1190# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1191 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1192 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1193 #error Need to adjust order of swash_property_names[]
1194# endif
1195
b44daeeb
KW
1196/* This is declared static in each of the few files that this is #defined for
1197 * to keep them from being publicly accessible. Hence there is a small amount
1198 * of wasted space */
1199
922e8cb4
KW
1200static const char* const swash_property_names[] = {
1201 "XPosixWord",
1202 "XPosixDigit",
1203 "XPosixAlpha",
1204 "XPosixLower",
1205 "XPosixUpper",
1206 "XPosixPunct",
1207 "XPosixPrint",
1208 "XPosixAlnum",
b0d691b2
KW
1209 "XPosixGraph",
1210 "Cased"
922e8cb4
KW
1211};
1212#endif
1213
6635f04f 1214START_EXTERN_C
96ac0975
NC
1215# ifdef DOINIT
1216EXTCONST U32 PL_charclass[] = {
1217# include "l1_char_class_tab.h"
1218};
1219
1220# else /* ! DOINIT */
1221EXTCONST U32 PL_charclass[];
1222# endif
6635f04f 1223END_EXTERN_C
96ac0975 1224
265c1f46 1225 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1226# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1227
1228 /* For internal core Perl use only: the base macro for defining macros like
1229 * isALPHA */
ff7ecfc3 1230# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1231 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1232
f4cdb42c
KW
1233 /* The mask for the _A versions of the macros; it just adds in the bit for
1234 * ASCII. */
1235# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1236
4650c663
KW
1237 /* For internal core Perl use only: the base macro for defining macros like
1238 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1239 * the ASCII bit are present */
b7d90381
KW
1240# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1241 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1242 == _CC_mask_A(classnum)))
f4cdb42c 1243
f4cdb42c 1244# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1245# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1246# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1247# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1248# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1249# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1250# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1251# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1252# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1253# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1254# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1255# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1256# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1257 */
d95f8b6a 1258# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1259# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1260# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1261# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1262
1263 /* continuation character for legal NAME in \N{NAME} */
1264# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1265
1266# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1267# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1268# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1269# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1270# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1271# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1272# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1273# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1274# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1275# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1276
0852beac
KW
1277# ifdef EBCDIC
1278# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1279# endif
1280
f12c0118
KW
1281 /* Participates in a single-character fold with a character above 255 */
1282# 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)))
1283
1284 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1285# 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 1286
4eeeb416 1287# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1288# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1289 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1290# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1291 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1292# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1293 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1294#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1295
1296 /* If we don't have perl.h, we are compiling a utility program. Below we
1297 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1298 * to it. Most are coded based on first principles. These are written to
74665a89 1299 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
0852beac
KW
1300# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1301# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1302# define isSPACE_A(c) (isBLANK_A(c) \
1303 || (c) == '\n' \
1304 || (c) == '\r' \
1305 || (c) == '\v' \
0852beac 1306 || (c) == '\f')
74665a89
KW
1307 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1308 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1309 * performance (if optimization isn't on), and allow the same code to be
1310 * used for both platform types */
1311# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1312 && ( (c) <= 'i' \
1313 || ((c) >= 'j' && (c) <= 'r') \
1314 || (c) >= 's'))
1315# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1316 && ( (c) <= 'I' \
1317 || ((c) >= 'J' && (c) <= 'R') \
1318 || (c) >= 'S'))
a4d7a999
KW
1319# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1320# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1321# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1322# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
74665a89
KW
1323# define isXDIGIT_A(c) (isDIGIT_A(c) \
1324 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1325 || ((c) <= 'F' && (c) >= 'A'))
74665a89
KW
1326# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1327 || (c) == '#' || (c) == '$' || (c) == '%' \
1328 || (c) == '&' || (c) == '\'' || (c) == '(' \
1329 || (c) == ')' || (c) == '*' || (c) == '+' \
1330 || (c) == ',' || (c) == '.' || (c) == '/' \
1331 || (c) == ':' || (c) == ';' || (c) == '<' \
1332 || (c) == '=' || (c) == '>' || (c) == '?' \
1333 || (c) == '@' || (c) == '[' || (c) == '\\' \
1334 || (c) == ']' || (c) == '^' || (c) == '_' \
1335 || (c) == '`' || (c) == '{' || (c) == '|' \
1336 || (c) == '}' || (c) == '~')
1337# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1338# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1339
0852beac 1340# ifdef EBCDIC
74665a89
KW
1341 /* The below is accurate for the 3 EBCDIC code pages traditionally
1342 * supported by perl. The only difference between them in the controls
1343 * is the position of \n, and that is represented symbolically below */
1344# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1345 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1346 || (c) == '\t' || (c) == '\v' \
1347 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1348 || (c) == 7 /* U+7F DEL */ \
1349 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1350 /* DLE, DC[1-3] */ \
1351 || (c) == 0x18 /* U+18 CAN */ \
1352 || (c) == 0x19 /* U+19 EOM */ \
1353 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1354 || (c) == 0x26 /* U+17 ETB */ \
1355 || (c) == 0x27 /* U+1B ESC */ \
1356 || (c) == 0x2D /* U+05 ENQ */ \
1357 || (c) == 0x2E /* U+06 ACK */ \
1358 || (c) == 0x32 /* U+16 SYN */ \
1359 || (c) == 0x37 /* U+04 EOT */ \
1360 || (c) == 0x3C /* U+14 DC4 */ \
1361 || (c) == 0x3D /* U+15 NAK */ \
