This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Carp: Fix .t failure to compile in v5.6
[perl5.git] / handy.h
CommitLineData
a0d0e21e 1/* handy.h
a687059c 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
da5d8dbb 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
a687059c 5 *
6e21c824
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8d063cd8 8 *
8d063cd8
LW
9 */
10
4650c663
KW
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
9d745869
KW
14#ifndef HANDY_H /* Guard against nested #inclusion */
15#define HANDY_H
16
85e6fe83 17#if !defined(__STDC__)
378cc40b
LW
18#ifdef NULL
19#undef NULL
20#endif
a687059c 21# define NULL 0
85e6fe83
LW
22#endif
23
24792b8d
NC
24#ifndef PERL_CORE
25# define Null(type) ((type)NULL)
954c1994
GS
26
27/*
ccfc67b7 28=head1 Handy Values
954c1994 29
3609ea0d 30=for apidoc AmU||Nullch
72d33970
FC
31Null character pointer. (No longer available when C<PERL_CORE> is
32defined.)
2307c6d0 33
954c1994 34=for apidoc AmU||Nullsv
72d33970 35Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
36
37=cut
38*/
39
24792b8d
NC
40# define Nullch Null(char*)
41# define Nullfp Null(PerlIO*)
42# define Nullsv Null(SV*)
43#endif
8d063cd8 44
641d3f0b
PP
45#ifdef TRUE
46#undef TRUE
47#endif
48#ifdef FALSE
49#undef FALSE
50#endif
51#define TRUE (1)
52#define FALSE (0)
53
cf3f0ffb
DM
54/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
55 * (compiler permitting) that casting away const-ness will give a warning;
56 * e.g.:
57 *
58 * const SV *sv = ...;
59 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
60 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
61 */
62
b1bc3f34 63#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
f065df04 64# define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
b1bc3f34
NC
65#else
66# define MUTABLE_PTR(p) ((void *) (p))
67#endif
68
a062e10d 69#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 70#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 71#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 72#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 73#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 74#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 75
f789f6a4 76#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
77# include <stdbool.h>
78# ifndef HAS_BOOL
79# define HAS_BOOL 1
80# endif
81#endif
82
8e84507e 83/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
84 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
85 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 86 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
87 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
88 g++ can be identified by __GNUG__.
89 Andy Dougherty February 2000
5d94fbed 90*/
3609ea0d 91#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 92# ifndef PERL_BOOL_AS_CHAR
5d94fbed 93# ifndef HAS_BOOL
c1d22f6b 94# define HAS_BOOL 1
5d94fbed 95# endif
f789f6a4 96# endif
5d94fbed
AD
97#endif
98
99#ifndef HAS_BOOL
f789f6a4
FC
100# ifdef bool
101# undef bool
102# endif
70d5cb32 103# define bool char
c1d22f6b 104# define HAS_BOOL 1
a687059c 105#endif
0d3e774c 106
236415e2
KW
107/* cast-to-bool. A simple (bool) cast may not do the right thing: if bool is
108 * defined as char for example, then the cast from int is
109 * implementation-defined (bool)!!(cbool) in a ternary triggers a bug in xlc on
110 * AIX */
18f5643b 111#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 112
46c6c7e2 113/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
114 * XXX Should really be a Configure probe, with HAS__FUNCTION__
115 * and FUNCTION__ as results.
116 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
46c6c7e2
JH
117#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
118# define FUNCTION__ __func__
119#else
9ea40801 120# if (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
be73ebb0 121 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
46c6c7e2
JH
122# define FUNCTION__ ""
123# else
124# define FUNCTION__ __FUNCTION__ /* Common extension. */
125# endif
126#endif
127
27d4fb96
PP
128/* XXX A note on the perl source internal type system. The
129 original intent was that I32 be *exactly* 32 bits.
130
131 Currently, we only guarantee that I32 is *at least* 32 bits.
132 Specifically, if int is 64 bits, then so is I32. (This is the case
133 for the Cray.) This has the advantage of meshing nicely with
134 standard library calls (where we pass an I32 and the library is
135 expecting an int), but the disadvantage that an I32 is not 32 bits.
136 Andy Dougherty August 1996
24fef2a7 137
dc45a647
MB
138 There is no guarantee that there is *any* integral type with
139 exactly 32 bits. It is perfectly legal for a system to have
140 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 141
dc45a647
MB
142 Similarly, there is no guarantee that I16 and U16 have exactly 16
143 bits.
693762b4 144
8e84507e
NIS
145 For dealing with issues that may arise from various 32/64-bit
146 systems, we will ask Configure to check out
8175356b 147
3609ea0d
JH
148 SHORTSIZE == sizeof(short)
149 INTSIZE == sizeof(int)
150 LONGSIZE == sizeof(long)
dc45a647 151 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 152 PTRSIZE == sizeof(void *)
dc45a647
MB
153 DOUBLESIZE == sizeof(double)
154 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 155
27d4fb96
PP
156*/
157
69512466
JH
158#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
159# include <inttypes.h>
dd0eed91
JH
160# ifdef INT32_MIN_BROKEN
161# undef INT32_MIN
162# define INT32_MIN (-2147483647-1)
163# endif
164# ifdef INT64_MIN_BROKEN
165# undef INT64_MIN
166# define INT64_MIN (-9223372036854775807LL-1)
167# endif
69512466
JH
168#endif
169
8175356b
JH
170typedef I8TYPE I8;
171typedef U8TYPE U8;
172typedef I16TYPE I16;
173typedef U16TYPE U16;
174typedef I32TYPE I32;
175typedef U32TYPE U32;
16d89be8
JH
176
177#ifdef HAS_QUAD
8175356b
JH
178typedef I64TYPE I64;
179typedef U64TYPE U64;
16d89be8 180#endif
8175356b 181
9c39b21c
JH
182/* INT64_C/UINT64_C are C99 from <stdint.h> (so they will not be
183 * available in strict C89 mode), but they are nice, so let's define
184 * them if necessary. */
0d3bd0a9 185#if defined(HAS_QUAD)
9c39b21c
JH
186# undef PeRl_INT64_C
187# undef PeRl_UINT64_C
188/* Prefer the native integer types (int and long) over long long
189 * (which is not C89) and Win32-specific __int64. */
190# if QUADKIND == QUAD_IS_INT && INTSIZE == 8
191# define PeRl_INT64_C(c) (c)
192# define PeRl_UINT64_C(c) CAT2(c,U)
193# endif
194# if QUADKIND == QUAD_IS_LONG && LONGSIZE == 8
195# define PeRl_INT64_C(c) CAT2(c,L)
196# define PeRl_UINT64_C(c) CAT2(c,UL)
197# endif
198# if QUADKIND == QUAD_IS_LONG_LONG && defined(HAS_LONG_LONG)
199# define PeRl_INT64_C(c) CAT2(c,LL)
200# define PeRl_UINT64_C(c) CAT2(c,ULL)
201# endif
202# if QUADKIND == QUAD_IS___INT64
203# define PeRl_INT64_C(c) CAT2(c,I64)
204# define PeRl_UINT64_C(c) CAT2(c,UI64)
205# endif
206# ifndef PeRl_INT64_C
702c92eb
JH
207# define PeRl_INT64_C(c) ((I64)(c)) /* last resort */
208# define PeRl_UINT64_C(c) ((U64)(c))
9c39b21c
JH
209# endif
210/* In OS X the INT64_C/UINT64_C are defined with LL/ULL, which will
211 * not fly with C89-pedantic gcc, so let's undefine them first so that
212 * we can redefine them with our native integer preferring versions. */
8eafc6eb 213# if defined(PERL_DARWIN) && defined(PERL_GCC_PEDANTIC)
9c39b21c
JH
214# undef INT64_C
215# undef UINT64_C
216# endif
217# ifndef INT64_C
218# define INT64_C(c) PeRl_INT64_C(c)
219# endif
220# ifndef UINT64_C
221# define UINT64_C(c) PeRl_UINT64_C(c)
222# endif
e8c95190 223#endif
dc750f2e 224
d8668976 225#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 226
5ff3f7a4
GS
227/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
228 Please search CHAR_MAX in perl.h for further details. */
229#define U8_MAX UINT8_MAX
230#define U8_MIN UINT8_MIN
231
5ff3f7a4
GS
232#define I16_MAX INT16_MAX
233#define I16_MIN INT16_MIN
234#define U16_MAX UINT16_MAX
235#define U16_MIN UINT16_MIN
236
5ff3f7a4
GS
237#define I32_MAX INT32_MAX
238#define I32_MIN INT32_MIN
0e983133
GS
239#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
240# define U32_MAX UINT32_MAX
241#else
242# define U32_MAX 4294967295U
243#endif
5ff3f7a4
GS
244#define U32_MIN UINT32_MIN
245
246#else
247
5c9fa16e
KA
248/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
249 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
250#define U8_MAX PERL_UCHAR_MAX
251#define U8_MIN PERL_UCHAR_MIN
79072805 252
27d4fb96
PP
253#define I16_MAX PERL_SHORT_MAX
254#define I16_MIN PERL_SHORT_MIN
255#define U16_MAX PERL_USHORT_MAX
256#define U16_MIN PERL_USHORT_MIN
79072805 257
c4f23d77 258#if LONGSIZE > 4
27d4fb96
PP
259# define I32_MAX PERL_INT_MAX
260# define I32_MIN PERL_INT_MIN
261# define U32_MAX PERL_UINT_MAX
262# define U32_MIN PERL_UINT_MIN
79072805 263#else
27d4fb96
PP
264# define I32_MAX PERL_LONG_MAX
265# define I32_MIN PERL_LONG_MIN
266# define U32_MAX PERL_ULONG_MAX
267# define U32_MIN PERL_ULONG_MIN
79072805
LW
268#endif
269
5ff3f7a4
GS
270#endif
271
58a9a5d5 272/* log(2) is pretty close to 0.30103, just in case anyone is grepping for it */
fc36a67e
PP
273#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
274#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
275#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
276
88794300 277/* Unused by core; should be deprecated */
ff68c719 278#define Ctl(ch) ((ch) & 037)
8d063cd8 279
4650c663
KW
280/* This is a helper macro to avoid preprocessor issues, replaced by nothing
281 * unless under DEBUGGING, where it expands to an assert of its argument,
282 * followed by a comma (hence the comma operator). If we just used a straight
3e94db23
JH
283 * assert(), we would get a comma with nothing before it when not DEBUGGING.
