This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Add is_utf8_fixed_width_buf_flags() and use it
[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
085b7534 492/* memEQ and memNE where second comparand is a string constant */
568a785a 493#define memEQs(s1, l, s2) \
8e6497b6 494 (sizeof(s2)-1 == l && memEQ(s1, ("" s2 ""), (sizeof(s2)-1)))
568a785a
NC
495#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
496
062b6850
KW
497#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
498#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
499#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
500#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
501
bbce6d69
PP
502/*
503 * Character classes.
504 *
505 * Unfortunately, the introduction of locales means that we
506 * can't trust isupper(), etc. to tell the truth. And when
507 * it comes to /\w+/ with tainting enabled, we *must* be able
508 * to trust our character classes.
509 *
510 * Therefore, the default tests in the text of Perl will be
511 * independent of locale. Any code that wants to depend on
512 * the current locale will use the tests that begin with "lc".
513 */
514
2304df62
AD
515#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
516# ifndef CTYPE256
517# define CTYPE256
518# endif
519#endif
520
954c1994 521/*
ccfc67b7 522
dcccc8ff 523=head1 Character classification
243effed
KW
524This section is about functions (really macros) that classify characters
525into types, such as punctuation versus alphabetic, etc. Most of these are
526analogous to regular expression character classes. (See
527L<perlrecharclass/POSIX Character Classes>.) There are several variants for
528each class. (Not all macros have all variants; each item below lists the
529ones valid for it.) None are affected by C<use bytes>, and only the ones
530with C<LC> in the name are affected by the current locale.
531
532The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
533C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
534represented by that octet is (or on non-ASCII platforms, corresponds to) an
535ASCII character in the named class based on platform, Unicode, and Perl rules.
536If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 537
6aff1f14
KW
538Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
539with no suffix C<"_A">.
4b9734bf
KW
540
541Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
542onto the platform. That is, the code points that are ASCII are unaffected,
543since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
544as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
545true when called with the code point 0xDF, which is a word character in both
4650c663 546ASCII and EBCDIC (though it represents different characters in each).
243effed 547
d0da05db 548Variant C<isFOO_uvchr> is like the C<isFOO_L1> variant, but accepts any UV code
243effed
KW
549point as input. If the code point is larger than 255, Unicode rules are used
550to determine if it is in the character class. For example,
d0da05db 551C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 552WITH MACRON in Unicode, and is a word character.
243effed 553
d0da05db 554Variant C<isFOO_utf8> is like C<isFOO_uvchr>, but the input is a pointer to a
dc6b0978
KW
555(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
556possibly containing embedded C<NUL> characters). The classification of just
557the first (possibly multi-byte) character in the string is tested.
243effed 558
31f05a37
KW
559Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
560result is based on the current locale, which is what C<LC> in the name stands
561for. If Perl can determine that the current locale is a UTF-8 locale, it uses
562the published Unicode rules; otherwise, it uses the C library function that
563gives the named classification. For example, C<isDIGIT_LC()> when not in a
564UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
565returned if the input won't fit into an octet. On some platforms where the C
566library function is known to be defective, Perl changes its result to follow
567the POSIX standard's rules.
243effed
KW
568
569Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
570returns the same as C<isFOO_LC> for input code points less than 256, and
571returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
572
b7d90381 573Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer
dc6b0978
KW
574to a (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
575possibly containing embedded C<NUL> characters). The classification of just
576the first (possibly multi-byte) character in the string is tested.
ccfc67b7 577
954c1994 578=for apidoc Am|bool|isALPHA|char ch
2787a470 579Returns a boolean indicating whether the specified character is an
6aff1f14 580alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
581See the L<top of this section|/Character classification> for an explanation of
582variants
d0da05db 583C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>, C<isALPHA_LC>,
243effed 584C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
8a58bdcf 585
15861f94
KW
586=for apidoc Am|bool|isALPHANUMERIC|char ch
587Returns a boolean indicating whether the specified character is a either an
6aff1f14 588alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
589See the L<top of this section|/Character classification> for an explanation of
590variants
d0da05db 591C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
e5ad6aba 592C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
15861f94
KW
593C<isALPHANUMERIC_LC_utf8>.
594
8a58bdcf
KW
595=for apidoc Am|bool|isASCII|char ch
596Returns a boolean indicating whether the specified character is one of the 128
243effed 597characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 598On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
599character corresponds to an ASCII character. Variants C<isASCII_A()> and
600C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
601See the L<top of this section|/Character classification> for an explanation of
602variants
d0da05db 603C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
243effed
KW
604C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
605library routine C<isascii()>. In these cases, the variants whose names contain
606C<LC> are the same as the corresponding ones without.