1362 || (c) == 0x3F)/* U+1A SUB */
0852beac 1363# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1364# else /* isASCII is already defined for ASCII platforms, so can use that to
1365 define isCNTRL */
1366# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1367# endif
1368
3ffc8c70 1369 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1370 * during debugging, and it seems best to keep them. We may be called
1371 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1372 * do anything anyway, so make it not a problem */
1373# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1374# define NATIVE_TO_LATIN1(ch) (ch)
1375# endif
3ded5eb0
KW
1376# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1377# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1378# define isBLANK_L1(c) (isBLANK_A(c) \
1379 || (FITS_IN_8_BITS(c) \
1380 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1381# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1382# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1383# define isLOWER_L1(c) (isLOWER_A(c) \
1384 || (FITS_IN_8_BITS(c) \
ae683a5f 1385 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1386 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1387 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1388 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1389 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1390# define isPRINT_L1(c) (isPRINT_A(c) \
1391 || (FITS_IN_8_BITS(c) \
1392 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1393# define isPUNCT_L1(c) (isPUNCT_A(c) \
1394 || (FITS_IN_8_BITS(c) \
ae683a5f 1395 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1396 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1397 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1398 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1399 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1400 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1401 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1402# define isSPACE_L1(c) (isSPACE_A(c) \
1403 || (FITS_IN_8_BITS(c) \
ae683a5f 1404 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1405 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1406# define isUPPER_L1(c) (isUPPER_A(c) \
1407 || (FITS_IN_8_BITS(c) \
ae683a5f 1408 && ( NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
3ded5eb0
KW
1409 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1410 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1411# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1412# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1413# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1414 || isBLANK_L1(c) \
1415 || (c) == '-' \
1416 || (c) == '(' \
1417 || (c) == ')')
1418 /* The following are not fully accurate in the above-ASCII range. I (khw)
1419 * don't think it's necessary to be so for the purposes where this gets
1420 * compiled */
1421# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1422# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1423
1424 /* And these aren't accurate at all. They are useful only for above
1425 * Latin1, which utilities and bootstrapping don't deal with */
1426# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1427# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1428# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1429
1430 /* Many of the macros later in this file are defined in terms of these. By
1431 * implementing them with a function, which converts the class number into
1432 * a call to the desired macro, all of the later ones work. However, that
1433 * function won't be actually defined when building a utility program (no
1434 * perl.h), and so a compiler error will be generated if one is attempted
1435 * to be used. And the above-Latin1 code points require Unicode tables to
1436 * be present, something unlikely to be the case when bootstrapping */
1437# define _generic_isCC(c, classnum) \
1438 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1439# define _generic_isCC_A(c, classnum) \
1440 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1441#endif /* End of no perl.h H_PERL */
8a58bdcf 1442
e66b99e9
KW
1443#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1444#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1445#define isASCII_A(c) isASCII(c)
1446#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1447#define isBLANK(c) isBLANK_A(c)
1448#define isCNTRL(c) isCNTRL_A(c)
1449#define isDIGIT(c) isDIGIT_A(c)
1450#define isGRAPH(c) isGRAPH_A(c)
1451#define isIDFIRST(c) isIDFIRST_A(c)
1452#define isLOWER(c) isLOWER_A(c)
1453#define isPRINT(c) isPRINT_A(c)
779cf272 1454#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1455#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1456#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1457#define isPUNCT(c) isPUNCT_A(c)
1458#define isSPACE(c) isSPACE_A(c)
1459#define isUPPER(c) isUPPER_A(c)
1460#define isWORDCHAR(c) isWORDCHAR_A(c)
1461#define isXDIGIT(c) isXDIGIT_A(c)
1462
1463/* ASCII casing. These could also be written as
1464 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1465 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1466 which uses table lookup and mask instead of subtraction. (This would
1467 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1468#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1469#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1470
25200305
KW
1471/* In the ASCII range, these are equivalent to what they're here defined to be.
1472 * But by creating these definitions, other code doesn't have to be aware of
1473 * this detail */
1474#define toFOLD(c) toLOWER(c)
25200305
KW
1475#define toTITLE(c) toUPPER(c)
1476
c753c8d3
KW
1477#define toLOWER_A(c) toLOWER(c)
1478#define toUPPER_A(c) toUPPER(c)
25200305
KW
1479#define toFOLD_A(c) toFOLD(c)
1480#define toTITLE_A(c) toTITLE(c)
1a0901db 1481
4650c663 1482/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1483#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1484 ? (c) \
f4cd282c 1485 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1486#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1487
1a0901db 1488/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1489 * all mapped to one of them, and these need special handling; returns the
1490 * input itself if is out-of-range */
b2bf251f 1491#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1492 ? (c) \
f4cd282c 1493 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1494#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1495
beab9ebe
KW
1496/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1497
1498/* For internal core Perl use only: the base macro for defining macros like
1499 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1500 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1501 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1502 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1503 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1504 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1505 * code points under locale rules, use the macros based on _generic_LC_uvchr
1506 * instead of this. */
beab9ebe
KW
1507#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1508 (! FITS_IN_8_BITS(c) \
1509 ? 0 \
31f05a37
KW
1510 : IN_UTF8_CTYPE_LOCALE \
1511 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1512 : cBOOL(non_utf8))
1513
1514/* For internal core Perl use only: a helper macro for defining macros like
1515 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1516 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1517 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1518 * _generic_LC_base for more info */