284 *
285 * We also use empty definition under Coverity since the __ASSERT__
286 * checks often check for things that Really Cannot Happen, and Coverity
287 * detects that and gets all excited. */
288
289#if defined(DEBUGGING) && !defined(__COVERITY__)
0f092d08
KW
290# define __ASSERT_(statement) assert(statement),
291#else
292# define __ASSERT_(statement)
293#endif
294
3fe05580
MHM
295/*
296=head1 SV-Body Allocation
297
298=for apidoc Ama|SV*|newSVpvs|const char* s
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)
a038e571
NC
411#define hv_fetchs(hv,key,lval) \
412 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
413 (lval) ? (HV_FETCH_JUST_SV | HV_FETCH_LVALUE) \
414 : HV_FETCH_JUST_SV, NULL, 0))
415
416#define hv_stores(hv,key,val) \
417 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
418 (HV_FETCH_ISSTORE|HV_FETCH_JUST_SV), (val), 0))
ba3a79e7 419
510966aa
Z
420#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
421
b96d8cd9
NC
422#define get_cvs(str, flags) \
423 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 424
954c1994 425/*
ccfc67b7
JH
426=head1 Miscellaneous Functions
427
954c1994 428=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
429Test two C<NUL>-terminated strings to see if they are different. Returns true
430or false.
954c1994
GS
431
432=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
433Test two C<NUL>-terminated strings to see if they are equal. Returns true or
434false.
954c1994
GS
435
436=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
437Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
438second, C<s2>. Returns true or false.
954c1994
GS
439
440=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
441Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
442equal to the second, C<s2>. Returns true or false.
954c1994
GS
443
444=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
445Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
446the second, C<s2>. Returns true or false.
954c1994
GS
447
448=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
449Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
450or equal to the second, C<s2>. Returns true or false.
954c1994
GS
451
452=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
453Test two C<NUL>-terminated strings to see if they are different. The C<len>
454parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
455wrapper for C<strncmp>).
456
457=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
458Test two C<NUL>-terminated strings to see if they are equal. The C<len>
459parameter indicates the number of bytes to compare. Returns true or false. (A
460wrapper for C<strncmp>).
954c1994 461
bd18bd40
KW
462=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
463Test two buffers (which may contain embedded C<NUL> characters, to see if they
464are equal. The C<len> parameter indicates the number of bytes to compare.
465Returns zero if equal, or non-zero if non-equal.
466
467=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
468Test two buffers (which may contain embedded C<NUL> characters, to see if they
469are not equal. The C<len> parameter indicates the number of bytes to compare.
470Returns zero if non-equal, or non-zero if equal.
471
954c1994
GS
472=cut
473*/
474
8d063cd8
LW
475#define strNE(s1,s2) (strcmp(s1,s2))
476#define strEQ(s1,s2) (!strcmp(s1,s2))
477#define strLT(s1,s2) (strcmp(s1,s2) < 0)
478#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
479#define strGT(s1,s2) (strcmp(s1,s2) > 0)
480#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
481#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
482#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
378cc40b 483
36477c24
PP
484#ifdef HAS_MEMCMP
485# define memNE(s1,s2,l) (memcmp(s1,s2,l))
486# define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
487#else
488# define memNE(s1,s2,l) (bcmp(s1,s2,l))
489# define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
490#endif
491
568a785a 492#define memEQs(s1, l, s2) \
8e6497b6 493 (sizeof(s2)-1 == l && memEQ(s1, ("" s2 ""), (sizeof(s2)-1)))
568a785a
NC
494#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
495
bbce6d69
PP
496/*
497 * Character classes.
498 *
499 * Unfortunately, the introduction of locales means that we
500 * can't trust isupper(), etc. to tell the truth. And when
501 * it comes to /\w+/ with tainting enabled, we *must* be able
502 * to trust our character classes.
503 *
504 * Therefore, the default tests in the text of Perl will be
505 * independent of locale. Any code that wants to depend on
506 * the current locale will use the tests that begin with "lc".
507 */
508
2304df62
AD
509#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
510# ifndef CTYPE256
511# define CTYPE256
512# endif
513#endif
514
954c1994 515/*
ccfc67b7 516
dcccc8ff 517=head1 Character classification
243effed
KW
518This section is about functions (really macros) that classify characters
519into types, such as punctuation versus alphabetic, etc. Most of these are
520analogous to regular expression character classes. (See
521L<perlrecharclass/POSIX Character Classes>.) There are several variants for
522each class. (Not all macros have all variants; each item below lists the
523ones valid for it.) None are affected by C<use bytes>, and only the ones
524with C<LC> in the name are affected by the current locale.
525
526The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
527C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
528represented by that octet is (or on non-ASCII platforms, corresponds to) an
529ASCII character in the named class based on platform, Unicode, and Perl rules.
530If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 531
6aff1f14
KW
532Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
533with no suffix C<"_A">.
4b9734bf
KW
534
535Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
536onto the platform. That is, the code points that are ASCII are unaffected,
537since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
538as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
539true when called with the code point 0xDF, which is a word character in both
4650c663 540ASCII and EBCDIC (though it represents different characters in each).
243effed
KW
541
542Variant C<isFOO_uni> is like the C<isFOO_L1> variant, but accepts any UV code
543point as input. If the code point is larger than 255, Unicode rules are used
544to determine if it is in the character class. For example,
6aff1f14
KW
545C<isWORDCHAR_uni(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
546WITH MACRON in Unicode, and is a word character.
243effed
KW
547
548Variant C<isFOO_utf8> is like C<isFOO_uni>, but the input is a pointer to a
dc6b0978
KW
549(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
550possibly containing embedded C<NUL> characters). The classification of just
551the first (possibly multi-byte) character in the string is tested.
243effed 552
31f05a37
KW
553Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
554result is based on the current locale, which is what C<LC> in the name stands
555for. If Perl can determine that the current locale is a UTF-8 locale, it uses
556the published Unicode rules; otherwise, it uses the C library function that
557gives the named classification. For example, C<isDIGIT_LC()> when not in a
558UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
559returned if the input won't fit into an octet. On some platforms where the C
560library function is known to be defective, Perl changes its result to follow
561the POSIX standard's rules.
243effed
KW
562
563Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
564returns the same as C<isFOO_LC> for input code points less than 256, and
565returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
566
b7d90381 567Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer
dc6b0978
KW
568to a (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
569possibly containing embedded C<NUL> characters). The classification of just
570the first (possibly multi-byte) character in the string is tested.
ccfc67b7 571
954c1994 572=for apidoc Am|bool|isALPHA|char ch
2787a470 573Returns a boolean indicating whether the specified character is an
6aff1f14 574alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
575See the L<top of this section|/Character classification> for an explanation of
576variants
e5ad6aba 577C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uni>, C<isALPHA_utf8>, C<isALPHA_LC>,
243effed 578C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
8a58bdcf 579
15861f94
KW
580=for apidoc Am|bool|isALPHANUMERIC|char ch
581Returns a boolean indicating whether the specified character is a either an
6aff1f14 582alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
583See the L<top of this section|/Character classification> for an explanation of
584variants
15861f94 585C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uni>,
e5ad6aba 586C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
15861f94
KW
587C<isALPHANUMERIC_LC_utf8>.
588
8a58bdcf
KW
589=for apidoc Am|bool|isASCII|char ch
590Returns a boolean indicating whether the specified character is one of the 128
243effed 591characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 592On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
593character corresponds to an ASCII character. Variants C<isASCII_A()> and
594C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
595See the L<top of this section|/Character classification> for an explanation of
596variants
243effed
KW
597C<isASCII_uni>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
598C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
599library routine C<isascii()>. In these cases, the variants whose names contain
600C<LC> are the same as the corresponding ones without.
601
d98532ea
KW
602Also note, that because all ASCII characters are UTF-8 invariant (meaning they
603have the exact same representation (always a single byte) whether encoded in
604UTF-8 or not), C<isASCII> will give the correct results when called with any
605byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
606work properly on any string encoded or not in UTF-8.
607
243effed
KW
608=for apidoc Am|bool|isBLANK|char ch
609Returns a boolean indicating whether the specified character is a
6aff1f14 610character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
611See the L<top of this section|/Character classification> for an explanation of
612variants
e5ad6aba 613C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uni>, C<isBLANK_utf8>, C<isBLANK_LC>,
243effed
KW
614C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
615platforms do not have the C library routine C<isblank()>. In these cases, the
616variants whose names contain C<LC> are the same as the corresponding ones
617without.