607
d98532ea
KW
608Also note, that because all ASCII characters are UTF-8 invariant (meaning they
609have the exact same representation (always a single byte) whether encoded in
610UTF-8 or not), C<isASCII> will give the correct results when called with any
611byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
612work properly on any string encoded or not in UTF-8.
613
243effed
KW
614=for apidoc Am|bool|isBLANK|char ch
615Returns a boolean indicating whether the specified character is a
6aff1f14 616character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
617See the L<top of this section|/Character classification> for an explanation of
618variants
d0da05db 619C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>, C<isBLANK_LC>,
243effed
KW
620C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
621platforms do not have the C library routine C<isblank()>. In these cases, the
622variants whose names contain C<LC> are the same as the corresponding ones
623without.
624
625=for apidoc Am|bool|isCNTRL|char ch
626Returns a boolean indicating whether the specified character is a
6aff1f14 627control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
628See the L<top of this section|/Character classification> for an explanation of
629variants
d0da05db 630C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
6aff1f14
KW
631C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
632On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 633
954c1994 634=for apidoc Am|bool|isDIGIT|char ch
2787a470 635Returns a boolean indicating whether the specified character is a
6aff1f14 636digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 637Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
638See the L<top of this section|/Character classification> for an explanation of
639variants
d0da05db 640C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
243effed
KW
641C<isDIGIT_LC_utf8>.
642
643=for apidoc Am|bool|isGRAPH|char ch
644Returns a boolean indicating whether the specified character is a
6aff1f14 645graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff
KW
646See the L<top of this section|/Character classification> for an explanation of
647variants
d0da05db 648C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
243effed 649C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
954c1994 650
0c82b6df 651=for apidoc Am|bool|isLOWER|char ch
2787a470 652Returns a boolean indicating whether the specified character is a
6aff1f14 653lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
654See the L<top of this section|/Character classification> for an explanation of
655variants
d0da05db 656C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>, C<isLOWER_LC>,
243effed 657C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
0c82b6df 658
c99e91e9 659=for apidoc Am|bool|isOCTAL|char ch
2787a470 660Returns a boolean indicating whether the specified character is an
6aff1f14 661octal digit, [0-7].
243effed
KW
662The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
663C<isOCTAL>.
664
665=for apidoc Am|bool|isPUNCT|char ch
666Returns a boolean indicating whether the specified character is a
6aff1f14
KW
667punctuation character, analogous to C<m/[[:punct:]]/>.
668Note that the definition of what is punctuation isn't as
243effed
KW
669straightforward as one might desire. See L<perlrecharclass/POSIX Character
670Classes> for details.
dcccc8ff
KW
671See the L<top of this section|/Character classification> for an explanation of
672variants
d0da05db 673C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
243effed 674C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
c99e91e9 675
0c82b6df 676=for apidoc Am|bool|isSPACE|char ch
2787a470 677Returns a boolean indicating whether the specified character is a
6aff1f14 678whitespace character. This is analogous
398d098a 679to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 680this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
681locale forms of this macro (the ones with C<LC> in their names) matched
682precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
683in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
684(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
685See the L<top of this section|/Character classification> for an explanation of
686variants
d0da05db 687C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>, C<isSPACE_LC>,
243effed 688C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
0c82b6df 689
398d098a
KW
690=for apidoc Am|bool|isPSXSPC|char ch
691(short for Posix Space)
779cf272
KW
692Starting in 5.18, this is identical in all its forms to the
693corresponding C<isSPACE()> macros.
398d098a
KW
694The locale forms of this macro are identical to their corresponding
695C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
696non-locale forms differ from their C<isSPACE()> forms only in that the
697C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
698Otherwise they are identical. Thus this macro is analogous to what
699C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 700See the L<top of this section|/Character classification> for an explanation of
d0da05db 701variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
b7d90381 702C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
398d098a 703
954c1994 704=for apidoc Am|bool|isUPPER|char ch
2787a470 705Returns a boolean indicating whether the specified character is an
6aff1f14 706uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff
KW
707See the L<top of this section|/Character classification> for an explanation of
708variants
d0da05db 709C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>, C<isUPPER_LC>,
243effed 710C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
954c1994 711
243effed 712=for apidoc Am|bool|isPRINT|char ch
8eea39dd 713Returns a boolean indicating whether the specified character is a
6aff1f14 714printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
715See the L<top of this section|/Character classification> for an explanation of
716variants
d0da05db 717C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>, C<isPRINT_LC>,
243effed
KW
718C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
719
720=for apidoc Am|bool|isWORDCHAR|char ch
721Returns a boolean indicating whether the specified character is a character
722that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
723in a regular expression. A word character is an alphabetic character, a
724decimal digit, a connecting punctuation character (such as an underscore), or
725a "mark" character that attaches to one of those (like some sort of accent).