1519#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1520 _generic_LC_base(c,utf8_locale_classnum, \
1521 non_utf8_func( (_LC_CAST) (c)))
1522
1523/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1524 * 'c' is the platform's native underscore character */
1525#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1526 _generic_LC_base(c, utf8_locale_classnum, \
1527 (non_utf8_func( (_LC_CAST) (c)) \
1528 || (char)(c) == '_'))
1529
1530/* These next three are also for internal core Perl use only: case-change
1531 * helper macros */
1532#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1533 ? (c) \
31f05a37
KW
1534 : (IN_UTF8_CTYPE_LOCALE) \
1535 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1536 : (cast)function((cast)(c)))
beab9ebe 1537
31f05a37
KW
1538/* Note that the result can be larger than a byte in a UTF-8 locale. It
1539 * returns a single value, so can't adequately return the upper case of LATIN
1540 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1541 * values "SS"); instead it asserts against that under DEBUGGING, and
1542 * otherwise returns its input */
beab9ebe
KW
1543#define _generic_toUPPER_LC(c, function, cast) \
1544 (! FITS_IN_8_BITS(c) \
1545 ? (c) \
31f05a37 1546 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1547 ? (cast)function((cast)(c)) \
31f05a37
KW
1548 : ((((U8)(c)) == MICRO_SIGN) \
1549 ? GREEK_CAPITAL_LETTER_MU \
1550 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1551 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1552 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1553 ? (__ASSERT_(0) (c)) \
1554 : PL_mod_latin1_uc[ (U8) (c) ])))))
1555
1556/* Note that the result can be larger than a byte in a UTF-8 locale. It
1557 * returns a single value, so can't adequately return the fold case of LATIN
1558 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1559 * values "ss"); instead it asserts against that under DEBUGGING, and
1560 * otherwise returns its input */
beab9ebe 1561#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1562 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1563 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1564 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1565 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1566 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1567
84061b6a 1568/* Use the libc versions for these if available. */
f05550c0 1569#if defined(HAS_ISASCII)
84061b6a
KW
1570# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1571#else
1572# define isASCII_LC(c) isASCII(c)
1573#endif
1574
f05550c0 1575#if defined(HAS_ISBLANK)
84061b6a 1576# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1577#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1578# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1579#endif
1580
f05550c0 1581#define _LC_CAST U8
bbce6d69 1582
f05550c0 1583#ifdef WIN32
375f5f06
KW
1584 /* The Windows functions don't bother to follow the POSIX standard, which
1585 * for example says that something can't both be a printable and a control.
1586 * But Windows treats the \t control as a printable, and does such things
1587 * as making superscripts into both digits and punctuation. This tames
1588 * these flaws by assuming that the definitions of both controls and space
1589 * are correct, and then making sure that other definitions don't have
1590 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1591 * Not all possible weirdnesses are checked for, just the ones that were
1592 * detected on actual Microsoft code pages */
1593
b7d90381
KW
1594# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1595# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1596
1597# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1598 && isALPHANUMERIC_LC(c))
1599# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1600 ! isPUNCT_LC(c))
1601# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1602 isALPHANUMERIC_LC(c))
1603# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1604# define isIDFIRST_LC(c) (((c) == '_') \
1605 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1606# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1607# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1608# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1609# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1610# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1611# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1612 && isALPHANUMERIC_LC(c))
f05550c0
BF
1613
1614# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1615# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1616# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1617
1618#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1619 /* For most other platforms */
beab9ebe 1620
f05550c0
BF
1621# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1622# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1623# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1624# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1625# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1626# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1627# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1628# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1629# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1630# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1631# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1632# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1633# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1634
1635
1636# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1637# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1638# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1639
1640#else /* The final fallback position */
1641
b7d90381
KW
1642# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1643# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1644# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1645# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1646# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1647# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1648# define isLOWER_LC(c) (isascii(c) && islower(c))
1649# define isPRINT_LC(c) (isascii(c) && isprint(c))
1650# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1651# define isSPACE_LC(c) (isascii(c) && isspace(c))
1652# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1653# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1654# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1655
1656# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1657# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1658# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1659
f05550c0 1660#endif
55204971 1661
eba68aa0
KW
1662#define isIDCONT(c) isWORDCHAR(c)
1663#define isIDCONT_A(c) isWORDCHAR_A(c)
1664#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1665#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1666#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1667
4650c663 1668/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1669 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1670 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1671 * classes where there is a macro or function 'above_latin1' that takes the
1672 * single argument 'c' and returns the desired value. These exist for those
1673 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1674 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1675 * for classes where that overhead is faster than a direct lookup.