618
619=for apidoc Am|bool|isCNTRL|char ch
620Returns a boolean indicating whether the specified character is a
6aff1f14 621control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
622See the L<top of this section|/Character classification> for an explanation of
623variants
e5ad6aba 624C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uni>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
6aff1f14
KW
625C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
626On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 627
954c1994 628=for apidoc Am|bool|isDIGIT|char ch
2787a470 629Returns a boolean indicating whether the specified character is a
6aff1f14 630digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 631Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
632See the L<top of this section|/Character classification> for an explanation of
633variants
e5ad6aba 634C<isDIGIT_uni>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
243effed
KW
635C<isDIGIT_LC_utf8>.
636
637=for apidoc Am|bool|isGRAPH|char ch
638Returns a boolean indicating whether the specified character is a
6aff1f14 639graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff
KW
640See the L<top of this section|/Character classification> for an explanation of
641variants
e5ad6aba 642C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uni>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
243effed 643C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
954c1994 644
0c82b6df 645=for apidoc Am|bool|isLOWER|char ch
2787a470 646Returns a boolean indicating whether the specified character is a
6aff1f14 647lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
648See the L<top of this section|/Character classification> for an explanation of
649variants
e5ad6aba 650C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uni>, C<isLOWER_utf8>, C<isLOWER_LC>,
243effed 651C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
0c82b6df 652
c99e91e9 653=for apidoc Am|bool|isOCTAL|char ch
2787a470 654Returns a boolean indicating whether the specified character is an
6aff1f14 655octal digit, [0-7].
243effed
KW
656The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
657C<isOCTAL>.
658
659=for apidoc Am|bool|isPUNCT|char ch
660Returns a boolean indicating whether the specified character is a
6aff1f14
KW
661punctuation character, analogous to C<m/[[:punct:]]/>.
662Note that the definition of what is punctuation isn't as
243effed
KW
663straightforward as one might desire. See L<perlrecharclass/POSIX Character
664Classes> for details.
dcccc8ff
KW
665See the L<top of this section|/Character classification> for an explanation of
666variants
e5ad6aba 667C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uni>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
243effed 668C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
c99e91e9 669
0c82b6df 670=for apidoc Am|bool|isSPACE|char ch
2787a470 671Returns a boolean indicating whether the specified character is a
6aff1f14 672whitespace character. This is analogous
398d098a 673to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 674this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
675locale forms of this macro (the ones with C<LC> in their names) matched
676precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
677in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
678(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
679See the L<top of this section|/Character classification> for an explanation of
680variants
e5ad6aba 681C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uni>, C<isSPACE_utf8>, C<isSPACE_LC>,
243effed 682C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
0c82b6df 683
398d098a
KW
684=for apidoc Am|bool|isPSXSPC|char ch
685(short for Posix Space)
779cf272
KW
686Starting in 5.18, this is identical in all its forms to the
687corresponding C<isSPACE()> macros.
398d098a
KW
688The locale forms of this macro are identical to their corresponding
689C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
690non-locale forms differ from their C<isSPACE()> forms only in that the
691C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
692Otherwise they are identical. Thus this macro is analogous to what
693C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 694See the L<top of this section|/Character classification> for an explanation of
b7d90381
KW
695variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uni>, C<isPSXSPC_utf8>,
696C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
398d098a 697
954c1994 698=for apidoc Am|bool|isUPPER|char ch
2787a470 699Returns a boolean indicating whether the specified character is an
6aff1f14 700uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff
KW
701See the L<top of this section|/Character classification> for an explanation of
702variants
e5ad6aba 703C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uni>, C<isUPPER_utf8>, C<isUPPER_LC>,
243effed 704C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
954c1994 705
243effed 706=for apidoc Am|bool|isPRINT|char ch
8eea39dd 707Returns a boolean indicating whether the specified character is a
6aff1f14 708printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
709See the L<top of this section|/Character classification> for an explanation of
710variants
e5ad6aba 711C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uni>, C<isPRINT_utf8>, C<isPRINT_LC>,
243effed
KW
712C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
713
714=for apidoc Am|bool|isWORDCHAR|char ch
715Returns a boolean indicating whether the specified character is a character
716that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
717in a regular expression. A word character is an alphabetic character, a
718decimal digit, a connecting punctuation character (such as an underscore), or
719a "mark" character that attaches to one of those (like some sort of accent).
720C<isALNUM()> is a synonym provided for backward compatibility, even though a
721word character includes more than the standard C language meaning of
722alphanumeric.
dcccc8ff
KW
723See the L<top of this section|/Character classification> for an explanation of
724variants
1a83413c
KW
725C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uni>, and C<isWORDCHAR_utf8>.
726C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
727described there, but additionally include the platform's native underscore.
8a58bdcf
KW
728
729=for apidoc Am|bool|isXDIGIT|char ch
730Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
731digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
732and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
733See the L<top of this section|/Character classification> for an explanation of
734variants
243effed
KW
735C<isXDIGIT_uni>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
736C<isXDIGIT_LC_utf8>.
737
3c3ecf18
KW
738=for apidoc Am|bool|isIDFIRST|char ch
739Returns a boolean indicating whether the specified character can be the first
740character of an identifier. This is very close to, but not quite the same as
741the official Unicode property C<XID_Start>. The difference is that this
742returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
743See the L<top of this section|/Character classification> for an explanation of
744variants
3c3ecf18
KW
745C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uni>, C<isIDFIRST_utf8>,
746C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
747
748=for apidoc Am|bool|isIDCONT|char ch
749Returns a boolean indicating whether the specified character can be the
750second or succeeding character of an identifier. This is very close to, but
751not quite the same as the official Unicode property C<XID_Continue>. The
752difference is that this returns true only if the input character also matches
dcccc8ff
KW
753L</isWORDCHAR>. See the L<top of this section|/Character classification> for
754an
3c3ecf18
KW
755explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uni>,
756C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
757C<isIDCONT_LC_utf8>.
758
243effed 759=head1 Miscellaneous Functions
8eea39dd 760
95a59cab 761=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 762Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
763Behaviour is only well defined when isXDIGIT(*str) is true.
764
e7c1e6c1
KW
765=head1 Character case changing
766
1f607577
KW
767=for apidoc Am|U8|toUPPER|U8 ch
768Converts the specified character to uppercase. If the input is anything but an
769ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 770C<toUPPER_A> is equivalent.
954c1994 771
1f607577
KW
772=for apidoc Am|UV|toUPPER_uni|UV cp|U8* s|STRLEN* lenp
773Converts the Unicode code point C<cp> to its uppercase version, and
774stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
775that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
776bytes since the uppercase version may be longer than the original character.
777
778The first code point of the uppercased version is returned
779(but note, as explained just above, that there may be more.)
780
781=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
782Converts the UTF-8 encoded character at C<p> to its uppercase version, and
783stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
784that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
785bytes since the uppercase version may be longer than the original character.
786
787The first code point of the uppercased version is returned
788(but note, as explained just above, that there may be more.)
789
790The input character at C<p> is assumed to be well-formed.
791
25200305
KW
792=for apidoc Am|U8|toFOLD|U8 ch
793Converts the specified character to foldcase. If the input is anything but an
794ASCII uppercase character, that input character itself is returned. Variant
795C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
796Latin1 range, as the full generality of L</toFOLD_uni> is needed there.)
797
1f607577
KW
798=for apidoc Am|UV|toFOLD_uni|UV cp|U8* s|STRLEN* lenp
799Converts the Unicode code point C<cp> to its foldcase version, and
800stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
801that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
802bytes since the foldcase version may be longer than the original character.
803
804The first code point of the foldcased version is returned
805(but note, as explained just above, that there may be more.)
806
807=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
808Converts the UTF-8 encoded character at C<p> to its foldcase version, and
809stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
810that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
811bytes since the foldcase version may be longer than the original character.
812
813The first code point of the foldcased version is returned
814(but note, as explained just above, that there may be more.)
815
816The input character at C<p> is assumed to be well-formed.
817
818=for apidoc Am|U8|toLOWER|U8 ch
819Converts the specified character to lowercase. If the input is anything but an
820ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 821C<toLOWER_A> is equivalent.
954c1994 822
1f607577 823=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
824Converts the specified Latin1 character to lowercase. The results are
825undefined if the input doesn't fit in a byte.
1f607577
KW
826
827=for apidoc Am|U8|toLOWER_LC|U8 ch
828Converts the specified character to lowercase using the current locale's rules,
829if possible; otherwise returns the input character itself.
830
831=for apidoc Am|UV|toLOWER_uni|UV cp|U8* s|STRLEN* lenp
832Converts the Unicode code point C<cp> to its lowercase version, and
833stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
834that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
835bytes since the lowercase version may be longer than the original character.
836
837The first code point of the lowercased version is returned
838(but note, as explained just above, that there may be more.)
839
840=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
841Converts the UTF-8 encoded character at C<p> to its lowercase version, and
842stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
843that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
844bytes since the lowercase version may be longer than the original character.
845
846The first code point of the lowercased version is returned
847(but note, as explained just above, that there may be more.)
848
849The input character at C<p> is assumed to be well-formed.
850
25200305
KW
851=for apidoc Am|U8|toTITLE|U8 ch
852Converts the specified character to titlecase. If the input is anything but an
853ASCII lowercase character, that input character itself is returned. Variant
b7d90381
KW
854C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
855range, as the full generality of L</toTITLE_uni> is needed there. Titlecase is
856not a concept used in locale handling, so there is no functionality for that.)