d0da05db 1676 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1677 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1678 * have duplicate checks here, so could create versions of the macros that
1679 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1680
1681/* Note that all ignore 'use bytes' */
1e222e4f
KW
1682#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1683 ? _generic_isCC(c, classnum) \
cd500f2f 1684 : above_latin1(c))
1e222e4f
KW
1685#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
1686 ? _generic_isCC(c, classnum) \
922e8cb4 1687 : _is_uni_FOO(classnum, c))
d0da05db
KW
1688#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1689#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1690#define isASCII_uvchr(c) isASCII(c)
1691#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1692#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1693#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1694#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1695#define isIDCONT_uvchr(c) \
1696 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1697#define isIDFIRST_uvchr(c) \
1698 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
d0da05db
KW
1699#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1700#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1701
1702#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1703#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1704#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1705
1706#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1707#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1708#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1709#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1710
1711#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1712#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1713#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1714#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1715
1716/* For backwards compatibility, even though '_uni' should mean official Unicode
1717 * code points, in Perl it means native for those below 256 */
1718#define isALPHA_uni(c) isALPHA_uvchr(c)
1719#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1720#define isASCII_uni(c) isASCII_uvchr(c)
1721#define isBLANK_uni(c) isBLANK_uvchr(c)
1722#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1723#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1724#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1725#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1726#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1727#define isLOWER_uni(c) isLOWER_uvchr(c)
1728#define isPRINT_uni(c) isPRINT_uvchr(c)
1729#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1730#define isSPACE_uni(c) isSPACE_uvchr(c)
1731#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1732#define isUPPER_uni(c) isUPPER_uvchr(c)
1733#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1734#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1735#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1736#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1737#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1738#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1739#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1740
4650c663
KW
1741/* For internal core Perl use only: the base macros for defining macros like
1742 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1743 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1744 * simple class definitions; the other for more complex. These are like
d0da05db 1745 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1746#define _generic_LC_uvchr(latin1, above_latin1, c) \
1747 (c < 256 ? latin1(c) : above_latin1(c))
1748#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1749 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1750
1751#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1752#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1753 _CC_ALPHANUMERIC, c)
b7d90381
KW
1754#define isASCII_LC_uvchr(c) isASCII_LC(c)
1755#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1756 is_HORIZWS_cp_high, c)
feeab5a9 1757#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1758#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1759#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1760#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1761 _is_uni_perl_idcont, c)
b7d90381 1762#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1763 _is_uni_perl_idstart, c)
1764#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1765#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1766#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1767#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1768#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1769 is_XPERLSPACE_cp_high, c)
cd500f2f 1770#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1771#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1772 _CC_WORDCHAR, c)
b7d90381
KW
1773#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1774 is_XDIGIT_cp_high, c)
e712593e 1775
b7d90381 1776#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1777
4650c663
KW
1778/* For internal core Perl use only: the base macros for defining macros like
1779 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1780 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1781 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1782 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1783 * same representation whether utf8 or not. Note that it assumes that the utf8
1784 * has been validated, and ignores 'use bytes' */
34aeb2e9
KW
1785#define _base_generic_utf8(enum_name, name, p, use_locale ) \
1786 _is_utf8_FOO(CAT2(_CC_, enum_name), \
1787 (const U8 *) p, \
1788 "is" STRINGIFY(name) "_utf8", \
1789 "is" STRINGIFY(name) "_utf8_safe", \
1790 1, use_locale, __FILE__,__LINE__)
1791
1792#define _generic_utf8(name, p) _base_generic_utf8(name, name, p, 0)
da8c1a98
KW
1793
1794/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
1795 * they don't otherwise go out of their way to look for malformed UTF-8. If
1796 * they can return accurate results without knowing if the input is otherwise
1797 * malformed, they do so. For example isASCII is accurate in spite of any
1798 * non-length malformations because it looks only at a single byte. Likewise
1799 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
1800 * variant ones return FALSE. But, if the input has to be well-formed in order
1801 * for the results to be accurate, the macros will test and if malformed will
1802 * call a routine to die
1803 *
1804 * Except for toke.c, the macros do assume that e > p, asserting that on
1805 * DEBUGGING builds. Much code that calls these depends on this being true,
1806 * for other reasons. toke.c is treated specially as using the regular
1807 * assertion breaks it in many ways. All strings that these operate on there
1808 * are supposed to have an extra NUL character at the end, so that *e = \0. A
1809 * bunch of code in toke.c assumes that this is true, so the assertion allows
1810 * for that */
1811#ifdef PERL_IN_TOKE_C
1812# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
1813#else
1814# define _utf8_safe_assert(p,e) ((e) > (p))
1815#endif
1816
1817#define _generic_utf8_safe(classnum, p, e, above_latin1) \
1818 (__ASSERT_(_utf8_safe_assert(p, e)) \
1819 (UTF8_IS_INVARIANT(*(p))) \
1820 ? _generic_isCC(*(p), classnum) \
1821 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1822 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1823 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
1824 classnum) \
1825 : (_force_out_malformed_utf8_message( \
1826 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1827 : above_latin1))
b7d90381
KW
1828/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1829 * 'above_latin1' can be a macro */
da8c1a98
KW
1830#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
1831 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
1832#define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1833 _generic_utf8_safe(classnum, p, e, \
1834 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1835 ? (_force_out_malformed_utf8_message( \
1836 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1837 : above_latin1(p)))
4650c663 1838/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1839 * 'above_latin1' parameter */
da8c1a98
KW
1840#define _generic_swash_utf8_safe(classnum, p, e) \
1841_generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e))
922e8cb4 1842
cc8ab7c0 1843/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1844 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1845 * class is TRUE for. Hence it can skip the tests for this range.
1846 * 'above_latin1' should include its arguments */
da8c1a98
KW
1847#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
1848 (__ASSERT_(_utf8_safe_assert(p, e)) \
1849 (UTF8_IS_INVARIANT(*(p))) \
1850 ? _generic_isCC(*(p), classnum) \
1851 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1852 ? 0 /* Note that doesn't check validity for latin1 */ \
1853 : above_latin1)
1854
84238efa
KW
1855/* NOTE that some of these macros have very similar ones in regcharclass.h.