25200305 857
1f607577
KW
858=for apidoc Am|UV|toTITLE_uni|UV cp|U8* s|STRLEN* lenp
859Converts the Unicode code point C<cp> to its titlecase version, and
860stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
861that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
862bytes since the titlecase version may be longer than the original character.
863
864The first code point of the titlecased version is returned
865(but note, as explained just above, that there may be more.)
866
867=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
868Converts the UTF-8 encoded character at C<p> to its titlecase version, and
869stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
870that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
871bytes since the titlecase version may be longer than the original character.
872
873The first code point of the titlecased version is returned
874(but note, as explained just above, that there may be more.)
875
876The input character at C<p> is assumed to be well-formed.
877
954c1994 878=cut
353c9b6f 879
1f607577 880XXX Still undocumented isVERTWS_uni and _utf8; it's unclear what their names
25200305 881really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 882
8a58bdcf 883Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
884patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
885
954c1994
GS
886*/
887
7c062697
KW
888/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
889 * is known only in the perl core, and this macro can be called from outside
890 * that */
891#ifdef HAS_QUAD
892# define WIDEST_UTYPE U64TYPE
893#else
894# define WIDEST_UTYPE U32
895#endif
896
3912bc88
KW
897/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
898 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
899 * bits of information are lost even on a 64-bit machine, but to get the
900 * compiler to optimize it out if possible. This is because Configure makes
901 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
902 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
903 *
904 * For Coverity, be always true, because otherwise Coverity thinks
905 * it finds several expressions that are always true, independent
906 * of operands. Well, they are, but that is kind of the point.
220c71bf 907 */
7e75d1a1 908#ifndef __COVERITY__
220c71bf 909#define FITS_IN_8_BITS(c) ((sizeof(c) == 1) || !(((WIDEST_UTYPE)(c)) & ~0xFF))
7e75d1a1
JH
910#else
911#define FITS_IN_8_BITS(c) (1)
912#endif
cf301eb7 913
41f43cc2 914#ifdef EBCDIC
b6340bd0 915# ifndef _ALL_SOURCE
0852beac
KW
916 /* The native libc isascii() et.al. functions return the wrong results
917 * on at least z/OS unless this is defined. */
b6340bd0
KW
918# error _ALL_SOURCE should probably be defined
919# endif
41f43cc2 920#else
0852beac
KW
921 /* There is a simple definition of ASCII for ASCII platforms. But the
922 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59
LM
923 * other macros below.
924 * The '| 0' part ensures that c is an integer (and not e.g. a pointer) */
925# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
926#endif
927
c2da0b36
KW
928/* The lower 3 bits in both the ASCII and EBCDIC representations of '0' are 0,
929 * and the 8 possible permutations of those bits exactly comprise the 8 octal
930 * digits */
3d33b3e3 931#define isOCTAL_A(c) cBOOL(FITS_IN_8_BITS(c) && (0xF8 & (c)) == '0')
c2da0b36 932
9fb1bf9d 933#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 934
a500dc72
KW
935/* Character class numbers. For internal core Perl use only. The ones less
936 * than 32 are used in PL_charclass[] and the ones up through the one that
937 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
938 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
939 * their actual definitions are here. If that file has a name not used here,
940 * it won't compile.
1709d539
KW
941 *
942 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
943 * frequency of their use. This gives a slight edge to exiting a loop earlier
944 * (in reginclass() in regexec.c) */
1709d539
KW
945# define _CC_WORDCHAR 0 /* \w and [:word:] */
946# define _CC_DIGIT 1 /* \d and [:digit:] */
947# define _CC_ALPHA 2 /* [:alpha:] */
948# define _CC_LOWER 3 /* [:lower:] */
949# define _CC_UPPER 4 /* [:upper:] */
950# define _CC_PUNCT 5 /* [:punct:] */
951# define _CC_PRINT 6 /* [:print:] */
15861f94 952# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 953# define _CC_GRAPH 8 /* [:graph:] */
b0d691b2 954# define _CC_CASED 9 /* [:lower:] and [:upper:] under /i */
1709d539 955
b0d691b2 956#define _FIRST_NON_SWASH_CC 10
1709d539
KW
957/* The character classes above are implemented with swashes. The second group
958 * (just below) contains the ones implemented without. These are also sorted
959 * in rough order of the frequency of their use, except that \v should be last,
960 * as it isn't a real Posix character class, and some (small) inefficiencies in
961 * regular expression handling would be introduced by putting it in the middle
962 * of those that are. Also, cntrl and ascii come after the others as it may be
963 * useful to group these which have no members that match above Latin1, (or
964 * above ASCII in the latter case) */
965
779cf272 966# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
967# define _CC_BLANK 11 /* [:blank:] */
968# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
969# define _CC_CNTRL 13 /* [:cntrl:] */
970# define _CC_ASCII 14 /* [:ascii:] */
971# define _CC_VERTSPACE 15 /* \v */
1709d539 972
a0947d7b
KW
973# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
974
1709d539 975/* The members of the third group below do not need to be coordinated with data
3ffc8c70 976 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
977# define _CC_IDFIRST 16
978# define _CC_CHARNAME_CONT 17
979# define _CC_NONLATIN1_FOLD 18
980# define _CC_NONLATIN1_SIMPLE_FOLD 19
981# define _CC_QUOTEMETA 20
982# define _CC_NON_FINAL_FOLD 21
983# define _CC_IS_IN_SOME_FOLD 22
984# define _CC_MNEMONIC_CNTRL 23
073c22b3
KW
985
986/* This next group is only used on EBCDIC platforms, so theoretically could be
987 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
988# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
989# define _CC_UTF8_IS_START 29
990# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
991# define _CC_UTF8_IS_CONTINUATION 31
992/* Unused: 24-27
f4cdb42c
KW
993 * If more bits are needed, one could add a second word for non-64bit
994 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
995 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
996 * is used only for optimization (as of this writing), and differs in the
997 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 998 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
999 * would just cause /i regexes which match them to run less efficiently.
1000 * Similarly the EBCDIC-only bits are used just for speed, and could be
1001 * replaced by other means */
96ac0975 1002
3a371f2f
KW
1003#if defined(PERL_CORE) || defined(PERL_EXT)
1004/* An enum version of the character class numbers, to help compilers
1005 * optimize */
1006typedef enum {
3a371f2f 1007 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1008 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1009 _CC_ENUM_ASCII = _CC_ASCII,
1010 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1011 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1012 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1013 _CC_ENUM_DIGIT = _CC_DIGIT,
1014 _CC_ENUM_GRAPH = _CC_GRAPH,
1015 _CC_ENUM_LOWER = _CC_LOWER,
1016 _CC_ENUM_PRINT = _CC_PRINT,
1017 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1018 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1019 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1020 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1021 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1022 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1023} _char_class_number;
1024#endif
1025
63c61c3f 1026#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1027#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1028
b7d90381
KW
1029#if defined(PERL_IN_UTF8_C) \
1030 || defined(PERL_IN_REGCOMP_C) \
1031 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1032# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1033 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1034 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1035 #error Need to adjust order of swash_property_names[]
1036# endif
1037
b44daeeb
KW
1038/* This is declared static in each of the few files that this is #defined for
1039 * to keep them from being publicly accessible. Hence there is a small amount
1040 * of wasted space */
1041
922e8cb4
KW
1042static const char* const swash_property_names[] = {
1043 "XPosixWord",
1044 "XPosixDigit",
1045 "XPosixAlpha",
1046 "XPosixLower",
1047 "XPosixUpper",
1048 "XPosixPunct",
1049 "XPosixPrint",
1050 "XPosixAlnum",
b0d691b2
KW
1051 "XPosixGraph",
1052 "Cased"
922e8cb4
KW
1053};
1054#endif
1055
6635f04f 1056START_EXTERN_C
96ac0975
NC
1057# ifdef DOINIT
1058EXTCONST U32 PL_charclass[] = {
1059# include "l1_char_class_tab.h"
1060};
1061
1062# else /* ! DOINIT */
1063EXTCONST U32 PL_charclass[];
1064# endif
6635f04f 1065END_EXTERN_C
96ac0975 1066
265c1f46 1067 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1068# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1069
1070 /* For internal core Perl use only: the base macro for defining macros like
1071 * isALPHA */
ff7ecfc3 1072# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1073 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1074
f4cdb42c
KW
1075 /* The mask for the _A versions of the macros; it just adds in the bit for
1076 * ASCII. */
1077# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1078
4650c663
KW
1079 /* For internal core Perl use only: the base macro for defining macros like
1080 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1081 * the ASCII bit are present */
b7d90381
KW
1082# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1083 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1084 == _CC_mask_A(classnum)))
f4cdb42c 1085
f4cdb42c 1086# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1087# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1088# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1089# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1090# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1091# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1092# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1093# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1094# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1095# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1096# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1097# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1098# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1099 */