1856 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1857 * there, differing in name only by an underscore from the one here
1858 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1859 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1860 * points; the regcharclass.h ones are implemented as a series of
1861 * "if-else-if-else ..." */
1862
34aeb2e9
KW
1863#define isALPHA_utf8(p) _generic_utf8(ALPHA, p)
1864#define isALPHANUMERIC_utf8(p) _generic_utf8(ALPHANUMERIC, p)
1865#define isASCII_utf8(p) _generic_utf8(ASCII, p)
1866#define isBLANK_utf8(p) _generic_utf8(BLANK, p)
1867#define isCNTRL_utf8(p) _generic_utf8(CNTRL, p)
1868#define isDIGIT_utf8(p) _generic_utf8(DIGIT, p)
1869#define isGRAPH_utf8(p) _generic_utf8(GRAPH, p)
1870#define isIDCONT_utf8(p) _generic_utf8(IDCONT, p)
1871#define isIDFIRST_utf8(p) _generic_utf8(IDFIRST, p)
1872#define isLOWER_utf8(p) _generic_utf8(LOWER, p)
1873#define isPRINT_utf8(p) _generic_utf8(PRINT, p)
1874#define isPSXSPC_utf8(p) _generic_utf8(PSXSPC, p)
1875#define isPUNCT_utf8(p) _generic_utf8(PUNCT, p)
1876#define isSPACE_utf8(p) _generic_utf8(SPACE, p)
1877#define isUPPER_utf8(p) _generic_utf8(UPPER, p)
1878#define isVERTWS_utf8(p) _generic_utf8(VERTSPACE, p)
1879#define isWORDCHAR_utf8(p) _generic_utf8(WORDCHAR, p)
1880#define isXDIGIT_utf8(p) _generic_utf8(XDIGIT, p)
e8fa43e2 1881
da8c1a98
KW
1882#define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e)
1883#define isALPHANUMERIC_utf8_safe(p, e) \
1884 _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e)
1885#define isASCII_utf8_safe(p, e) \
1886 /* Because ASCII is invariant under utf8, the non-utf8 macro \
1887 * works */ \
1888 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
1889#define isBLANK_utf8_safe(p, e) \
1890 _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
1891
e8fa43e2
KW
1892#ifdef EBCDIC
1893 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1894 * more efficient macro instead of the more general one */
da8c1a98 1895# define isCNTRL_utf8_safe(p, e) \
56d02b8c 1896 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 1897#else
da8c1a98 1898# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
1899#endif
1900
da8c1a98
KW
1901#define isDIGIT_utf8_safe(p, e) \
1902 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
1903 _is_utf8_FOO_with_len(_CC_DIGIT, p, e))
1904#define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e)
1905#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
1906 _is_utf8_perl_idcont_with_len, p, e)
e5dcd934 1907
c11ff943
KW
1908/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1909 * IDFIRST is an alnum. See
c8362b00 1910 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1911 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1912 * This used to be not the XID version, but we decided to go with the more
1913 * modern Unicode definition */
da8c1a98
KW
1914#define isIDFIRST_utf8_safe(p, e) \
1915 _generic_func_utf8_safe(_CC_IDFIRST, \
1916 _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e))
1917
1918#define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e)
1919#define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e)
1920#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
1921#define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e)
1922#define isSPACE_utf8_safe(p, e) \
1923 _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
1924#define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e)
1925#define isVERTWS_utf8_safe(p, e) \
1926 _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
1927#define isWORDCHAR_utf8_safe(p, e) \
1928 _generic_swash_utf8_safe(_CC_WORDCHAR, p, e)
1929#define isXDIGIT_utf8_safe(p, e) \
1930 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
1931 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1932 ? (_force_out_malformed_utf8_message( \
1933 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1934 : is_XDIGIT_high(p)))
a0ed51b3 1935
7d2b61dd 1936#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1937#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1938#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1939#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1940
567b353c 1941/* For internal core use only, subject to change */
607313a1
KW
1942#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f, "", 0)
1943#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f, "", 0)
1944#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f, "", 0)
1945#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f, "", 0)
a1a5ec35
KW
1946
1947#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
1948#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
1949#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
1950#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 1951
4650c663
KW
1952/* For internal core Perl use only: the base macros for defining macros like
1953 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1954 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1955 * 'macro' parameter */
34aeb2e9
KW
1956#define _generic_LC_utf8(name, p) _base_generic_utf8(name, name, p, 1)
1957
1958#define isALPHA_LC_utf8(p) _generic_LC_utf8(ALPHA, p)
1959#define isALPHANUMERIC_LC_utf8(p) _generic_LC_utf8(ALPHANUMERIC, p)
1960#define isASCII_LC_utf8(p) _generic_LC_utf8(ASCII, p)
1961#define isBLANK_LC_utf8(p) _generic_LC_utf8(BLANK, p)
1962#define isCNTRL_LC_utf8(p) _generic_LC_utf8(CNTRL, p)
1963#define isDIGIT_LC_utf8(p) _generic_LC_utf8(DIGIT, p)
1964#define isGRAPH_LC_utf8(p) _generic_LC_utf8(GRAPH, p)
1965#define isIDCONT_LC_utf8(p) _generic_LC_utf8(IDCONT, p)
1966#define isIDFIRST_LC_utf8(p) _generic_LC_utf8(IDFIRST, p)
1967#define isLOWER_LC_utf8(p) _generic_LC_utf8(LOWER, p)
1968#define isPRINT_LC_utf8(p) _generic_LC_utf8(PRINT, p)
1969#define isPSXSPC_LC_utf8(p) _generic_LC_utf8(PSXSPC, p)
1970#define isPUNCT_LC_utf8(p) _generic_LC_utf8(PUNCT, p)
1971#define isSPACE_LC_utf8(p) _generic_LC_utf8(SPACE, p)
1972#define isUPPER_LC_utf8(p) _generic_LC_utf8(UPPER, p)
1973#define isWORDCHAR_LC_utf8(p) _generic_LC_utf8(WORDCHAR, p)
1974#define isXDIGIT_LC_utf8(p) _generic_LC_utf8(XDIGIT, p)
1975
da8c1a98
KW
1976/* For internal core Perl use only: the base macros for defining macros like
1977 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
1978 * point in 'p' is within the 0-255 range, it uses locale rules from the
1979 * passed-in 'macro' parameter */
1980#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
1981 (__ASSERT_(_utf8_safe_assert(p, e)) \
1982 (UTF8_IS_INVARIANT(*(p))) \
1983 ? macro(*(p)) \
1984 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1985 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1986 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
1987 : (_force_out_malformed_utf8_message( \
1988 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1989 : above_latin1))
1990
1991#define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \
1992 _generic_LC_utf8_safe(macro, p, e, \
1993 _is_utf8_FOO_with_len(classnum, p, e))
1994
1995#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
1996 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
1997
1998#define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1999 _generic_LC_utf8_safe(classnum, p, e, \
2000 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2001 ? (_force_out_malformed_utf8_message( \
2002 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2003 : above_latin1(p)))
2004
2005#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2006 _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \
2007 _CC_ALPHANUMERIC, p, e)
2008#define isALPHA_LC_utf8_safe(p, e) \
2009 _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2010#define isASCII_LC_utf8_safe(p, e) \
2011 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2012#define isBLANK_LC_utf8_safe(p, e) \
2013 _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2014#define isCNTRL_LC_utf8_safe(p, e) \
2015 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2016#define isDIGIT_LC_utf8_safe(p, e) \
2017 _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2018#define isGRAPH_LC_utf8_safe(p, e) \
2019 _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2020#define isIDCONT_LC_utf8_safe(p, e) \
2021 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2022 _is_utf8_perl_idcont_with_len, p, e)
2023#define isIDFIRST_LC_utf8_safe(p, e) \
2024 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2025 _is_utf8_perl_idstart_with_len, p, e)
2026#define isLOWER_LC_utf8_safe(p, e) \
2027 _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2028#define isPRINT_LC_utf8_safe(p, e) \
2029 _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2030#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2031#define isPUNCT_LC_utf8_safe(p, e) \
2032 _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2033#define isSPACE_LC_utf8_safe(p, e) \
2034 _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2035#define isUPPER_LC_utf8_safe(p, e) \
2036 _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2037#define isWORDCHAR_LC_utf8_safe(p, e) \
2038 _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2039#define isXDIGIT_LC_utf8_safe(p, e) \
2040 _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2041
fbc19f27
KW
2042/* Macros for backwards compatibility and for completeness when the ASCII and
2043 * Latin1 values are identical */
b7d90381
KW
2044#define isALPHAU(c) isALPHA_L1(c)
2045#define isDIGIT_L1(c) isDIGIT_A(c)
2046#define isOCTAL(c) isOCTAL_A(c)
2047#define isOCTAL_L1(c) isOCTAL_A(c)
2048#define isXDIGIT_L1(c) isXDIGIT_A(c)
2049#define isALNUM(c) isWORDCHAR(c)
2050#define isALNUMU(c) isWORDCHAR_L1(c)
2051#define isALNUM_LC(c) isWORDCHAR_LC(c)
2052#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2053#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
2054#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
2055#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
2056#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2057#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2058#define isALNUMC(c) isALPHANUMERIC(c)
2059#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2060#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2061#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 2062#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 2063#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 2064
2bd1cbf6
KW
2065/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2066 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2067 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2068 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2069 * the outlier from the block that contains the other controls, just like
2070 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2071 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2072 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2073 * etc. */
2bd1cbf6 2074#ifndef EBCDIC
75763b3a 2075# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2076#else
75763b3a
KW
2077# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2078 ((isPRINT_A(c)) \
2079 ? (UNLIKELY((c) == '?') \
2080 ? QUESTION_MARK_CTRL \
2081 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2082 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2083 ? '?' \
2084 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2085#endif
bbce6d69 2086
dea28490
JJ
2087/* Line numbers are unsigned, 32 bits. */
2088typedef U32 line_t;
e5dcd934 2089#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2090
91152fc1
DG
2091/* Helpful alias for version prescan */
2092#define is_LAX_VERSION(a,b) \
2093 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2094
2095#define is_STRICT_VERSION(a,b) \
2096 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2097
2098#define BADVERSION(a,b,c) \
2099 if (b) { \
2100 *b = c; \
2101 } \
2102 return a;
8c52afec 2103
cb27eebd
KW
2104/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2105 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2106 * which is advanced. The input is validated only by an assert() in DEBUGGING
2107 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2108 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2109#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2110 ? (c) \
2111 : ((c) + 9))))
902008b8
KW
2112#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2113 ? (*(s)++) \
2114 : (*(s)++ + 9))))
95a59cab 2115
cb27eebd
KW
2116/* Converts a character known to represent an octal digit (0-7) to its numeric
2117 * value. The input is validated only by an assert() in DEBUGGING builds. In
2118 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2119#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2120
305b8651
KW
2121/* Efficiently returns a boolean as to if two native characters are equivalent
2122 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2123 * the ALPHA in the name is to remind you of that. This is asserted() in
2124 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2125 * works (on valid input) for both non- and UTF-8-encoded bytes.
2126 *
2127 * When one of the inputs is a compile-time constant and gets folded by the
2128 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2129 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2130 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2131 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2132 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2133 * just a single 0, in the bit position where the upper- and lowercase differ.
2134 * */
305b8651
KW
2135#define isALPHA_FOLD_EQ(c1, c2) \
2136 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2137 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2138#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2139
8e84507e 2140/*
ccfc67b7
JH
2141=head1 Memory Management
2142
a02a5408 2143=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2144The XSUB-writer's interface to the C C<malloc> function.