d95f8b6a 1100# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1101# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1102# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1103# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1104
1105 /* continuation character for legal NAME in \N{NAME} */
1106# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1107
1108# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1109# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1110# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1111# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1112# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1113# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1114# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1115# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1116# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1117# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1118
0852beac
KW
1119# ifdef EBCDIC
1120# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1121# endif
1122
f12c0118
KW
1123 /* Participates in a single-character fold with a character above 255 */
1124# 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)))
1125
1126 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1127# 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 1128
4eeeb416 1129# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1130# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1131 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1132# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1133 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1134# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1135 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1136#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1137
1138 /* If we don't have perl.h, we are compiling a utility program. Below we
1139 * hard-code various macro definitions that wouldn't otherwise be available
0852beac
KW
1140 * to it. Most are coded based on first principals. First some ones common
1141 * to both ASCII and EBCDIC */
1142# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1143# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1144# define isSPACE_A(c) (isBLANK_A(c) \
1145 || (c) == '\n' \
1146 || (c) == '\r' \
1147 || (c) == '\v' \
1148 || (c) == '\f')
1149# ifdef EBCDIC /* There are gaps between 'i' and 'j'; 'r' and 's'. Same
1150 for uppercase. This is ordered to exclude most things
1151 early */
1152# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1153 && ((c) <= 'i' \
1154 || ((c) >= 'j' && (c) <= 'r') \
1155 || (c) >= 's'))
1156# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1157 && ((c) <= 'I' \
1158 || ((c) >= 'J' && (c) <= 'R') \
1159 || (c) >= 'S'))
1160# else /* ASCII platform. */
9fb1bf9d 1161# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z')
2181da29 1162# define isUPPER_A(c) ((c) <= 'Z' && (c) >= 'A')
0852beac
KW
1163# endif
1164
1165 /* Some more ASCII, non-ASCII common definitions */
a4d7a999
KW
1166# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1167# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1168# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac
KW
1169# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1170# define isXDIGIT_A(c) (isDIGIT_A(c) \
1171 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1172 || ((c) <= 'F' && (c) >= 'A'))
3ded5eb0 1173
0852beac
KW
1174# ifdef EBCDIC
1175# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1176 || (c) == '#' || (c) == '$' || (c) == '%' \
1177 || (c) == '&' || (c) == '\'' || (c) == '(' \
1178 || (c) == ')' || (c) == '*' || (c) == '+' \
1179 || (c) == ',' || (c) == '.' || (c) == '/' \
1180 || (c) == ':' || (c) == ';' || (c) == '<' \
1181 || (c) == '=' || (c) == '>' || (c) == '?' \
1182 || (c) == '@' || (c) == '[' || (c) == '\\' \
1183 || (c) == ']' || (c) == '^' || (c) == '_' \
1184 || (c) == '`' || (c) == '{' || (c) == '|' \
1185 || (c) == '}' || (c) == '~')
1186# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1187# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1188
1189# ifdef QUESTION_MARK_CTRL
1190# define _isQMC(c) ((c) == QUESTION_MARK_CTRL)
1191# else
1192# define _isQMC(c) 0
1193# endif
1194
1195 /* I (khw) can't think of a way to define all the ASCII controls
1196 * without resorting to a libc (locale-sensitive) call. But we know
1197 * that all controls but the question-mark one are in the range 0-0x3f.
1198 * This makes sure that all the controls that have names are included,
1199 * and all controls that are also considered ASCII in the locale. This
1200 * may include more or fewer than what it actually should, but the
1201 * wrong ones are less-important controls, so likely won't impact
1202 * things (keep in mind that this is compiled only if perl.h isn't
1203 * available). The question mark control is included if available */
1204# define isCNTRL_A(c) (((c) < 0x40 && isascii(c)) \
1205 || (c) == '\0' || (c) == '\a' || (c) == '\b' \
1206 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1207 || (c) == '\t' || (c) == '\v' || _isQMC(c))
1208
1209# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1210# else /* ASCII platform; things are simpler, and isASCII has already
1211 been defined */
1212# define isGRAPH_A(c) (((c) > ' ' && (c) < 127))
1213# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1214# define isPUNCT_A(c) (isGRAPH_A(c) && (! isALPHANUMERIC_A(c)))
1215# define isCNTRL_A(c) (isASCII(c) && (! isPRINT_A(c)))
1216# endif
1217
3ffc8c70 1218 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1219 * during debugging, and it seems best to keep them. We may be called
1220 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1221 * do anything anyway, so make it not a problem */
1222# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1223# define NATIVE_TO_LATIN1(ch) (ch)
1224# endif
3ded5eb0
KW
1225# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1226# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1227# define isBLANK_L1(c) (isBLANK_A(c) \
1228 || (FITS_IN_8_BITS(c) \
1229 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1230# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1231# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1232# define isLOWER_L1(c) (isLOWER_A(c) \
1233 || (FITS_IN_8_BITS(c) \
1234 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1235 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1236 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1237 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1238 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1239# define isPRINT_L1(c) (isPRINT_A(c) \
1240 || (FITS_IN_8_BITS(c) \
1241 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1242# define isPUNCT_L1(c) (isPUNCT_A(c) \
1243 || (FITS_IN_8_BITS(c) \
1244 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1245 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1246 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1247 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1248 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1249 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1250 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1251# define isSPACE_L1(c) (isSPACE_A(c) \
1252 || (FITS_IN_8_BITS(c) \
1253 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1254 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1255# define isUPPER_L1(c) (isUPPER_A(c) \
1256 || (FITS_IN_8_BITS(c) \
1257 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1258 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1259 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1260# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1261# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1262# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1263 || isBLANK_L1(c) \
1264 || (c) == '-' \
1265 || (c) == '(' \
1266 || (c) == ')')
1267 /* The following are not fully accurate in the above-ASCII range. I (khw)
1268 * don't think it's necessary to be so for the purposes where this gets
1269 * compiled */
1270# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1271# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1272
1273 /* And these aren't accurate at all. They are useful only for above
1274 * Latin1, which utilities and bootstrapping don't deal with */
1275# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1276# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1277# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1278
1279 /* Many of the macros later in this file are defined in terms of these. By
1280 * implementing them with a function, which converts the class number into
1281 * a call to the desired macro, all of the later ones work. However, that
1282 * function won't be actually defined when building a utility program (no
1283 * perl.h), and so a compiler error will be generated if one is attempted
1284 * to be used. And the above-Latin1 code points require Unicode tables to
1285 * be present, something unlikely to be the case when bootstrapping */
1286# define _generic_isCC(c, classnum) \
1287 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1288# define _generic_isCC_A(c, classnum) \
1289 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1290#endif /* End of no perl.h H_PERL */
8a58bdcf 1291
e66b99e9
KW
1292#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1293#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1294#define isASCII_A(c) isASCII(c)
1295#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1296#define isBLANK(c) isBLANK_A(c)
1297#define isCNTRL(c) isCNTRL_A(c)
1298#define isDIGIT(c) isDIGIT_A(c)
1299#define isGRAPH(c) isGRAPH_A(c)
1300#define isIDFIRST(c) isIDFIRST_A(c)
1301#define isLOWER(c) isLOWER_A(c)
1302#define isPRINT(c) isPRINT_A(c)
779cf272 1303#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1304#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1305#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1306#define isPUNCT(c) isPUNCT_A(c)
1307#define isSPACE(c) isSPACE_A(c)
1308#define isUPPER(c) isUPPER_A(c)
1309#define isWORDCHAR(c) isWORDCHAR_A(c)
1310#define isXDIGIT(c) isXDIGIT_A(c)
1311
1312/* ASCII casing. These could also be written as
1313 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1314 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1315 which uses table lookup and mask instead of subtraction. (This would
1316 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1317#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1318#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1319
25200305
KW
1320/* In the ASCII range, these are equivalent to what they're here defined to be.
1321 * But by creating these definitions, other code doesn't have to be aware of
1322 * this detail */
1323#define toFOLD(c) toLOWER(c)
25200305
KW
1324#define toTITLE(c) toUPPER(c)
1325
c753c8d3
KW
1326#define toLOWER_A(c) toLOWER(c)
1327#define toUPPER_A(c) toUPPER(c)
25200305
KW
1328#define toFOLD_A(c) toFOLD(c)
1329#define toTITLE_A(c) toTITLE(c)
1a0901db 1330
4650c663 1331/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1332#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1333 ? (c) \
f4cd282c 1334 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1335#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1336
1a0901db 1337/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1338 * all mapped to one of them, and these need special handling; returns the
1339 * input itself if is out-of-range */
b2bf251f 1340#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1341 ? (c) \
f4cd282c 1342 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1343#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1344
beab9ebe
KW
1345/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1346
1347/* For internal core Perl use only: the base macro for defining macros like
1348 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1349 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1350 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1351 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1352 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1353 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1354 * code points under locale rules, use the macros based on _generic_LC_uvchr
1355 * instead of this. */
beab9ebe
KW
1356#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1357 (! FITS_IN_8_BITS(c) \
1358 ? 0 \
31f05a37
KW
1359 : IN_UTF8_CTYPE_LOCALE \
1360 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1361 : cBOOL(non_utf8))
1362
1363/* For internal core Perl use only: a helper macro for defining macros like
1364 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1365 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1366 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1367 * _generic_LC_base for more info */
1368#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1369 _generic_LC_base(c,utf8_locale_classnum, \
1370 non_utf8_func( (_LC_CAST) (c)))
1371
1372/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1373 * 'c' is the platform's native underscore character */
1374#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1375 _generic_LC_base(c, utf8_locale_classnum, \
1376 (non_utf8_func( (_LC_CAST) (c)) \
1377 || (char)(c) == '_'))
1378
1379/* These next three are also for internal core Perl use only: case-change
1380 * helper macros */
1381#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1382 ? (c) \
31f05a37
KW
1383 : (IN_UTF8_CTYPE_LOCALE) \
1384 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1385 : (cast)function((cast)(c)))
beab9ebe 1386
31f05a37
KW
1387/* Note that the result can be larger than a byte in a UTF-8 locale. It
1388 * returns a single value, so can't adequately return the upper case of LATIN
1389 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1390 * values "SS"); instead it asserts against that under DEBUGGING, and
1391 * otherwise returns its input */
beab9ebe
KW
1392#define _generic_toUPPER_LC(c, function, cast) \
1393 (! FITS_IN_8_BITS(c) \
1394 ? (c) \
31f05a37 1395 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1396 ? (cast)function((cast)(c)) \
31f05a37
KW
1397 : ((((U8)(c)) == MICRO_SIGN) \
1398 ? GREEK_CAPITAL_LETTER_MU \
1399 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1400 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1401 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1402 ? (__ASSERT_(0) (c)) \
1403 : PL_mod_latin1_uc[ (U8) (c) ])))))
1404
1405/* Note that the result can be larger than a byte in a UTF-8 locale. It
1406 * returns a single value, so can't adequately return the fold case of LATIN
1407 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1408 * values "ss"); instead it asserts against that under DEBUGGING, and
1409 * otherwise returns its input */
beab9ebe 1410#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1411 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1412 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1413 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1414 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1415 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1416
84061b6a 1417/* Use the libc versions for these if available. */
f05550c0 1418#if defined(HAS_ISASCII)
84061b6a
KW
1419# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1420#else
1421# define isASCII_LC(c) isASCII(c)
1422#endif
1423
f05550c0 1424#if defined(HAS_ISBLANK)
84061b6a 1425# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1426#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1427# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1428#endif
1429
f05550c0 1430#define _LC_CAST U8
bbce6d69 1431
f05550c0 1432#ifdef WIN32
375f5f06
KW
1433 /* The Windows functions don't bother to follow the POSIX standard, which
1434 * for example says that something can't both be a printable and a control.
1435 * But Windows treats the \t control as a printable, and does such things
1436 * as making superscripts into both digits and punctuation. This tames
1437 * these flaws by assuming that the definitions of both controls and space
1438 * are correct, and then making sure that other definitions don't have
1439 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1440 * Not all possible weirdnesses are checked for, just the ones that were
1441 * detected on actual Microsoft code pages */
1442
b7d90381
KW
1443# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1444# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1445
1446# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1447 && isALPHANUMERIC_LC(c))
1448# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1449 ! isPUNCT_LC(c))
1450# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1451 isALPHANUMERIC_LC(c))
1452# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1453# define isIDFIRST_LC(c) (((c) == '_') \
1454 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1455# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1456# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1457# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1458# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1459# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1460# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1461 && isALPHANUMERIC_LC(c))
f05550c0
BF
1462
1463# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1464# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1465# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1466
1467#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1468 /* For most other platforms */
beab9ebe 1469
f05550c0
BF
1470# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1471# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1472# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1473# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1474# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1475# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1476# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1477# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1478# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1479# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1480# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1481# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1482# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1483
1484
1485# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1486# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1487# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1488
1489#else /* The final fallback position */
1490
b7d90381
KW
1491# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1492# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1493# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1494# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1495# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1496# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1497# define isLOWER_LC(c) (isascii(c) && islower(c))
1498# define isPRINT_LC(c) (isascii(c) && isprint(c))
1499# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1500# define isSPACE_LC(c) (isascii(c) && isspace(c))
1501# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1502# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1503# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1504
1505# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1506# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1507# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1508
f05550c0 1509#endif
55204971 1510
eba68aa0
KW
1511#define isIDCONT(c) isWORDCHAR(c)
1512#define isIDCONT_A(c) isWORDCHAR_A(c)
1513#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1514#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1515#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1516
4650c663
KW
1517/* For internal core Perl use only: the base macros for defining macros like
1518 * isALPHA_uni. 'c' is the code point to check. 'classnum' is the POSIX class
1519 * number defined earlier in this file. _generic_uni() is used for POSIX
1520 * classes where there is a macro or function 'above_latin1' that takes the
1521 * single argument 'c' and returns the desired value. These exist for those
1522 * classes which have simple definitions, avoiding the overhead of a hash
1523 * lookup or inversion list binary search. _generic_swash_uni() can be used
1524 * for classes where that overhead is faster than a direct lookup.
1525 * _generic_uni() won't compile if 'c' isn't unsigned, as it won't match the
1526 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1527 * have duplicate checks here, so could create versions of the macros that
1528 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1529
1530/* Note that all ignore 'use bytes' */
cd500f2f
KW
1531#define _generic_uni(classnum, above_latin1, c) ((c) < 256 \
1532 ? _generic_isCC(c, classnum) \
1533 : above_latin1(c))
1534#define _generic_swash_uni(classnum, c) ((c) < 256 \
1535 ? _generic_isCC(c, classnum) \
922e8cb4 1536 : _is_uni_FOO(classnum, c))
cd500f2f
KW
1537#define isALPHA_uni(c) _generic_swash_uni(_CC_ALPHA, c)
1538#define isALPHANUMERIC_uni(c) _generic_swash_uni(_CC_ALPHANUMERIC, c)
509fb054
KW
1539#define isASCII_uni(c) isASCII(c)
1540#define isBLANK_uni(c) _generic_uni(_CC_BLANK, is_HORIZWS_cp_high, c)
1541#define isCNTRL_uni(c) isCNTRL_L1(c) /* All controls are in Latin1 */
cd500f2f
KW
1542#define isDIGIT_uni(c) _generic_swash_uni(_CC_DIGIT, c)
1543#define isGRAPH_uni(c) _generic_swash_uni(_CC_GRAPH, c)
eba68aa0 1544#define isIDCONT_uni(c) _generic_uni(_CC_WORDCHAR, _is_uni_perl_idcont, c)
509fb054 1545#define isIDFIRST_uni(c) _generic_uni(_CC_IDFIRST, _is_uni_perl_idstart, c)
cd500f2f
KW
1546#define isLOWER_uni(c) _generic_swash_uni(_CC_LOWER, c)
1547#define isPRINT_uni(c) _generic_swash_uni(_CC_PRINT, c)
aea0c08c 1548
cd500f2f 1549#define isPUNCT_uni(c) _generic_swash_uni(_CC_PUNCT, c)
509fb054 1550#define isSPACE_uni(c) _generic_uni(_CC_SPACE, is_XPERLSPACE_cp_high, c)
779cf272
KW
1551#define isPSXSPC_uni(c) isSPACE_uni(c)
1552
cd500f2f 1553#define isUPPER_uni(c) _generic_swash_uni(_CC_UPPER, c)
509fb054 1554#define isVERTWS_uni(c) _generic_uni(_CC_VERTSPACE, is_VERTWS_cp_high, c)
cd500f2f 1555#define isWORDCHAR_uni(c) _generic_swash_uni(_CC_WORDCHAR, c)
509fb054 1556#define isXDIGIT_uni(c) _generic_uni(_CC_XDIGIT, is_XDIGIT_cp_high, c)
b18192e9 1557
b8d68ded 1558#define toFOLD_uni(c,s,l) to_uni_fold(c,s,l)
509fb054
KW
1559#define toLOWER_uni(c,s,l) to_uni_lower(c,s,l)
1560#define toTITLE_uni(c,s,l) to_uni_title(c,s,l)
1561#define toUPPER_uni(c,s,l) to_uni_upper(c,s,l)
a0ed51b3 1562
4650c663
KW
1563/* For internal core Perl use only: the base macros for defining macros like
1564 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1565 * point, not just 0-255. Like _generic_uni, there are two versions, one for
1566 * simple class definitions; the other for more complex. These are like
1567 * _generic_uni, so see it for more info. */
cd500f2f
KW
1568#define _generic_LC_uvchr(latin1, above_latin1, c) \
1569 (c < 256 ? latin1(c) : above_latin1(c))
1570#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1571 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1572
1573#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1574#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1575 _CC_ALPHANUMERIC, c)
b7d90381
KW
1576#define isASCII_LC_uvchr(c) isASCII_LC(c)
1577#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1578 is_HORIZWS_cp_high, c)
feeab5a9 1579#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1580#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1581#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1582#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1583 _is_uni_perl_idcont, c)
b7d90381 1584#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1585 _is_uni_perl_idstart, c)
1586#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1587#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1588#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1589#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1590#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1591 is_XPERLSPACE_cp_high, c)
cd500f2f 1592#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1593#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1594 _CC_WORDCHAR, c)
b7d90381
KW
1595#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1596 is_XDIGIT_cp_high, c)
e712593e 1597
b7d90381 1598#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1599
4650c663
KW
1600/* For internal core Perl use only: the base macros for defining macros like
1601 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1602 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1603 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1604 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1605 * same representation whether utf8 or not. Note that it assumes that the utf8
1606 * has been validated, and ignores 'use bytes' */
cd500f2f 1607#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1608 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1609 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1610 ? _generic_isCC( \
a62b247b 1611 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
aea0c08c 1612 *((p)+1 )), \
94bb8c36 1613 classnum) \
252810d0 1614 : utf8)
b7d90381
KW
1615/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1616 * 'above_latin1' can be a macro */
cd500f2f
KW
1617#define _generic_func_utf8(classnum, above_latin1, p) \
1618 _generic_utf8(classnum, p, above_latin1(p))
4650c663 1619/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1620 * 'above_latin1' parameter */
922e8cb4 1621#define _generic_swash_utf8(classnum, p) \
cd500f2f 1622 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
922e8cb4 1623
cc8ab7c0 1624/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1625 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1626 * class is TRUE for. Hence it can skip the tests for this range.