2145
596f7718 2146Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2147
c5008215
JC
2148In 5.9.3, Newx() and friends replace the older New() API, and drops
2149the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2150themselves. This aid has been superseded by a new build option,
d10b4965 2151PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2152there for use in XS modules supporting older perls.
2153
a02a5408 2154=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2155The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2156cast. See also C<L</Newx>>.
954c1994 2157
596f7718 2158Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2159
a02a5408 2160=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2161The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2162memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2163
596f7718 2164Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2165
954c1994
GS
2166=for apidoc Am|void|Renew|void* ptr|int nitems|type
2167The XSUB-writer's interface to the C C<realloc> function.
2168
596f7718 2169Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2170
954c1994
GS
2171=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2172The XSUB-writer's interface to the C C<realloc> function, with
2173cast.
2174
596f7718 2175Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2176
49b8b560 2177=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2178The XSUB-writer's interface to the C C<free> function.
2179
596f7718 2180This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2181
954c1994
GS
2182=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2183The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2184source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2185C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2186
e90e2364 2187=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2188Like C<Move> but returns C<dest>. Useful
72d33970 2189for encouraging compilers to tail-call
e90e2364
NC
2190optimise.
2191
954c1994
GS
2192=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2193The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2194source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2195C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2196
e90e2364
NC
2197=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2198
796b6530 2199Like C<Copy> but returns C<dest>. Useful
72d33970 2200for encouraging compilers to tail-call
e90e2364
NC
2201optimise.
2202
954c1994
GS
2203=for apidoc Am|void|Zero|void* dest|int nitems|type
2204
2205The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2206destination, C<nitems> is the number of items, and C<type> is the type.
2207
e90e2364
NC
2208=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2209
72d33970
FC
2210Like C<Zero> but returns dest. Useful
2211for encouraging compilers to tail-call
e90e2364
NC
2212optimise.
2213
da5d8dbb 2214=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2215This is an architecture-independent macro to copy one structure to another.
954c1994 2216
7e337ee0
JH
2217=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2218
2219Fill up memory with a byte pattern (a byte repeated over and over
2220again) that hopefully catches attempts to access uninitialized memory.
2221
2222=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2223
2224PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2225
1c12ffb4 2226=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2227
2228PoisonWith(0xEF) for catching access to freed memory.
2229
9965345d
JH
2230=for apidoc Am|void|Poison|void* dest|int nitems|type
2231
7e337ee0 2232PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2233
2234=cut */
954c1994 2235
561b68a9
SH
2236/* Maintained for backwards-compatibility only. Use newSV() instead. */
2237#ifndef PERL_CORE
ff06c60c 2238#define NEWSV(x,len) newSV(len)
561b68a9 2239#endif
ff06c60c 2240
19a94d75
JH
2241#define MEM_SIZE_MAX ((MEM_SIZE)~0)
2242
e6bdf523 2243
27d5b266 2244#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2245
2246/* This expression will be constant-folded at compile time. It checks
2247 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2248 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2249 * As well as avoiding the need for a run-time check in some cases, it's
2250 * designed to avoid compiler warnings like:
2251 * comparison is always false due to limited range of data type
73e8ff00
DM
2252 * It's mathematically equivalent to
2253 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2254 */
2255
2256# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
004073ba 2257 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
e6bdf523 2258
88f9f128 2259/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2260 * compiler warnings. We *want* to write the expression as
2261 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2262 * (for some compile-time constant C), but even when the LHS
2263 * constant-folds to false at compile-time, g++ insists on emitting
2264 * warnings about the RHS (e.g. "comparison is always false"), so instead
2265 * we write it as
e6bdf523 2266 *
d98e5cde 2267 * (cond ? n : X) > C
88f9f128 2268 *
d98e5cde
DM
2269 * where X is a constant with X > C always false. Choosing a value for X
2270 * is tricky. If 0, some compilers will complain about 0 > C always being
2271 * false; if 1, Coverity complains when n happens to be the constant value
2272 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2273 * for X and hope that nothing else whines.
e6bdf523
DM
2274 */
2275
2276# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2277 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2278 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2279
2280# define MEM_WRAP_CHECK(n,t) \
2281 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2282 && (croak_memory_wrap(),0))
2283
2284# define MEM_WRAP_CHECK_1(n,t,a) \
2285 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2286 && (Perl_croak_nocontext("%s",(a)),0))
2287
8b44ba4c 2288#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2289
d1decf2b 2290#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
2291#else
2292
410319be
NC
2293#define MEM_WRAP_CHECK(n,t)
2294#define MEM_WRAP_CHECK_1(n,t,a)
2295#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
2296#define MEM_WRAP_CHECK_(n,t)
2297
6fff79ce 2298#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 2299
1936d2a7 2300#endif
8b44ba4c 2301
fe4f188c 2302#ifdef PERL_MEM_LOG
46c6c7e2 2303/*
9f653bb5 2304 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2305 * go through functions, which are handy for debugging breakpoints, but
2306 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2307 * line number, and C function name if available) passed in. This info can
2308 * then be used for logging the calls, for which one gets a sample
73d1d973 2309 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2310 *
46c6c7e2 2311 * Known problems:
94e892a6 2312 * - not all memory allocs get logged, only those
46c6c7e2 2313 * that go through Newx() and derivatives (while all
94e892a6 2314 * Safefrees do get logged)
46c6c7e2
JH
2315 * - __FILE__ and __LINE__ do not work everywhere
2316 * - __func__ or __FUNCTION__ even less so
2317 * - I think more goes on after the perlio frees but
2318 * the thing is that STDERR gets closed (as do all
2319 * the file descriptors)
2320 * - no deeper calling stack than the caller of the Newx()
2321 * or the kind, but do I look like a C reflection/introspection
2322 * utility to you?