1627 * 'above_latin1' should include its arguments */
3ded5eb0 1628#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1629 (UTF8_IS_INVARIANT(*(p)) \
1630 ? _generic_isCC(*(p), classnum) \
1631 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1632 ? above_latin1 \
cc8ab7c0
KW
1633 : 0)
1634
84238efa
KW
1635/* NOTE that some of these macros have very similar ones in regcharclass.h.
1636 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1637 * there, differing in name only by an underscore from the one here
1638 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1639 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1640 * points; the regcharclass.h ones are implemented as a series of
1641 * "if-else-if-else ..." */
1642
b7d90381
KW
1643#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1644#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1645#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
509fb054
KW
1646 utf8, the non-utf8 macro works
1647 */
b7d90381 1648#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2
KW
1649
1650#ifdef EBCDIC
1651 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1652 * more efficient macro instead of the more general one */
88794300 1653# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
e8fa43e2
KW
1654#else
1655# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1656#endif
1657
cd500f2f
KW
1658#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1659 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1660#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1661#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1662 _is_utf8_perl_idcont, p)
e5dcd934 1663
c11ff943
KW
1664/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1665 * IDFIRST is an alnum. See
c8362b00 1666 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1667 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1668 * This used to be not the XID version, but we decided to go with the more
1669 * modern Unicode definition */
b7d90381 1670#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1671 _is_utf8_perl_idstart, p)
e5dcd934 1672
b7d90381
KW
1673#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1674#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1675#define isPSXSPC_utf8(p) isSPACE_utf8(p)
1676#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1677#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1678#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1679#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1680#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1681#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
cd500f2f 1682 is_XDIGIT_high(p))
a0ed51b3 1683
7d2b61dd 1684#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1685#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1686#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1687#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1688
4650c663
KW
1689/* For internal core Perl use only: the base macros for defining macros like
1690 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1691 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1692 * 'macro' parameter */
94bb8c36
KW
1693#define _generic_LC_utf8(macro, p, utf8) \
1694 (UTF8_IS_INVARIANT(*(p)) \
1695 ? macro(*(p)) \
1696 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
a62b247b 1697 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
9201a572
KW
1698 : utf8)
1699
cd500f2f
KW
1700#define _generic_LC_swash_utf8(macro, classnum, p) \
1701 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1702#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1703 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1704
b7d90381 1705#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
cd500f2f 1706 _CC_ALPHANUMERIC, p)
b7d90381
KW
1707#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1708#define isASCII_LC_utf8(p) isASCII_LC(*p)
1709#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1710 is_HORIZWS_high, p)
1711#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1712#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1713#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1714#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1715 _is_utf8_perl_idcont, p)
1716#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1717 _is_utf8_perl_idstart, p)
1718#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1719#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1720#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1721#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1722#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1723 is_XPERLSPACE_high, p)
1724#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
cd500f2f
KW
1725#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1726 _CC_WORDCHAR, p)
b7d90381
KW
1727#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1728 is_XDIGIT_high, p)
aaa51d5e 1729
fbc19f27
KW
1730/* Macros for backwards compatibility and for completeness when the ASCII and
1731 * Latin1 values are identical */
b7d90381
KW
1732#define isALPHAU(c) isALPHA_L1(c)
1733#define isDIGIT_L1(c) isDIGIT_A(c)
1734#define isOCTAL(c) isOCTAL_A(c)
1735#define isOCTAL_L1(c) isOCTAL_A(c)
1736#define isXDIGIT_L1(c) isXDIGIT_A(c)
1737#define isALNUM(c) isWORDCHAR(c)
1738#define isALNUMU(c) isWORDCHAR_L1(c)
1739#define isALNUM_LC(c) isWORDCHAR_LC(c)
1740#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 1741#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
1742#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1743#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1744#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1745#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1746#define isALNUMC(c) isALPHANUMERIC(c)
1747#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1748#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 1749#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 1750#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 1751#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1752
2bd1cbf6
KW
1753/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1754 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1755 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1756 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1757 * the outlier from the block that contains the other controls, just like
1758 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1759 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
1760 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1761 * etc. */
2bd1cbf6 1762#ifndef EBCDIC
75763b3a 1763# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 1764#else
75763b3a
KW
1765# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1766 ((isPRINT_A(c)) \
1767 ? (UNLIKELY((c) == '?') \
1768 ? QUESTION_MARK_CTRL \
1769 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1770 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1771 ? '?' \
1772 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 1773#endif
bbce6d69 1774
dea28490
JJ
1775/* Line numbers are unsigned, 32 bits. */
1776typedef U32 line_t;
e5dcd934 1777#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 1778
91152fc1
DG
1779/* Helpful alias for version prescan */
1780#define is_LAX_VERSION(a,b) \
1781 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1782
1783#define is_STRICT_VERSION(a,b) \
1784 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1785
1786#define BADVERSION(a,b,c) \
1787 if (b) { \
1788 *b = c; \
1789 } \
1790 return a;
8c52afec 1791
cb27eebd
KW
1792/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1793 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1794 * which is advanced. The input is validated only by an assert() in DEBUGGING
1795 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1796 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1797#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1798 ? (c) \
1799 : ((c) + 9))))
902008b8
KW
1800#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1801 ? (*(s)++) \
1802 : (*(s)++ + 9))))
95a59cab 1803
cb27eebd
KW
1804/* Converts a character known to represent an octal digit (0-7) to its numeric
1805 * value. The input is validated only by an assert() in DEBUGGING builds. In
1806 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1807#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1808
305b8651
KW
1809/* Efficiently returns a boolean as to if two native characters are equivalent
1810 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1811 * the ALPHA in the name is to remind you of that. This is asserted() in
1812 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1813 * works (on valid input) for both non- and UTF-8-encoded bytes.
1814 *
1815 * When one of the inputs is a compile-time constant and gets folded by the
1816 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1817 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1818 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
1819 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1820 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1821 * just a single 0, in the bit position where the upper- and lowercase differ.
1822 * */
305b8651
KW
1823#define isALPHA_FOLD_EQ(c1, c2) \
1824 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1825 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1826#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1827
8e84507e 1828/*
ccfc67b7
JH
1829=head1 Memory Management
1830
a02a5408 1831=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
1832The XSUB-writer's interface to the C C<malloc> function.
1833
0d7b2759
MH
1834Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1835
c5008215
JC
1836In 5.9.3, Newx() and friends replace the older New() API, and drops
1837the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 1838themselves. This aid has been superseded by a new build option,
d10b4965 1839PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
1840there for use in XS modules supporting older perls.
1841
a02a5408 1842=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 1843The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 1844cast. See also C<L</Newx>>.
954c1994 1845
0d7b2759
MH
1846Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1847
a02a5408 1848=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 1849The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 1850memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 1851
0d7b2759
MH
1852Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1853
954c1994
GS
1854=for apidoc Am|void|Renew|void* ptr|int nitems|type
1855The XSUB-writer's interface to the C C<realloc> function.
1856
0d7b2759
MH
1857Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1858
954c1994
GS
1859=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1860The XSUB-writer's interface to the C C<realloc> function, with
1861cast.
1862
0d7b2759
MH
1863Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1864
49b8b560 1865=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
1866The XSUB-writer's interface to the C C<free> function.
1867
0d7b2759
MH
1868This should B<ONLY> be used on memory obtained using L<"Newx"> and friends.
1869
954c1994
GS
1870=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1871The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 1872source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1873C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 1874
e90e2364 1875=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 1876Like C<Move> but returns C<dest>. Useful
72d33970 1877for encouraging compilers to tail-call
e90e2364
NC
1878optimise.
1879
954c1994
GS
1880=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1881The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 1882source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1883C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 1884
e90e2364
NC
1885=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1886
796b6530 1887Like C<Copy> but returns C<dest>. Useful
72d33970 1888for encouraging compilers to tail-call
e90e2364
NC
1889optimise.
1890
954c1994
GS
1891=for apidoc Am|void|Zero|void* dest|int nitems|type
1892
1893The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1894destination, C<nitems> is the number of items, and C<type> is the type.
1895
e90e2364
NC
1896=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1897
72d33970
FC
1898Like C<Zero> but returns dest. Useful
1899for encouraging compilers to tail-call
e90e2364
NC
1900optimise.
1901
da5d8dbb 1902=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 1903This is an architecture-independent macro to copy one structure to another.
954c1994 1904
7e337ee0
JH
1905=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1906
1907Fill up memory with a byte pattern (a byte repeated over and over
1908again) that hopefully catches attempts to access uninitialized memory.
1909
1910=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1911
1912PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1913
1c12ffb4 1914=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
1915
1916PoisonWith(0xEF) for catching access to freed memory.
1917
9965345d
JH
1918=for apidoc Am|void|Poison|void* dest|int nitems|type
1919
7e337ee0 1920PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
1921
1922=cut */
954c1994 1923
561b68a9
SH
1924/* Maintained for backwards-compatibility only. Use newSV() instead. */
1925#ifndef PERL_CORE
ff06c60c 1926#define NEWSV(x,len) newSV(len)
561b68a9 1927#endif
ff06c60c 1928
19a94d75
JH
1929#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1930
e6bdf523 1931
27d5b266 1932#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
1933
1934/* This expression will be constant-folded at compile time. It checks
1935 * whether or not the type of the count n is so small (e.g. U8 or U16, or
1936 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
1937 * As well as avoiding the need for a run-time check in some cases, it's
1938 * designed to avoid compiler warnings like:
1939 * comparison is always false due to limited range of data type
73e8ff00
DM
1940 * It's mathematically equivalent to
1941 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
1942 */
1943
1944# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
004073ba 1945 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
e6bdf523 1946
88f9f128 1947/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
1948 * compiler warnings. We *want* to write the expression as
1949 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
1950 * (for some compile-time constant C), but even when the LHS
1951 * constant-folds to false at compile-time, g++ insists on emitting
1952 * warnings about the RHS (e.g. "comparison is always false"), so instead
1953 * we write it as
e6bdf523 1954 *
d98e5cde 1955 * (cond ? n : X) > C
88f9f128 1956 *
d98e5cde
DM
1957 * where X is a constant with X > C always false. Choosing a value for X
1958 * is tricky. If 0, some compilers will complain about 0 > C always being
1959 * false; if 1, Coverity complains when n happens to be the constant value
1960 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
1961 * for X and hope that nothing else whines.
e6bdf523
DM
1962 */
1963
1964# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
1965 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
1966 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
1967
1968# define MEM_WRAP_CHECK(n,t) \
1969 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1970 && (croak_memory_wrap(),0))
1971
1972# define MEM_WRAP_CHECK_1(n,t,a) \
1973 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1974 && (Perl_croak_nocontext("%s",(a)),0))
1975
8b44ba4c 1976#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 1977
d1decf2b 1978#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
1979#else
1980
410319be
NC
1981#define MEM_WRAP_CHECK(n,t)
1982#define MEM_WRAP_CHECK_1(n,t,a)
1983#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
1984#define MEM_WRAP_CHECK_(n,t)
1985
6fff79ce 1986#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 1987
1936d2a7 1988#endif
8b44ba4c 1989
fe4f188c 1990#ifdef PERL_MEM_LOG
46c6c7e2 1991/*
9f653bb5 1992 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
1993 * go through functions, which are handy for debugging breakpoints, but
1994 * which more importantly get the immediate calling environment (file and
e352bcff
JH
1995 * line number, and C function name if available) passed in. This info can
1996 * then be used for logging the calls, for which one gets a sample
73d1d973 1997 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 1998 *
46c6c7e2 1999 * Known problems:
94e892a6 2000 * - not all memory allocs get logged, only those
46c6c7e2 2001 * that go through Newx() and derivatives (while all
94e892a6 2002 * Safefrees do get logged)
46c6c7e2
JH
2003 * - __FILE__ and __LINE__ do not work everywhere
2004 * - __func__ or __FUNCTION__ even less so
2005 * - I think more goes on after the perlio frees but
2006 * the thing is that STDERR gets closed (as do all
2007 * the file descriptors)
2008 * - no deeper calling stack than the caller of the Newx()
2009 * or the kind, but do I look like a C reflection/introspection
2010 * utility to you?
2011 * - the function prototypes for the logging functions
2012 * probably should maybe be somewhere else than handy.h
2013 * - one could consider inlining (macrofying) the logging
2014 * for speed, but I am too lazy
2015 * - one could imagine recording the allocations in a hash,
2016 * (keyed by the allocation address?), and maintain that
2017 * through reallocs and frees, but how to do that without
2018 * any News() happening...?
73d1d973 2019 * - lots of -Ddefines to get useful/controllable output
b953482e 2020 * - lots of ENV reads
46c6c7e2
JH
2021 */
2022
fc6bde6f 2023PERL_CALLCONV Malloc_t Perl_mem_log_alloc(const UV n, const UV typesize, const char *type_name, Malloc_t newalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 2024
fc6bde6f 2025PERL_CALLCONV Malloc_t Perl_mem_log_realloc(const UV n, const UV typesize, const char *type_name, Malloc_t oldalloc, Malloc_t newalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 2026
fc6bde6f 2027PERL_CALLCONV Malloc_t Perl_mem_log_free(Malloc_t oldalloc, const char *filename, const int linenumber, const char *funcname);
46c6c7e2 2028
0b0ab801 2029# ifdef PERL_CORE
73d1d973 2030# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2031enum mem_log_type {
2032 MLT_ALLOC,
2033 MLT_REALLOC,
d7a2c63c
MHM
2034 MLT_FREE,
2035 MLT_NEW_SV,
2036 MLT_DEL_SV
0b0ab801
MHM
2037};
2038# endif
12754f92 2039# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2040void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2041void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2042# endif
0b0ab801
MHM
2043# endif
2044
fe4f188c
JH
2045#endif
2046
2047#ifdef PERL_MEM_LOG
d1401ee9
MHM
2048#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2049#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2050#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2051#endif
2052
2053#ifndef MEM_LOG_ALLOC
2054#define MEM_LOG_ALLOC(n,t,a) (a)
2055#endif
2056#ifndef MEM_LOG_REALLOC
2057#define MEM_LOG_REALLOC(n,t,v,a) (a)
2058#endif
2059#ifndef MEM_LOG_FREE
2060#define MEM_LOG_FREE(a) (a)
2061#endif
2062
d1401ee9
MHM
2063#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2064#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2065#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2066
2067#ifndef PERL_CORE
a02a5408
JC
2068/* pre 5.9.x compatibility */
2069#define New(x,v,n,t) Newx(v,n,t)
2070#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2071#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2072#endif
a02a5408 2073
ff68c719 2074#define Renew(v,n,t) \
d1401ee9 2075 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2076#define Renewc(v,n,t,c) \
d1401ee9 2077 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2078
2079#ifdef PERL_POISON
2080#define Safefree(d) \
06c0cc96 2081 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2082#else
fe4f188c 2083#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2084#endif
55497cff 2085
8b44ba4c
NC
2086#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2087#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2088#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2089
8b44ba4c
NC
2090#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2091#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2092#ifdef HAS_MEMSET
8b44ba4c 2093#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2094#else
8b44ba4c
NC
2095/* Using bzero(), which returns void. */
2096#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2097#endif
2098
7e337ee0
JH
2099#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2100#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2101#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2102#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2103
caa674f3
DD
2104#ifdef PERL_POISON
2105# define PERL_POISON_EXPR(x) x
2106#else
2107# define PERL_POISON_EXPR(x)
2108#endif
2109
2304df62 2110#ifdef USE_STRUCT_COPY
ff68c719 2111#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2112#else
2113#define StructCopy(s,d,t) Copy(s,d,1,t)
2114#endif
2cc61e15 2115
c3caa5c3
JH
2116/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2117 * want your zero-based indices to be less than but not equal to).
2118 *
2119 * C_ARRAY_END is one past the last: half-open/half-closed range,
2120 * not last-inclusive range. */
622913ab 2121#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2122#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2123
2cc61e15
DD
2124#ifdef NEED_VA_COPY
2125# ifdef va_copy
2126# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2127# else
a1866d1b
JH
2128# if defined(__va_copy)
2129# define Perl_va_copy(s, d) __va_copy(d, s)
2130# else
2131# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2132# endif
2cc61e15
DD
2133# endif
2134#endif
2135
472d47bc
SB
2136/* convenience debug macros */
2137#ifdef USE_ITHREADS
2138#define pTHX_FORMAT "Perl interpreter: 0x%p"
2139#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2140#define pTHX_VALUE_ (void *)my_perl,
2141#define pTHX_VALUE (void *)my_perl
2142#define pTHX__VALUE_ ,(void *)my_perl,
2143#define pTHX__VALUE ,(void *)my_perl
472d47bc 2144#else
3609ea0d 2145#define pTHX_FORMAT
472d47bc 2146#define pTHX__FORMAT
3609ea0d 2147#define pTHX_VALUE_
472d47bc 2148#define pTHX_VALUE
3609ea0d 2149#define pTHX__VALUE_
472d47bc
SB
2150#define pTHX__VALUE
2151#endif /* USE_ITHREADS */
3609ea0d 2152
2acdbac1
NC
2153/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2154 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2155 CPAN::Unpack show any users outside the core. */
2156#ifdef PERL_CORE
b7d90381
KW
2157# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2158 "Use of " s " is deprecated")
2acdbac1
NC
2159#endif
2160
dfff4baf
BF
2161/* Internal macros to deal with gids and uids */
2162#ifdef PERL_CORE
2163
2164# if Uid_t_size > IVSIZE
2165# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2166# define SvUID(sv) SvNV(sv)
2167# else
2168# if Uid_t_sign <= 0
2169# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2170# define SvUID(sv) SvIV(sv)
2171# else
2172# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2173# define SvUID(sv) SvUV(sv)
2174# endif
2175# endif /* Uid_t_size */
2176
2177# if Gid_t_size > IVSIZE
2178# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2179# define SvGID(sv) SvNV(sv)
2180# else
2181# if Gid_t_sign <= 0
2182# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2183# define SvGID(sv) SvIV(sv)
2184# else
2185# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2186# define SvGID(sv) SvUV(sv)
2187# endif
2188# endif /* Gid_t_size */
2189
2190#endif
2191
9d745869
KW
2192#endif /* HANDY_H */
2193
e9a8c099 2194/*
14d04a33 2195 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2196 */