2323 * - the function prototypes for the logging functions
2324 * probably should maybe be somewhere else than handy.h
2325 * - one could consider inlining (macrofying) the logging
2326 * for speed, but I am too lazy
2327 * - one could imagine recording the allocations in a hash,
2328 * (keyed by the allocation address?), and maintain that
2329 * through reallocs and frees, but how to do that without
2330 * any News() happening...?
73d1d973 2331 * - lots of -Ddefines to get useful/controllable output
b953482e 2332 * - lots of ENV reads
46c6c7e2
JH
2333 */
2334
0b0ab801 2335# ifdef PERL_CORE
73d1d973 2336# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2337enum mem_log_type {
2338 MLT_ALLOC,
2339 MLT_REALLOC,
d7a2c63c
MHM
2340 MLT_FREE,
2341 MLT_NEW_SV,
2342 MLT_DEL_SV
0b0ab801
MHM
2343};
2344# endif
12754f92 2345# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2346void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2347void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2348# endif
0b0ab801
MHM
2349# endif
2350
fe4f188c
JH
2351#endif
2352
2353#ifdef PERL_MEM_LOG
d1401ee9
MHM
2354#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2355#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2356#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2357#endif
2358
2359#ifndef MEM_LOG_ALLOC
2360#define MEM_LOG_ALLOC(n,t,a) (a)
2361#endif
2362#ifndef MEM_LOG_REALLOC
2363#define MEM_LOG_REALLOC(n,t,v,a) (a)
2364#endif
2365#ifndef MEM_LOG_FREE
2366#define MEM_LOG_FREE(a) (a)
2367#endif
2368
d1401ee9
MHM
2369#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2370#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2371#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2372
2373#ifndef PERL_CORE
a02a5408
JC
2374/* pre 5.9.x compatibility */
2375#define New(x,v,n,t) Newx(v,n,t)
2376#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2377#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2378#endif
a02a5408 2379
ff68c719 2380#define Renew(v,n,t) \
d1401ee9 2381 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2382#define Renewc(v,n,t,c) \
d1401ee9 2383 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2384
2385#ifdef PERL_POISON
2386#define Safefree(d) \
06c0cc96 2387 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2388#else
fe4f188c 2389#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2390#endif
55497cff 2391
45908e4d 2392#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2393
45908e4d
YO
2394
2395#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)))
2396#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)))
2397#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2398
2399#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)))
2400#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 2401#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2402
7e337ee0
JH
2403#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2404#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2405#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2406#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2407
caa674f3
DD
2408#ifdef PERL_POISON
2409# define PERL_POISON_EXPR(x) x
2410#else
2411# define PERL_POISON_EXPR(x)
2412#endif
2413
ff68c719 2414#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2415
c3caa5c3
JH
2416/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2417 * want your zero-based indices to be less than but not equal to).
2418 *
2419 * C_ARRAY_END is one past the last: half-open/half-closed range,
2420 * not last-inclusive range. */
622913ab 2421#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2422#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2423
2cc61e15
DD
2424#ifdef NEED_VA_COPY
2425# ifdef va_copy
2426# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2427# elif defined(__va_copy)
2428# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2429# else
07798b17 2430# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2431# endif
2432#endif
2433
472d47bc
SB
2434/* convenience debug macros */
2435#ifdef USE_ITHREADS
2436#define pTHX_FORMAT "Perl interpreter: 0x%p"
2437#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2438#define pTHX_VALUE_ (void *)my_perl,
2439#define pTHX_VALUE (void *)my_perl
2440#define pTHX__VALUE_ ,(void *)my_perl,
2441#define pTHX__VALUE ,(void *)my_perl
472d47bc 2442#else
3609ea0d 2443#define pTHX_FORMAT
472d47bc 2444#define pTHX__FORMAT
3609ea0d 2445#define pTHX_VALUE_
472d47bc 2446#define pTHX_VALUE
3609ea0d 2447#define pTHX__VALUE_
472d47bc
SB
2448#define pTHX__VALUE
2449#endif /* USE_ITHREADS */
3609ea0d 2450
2acdbac1
NC
2451/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2452 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2453 CPAN::Unpack show any users outside the core. */
2454#ifdef PERL_CORE
dc6e8de0
A
2455# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2456 "Use of " s " is deprecated")
c9680906
A
2457# define deprecate_disappears_in(when,message) \
2458 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2459 message ", and will disappear in Perl " when)
ac641426
A
2460# define deprecate_fatal_in(when,message) \
2461 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2462 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2463#endif
2464
dfff4baf
BF
2465/* Internal macros to deal with gids and uids */
2466#ifdef PERL_CORE
2467
2468# if Uid_t_size > IVSIZE
2469# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2470# define SvUID(sv) SvNV(sv)
07798b17
AC
2471# elif Uid_t_sign <= 0
2472# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2473# define SvUID(sv) SvIV(sv)
dfff4baf 2474# else
07798b17
AC
2475# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2476# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2477# endif /* Uid_t_size */
2478
2479# if Gid_t_size > IVSIZE
2480# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2481# define SvGID(sv) SvNV(sv)
07798b17
AC
2482# elif Gid_t_sign <= 0
2483# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2484# define SvGID(sv) SvIV(sv)
dfff4baf 2485# else
07798b17
AC
2486# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2487# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2488# endif /* Gid_t_size */
2489
2490#endif
2491
6a5bc5ac 2492#endif /* PERL_HANDY_H_ */
9d745869 2493
e9a8c099 2494/*
14d04a33 2495 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2496 */