726C<isALNUM()> is a synonym provided for backward compatibility, even though a
727word character includes more than the standard C language meaning of
728alphanumeric.
dcccc8ff
KW
729See the L<top of this section|/Character classification> for an explanation of
730variants
d0da05db 731C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and C<isWORDCHAR_utf8>.
1a83413c
KW
732C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
733described there, but additionally include the platform's native underscore.
8a58bdcf
KW
734
735=for apidoc Am|bool|isXDIGIT|char ch
736Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
737digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
738and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
739See the L<top of this section|/Character classification> for an explanation of
740variants
d0da05db 741C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
243effed
KW
742C<isXDIGIT_LC_utf8>.
743
3c3ecf18
KW
744=for apidoc Am|bool|isIDFIRST|char ch
745Returns a boolean indicating whether the specified character can be the first
746character of an identifier. This is very close to, but not quite the same as
747the official Unicode property C<XID_Start>. The difference is that this
748returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
749See the L<top of this section|/Character classification> for an explanation of
750variants
d0da05db 751C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
3c3ecf18
KW
752C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
753
754=for apidoc Am|bool|isIDCONT|char ch
755Returns a boolean indicating whether the specified character can be the
756second or succeeding character of an identifier. This is very close to, but
757not quite the same as the official Unicode property C<XID_Continue>. The
758difference is that this returns true only if the input character also matches
dcccc8ff
KW
759L</isWORDCHAR>. See the L<top of this section|/Character classification> for
760an
d0da05db 761explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
3c3ecf18
KW
762C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
763C<isIDCONT_LC_utf8>.
764
243effed 765=head1 Miscellaneous Functions
8eea39dd 766
95a59cab 767=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 768Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
769Behaviour is only well defined when isXDIGIT(*str) is true.
770
e7c1e6c1
KW
771=head1 Character case changing
772
1f607577
KW
773=for apidoc Am|U8|toUPPER|U8 ch
774Converts the specified character to uppercase. If the input is anything but an
775ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 776C<toUPPER_A> is equivalent.
954c1994 777
d0da05db
KW
778=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
779Converts the code point C<cp> to its uppercase version, and
780stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
781point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
782that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
783bytes since the uppercase version may be longer than the original character.
784
785The first code point of the uppercased version is returned
786(but note, as explained just above, that there may be more.)
787
788=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
789Converts the UTF-8 encoded character at C<p> to its uppercase version, and
790stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
791that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
792bytes since the uppercase version may be longer than the original character.
793
794The first code point of the uppercased version is returned
795(but note, as explained just above, that there may be more.)
796
797The input character at C<p> is assumed to be well-formed.
798
25200305
KW
799=for apidoc Am|U8|toFOLD|U8 ch
800Converts the specified character to foldcase. If the input is anything but an
801ASCII uppercase character, that input character itself is returned. Variant
802C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 803Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 804
d0da05db
KW
805=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
806Converts the code point C<cp> to its foldcase version, and
807stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
808point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
809that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
810bytes since the foldcase version may be longer than the original character.
811
812The first code point of the foldcased version is returned
813(but note, as explained just above, that there may be more.)
814
815=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
816Converts the UTF-8 encoded character at C<p> to its foldcase version, and
817stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
818that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
819bytes since the foldcase version may be longer than the original character.
820
821The first code point of the foldcased version is returned
822(but note, as explained just above, that there may be more.)
823
824The input character at C<p> is assumed to be well-formed.
825
826=for apidoc Am|U8|toLOWER|U8 ch
827Converts the specified character to lowercase. If the input is anything but an
828ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 829C<toLOWER_A> is equivalent.
954c1994 830
1f607577 831=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
832Converts the specified Latin1 character to lowercase. The results are
833undefined if the input doesn't fit in a byte.
1f607577
KW
834
835=for apidoc Am|U8|toLOWER_LC|U8 ch
836Converts the specified character to lowercase using the current locale's rules,
837if possible; otherwise returns the input character itself.
838
d0da05db
KW
839=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
840Converts the code point C<cp> to its lowercase version, and
841stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
842point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
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
849=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
850Converts the UTF-8 encoded character at C<p> to its lowercase version, and
851stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
852that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
853bytes since the lowercase version may be longer than the original character.
854
855The first code point of the lowercased version is returned
856(but note, as explained just above, that there may be more.)
857
858The input character at C<p> is assumed to be well-formed.
859
25200305
KW
860=for apidoc Am|U8|toTITLE|U8 ch
861Converts the specified character to titlecase. If the input is anything but an
862ASCII lowercase character, that input character itself is returned. Variant
b7d90381 863C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 864range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 865not a concept used in locale handling, so there is no functionality for that.)
25200305 866
d0da05db
KW
867=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
868Converts the code point C<cp> to its titlecase version, and
869stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
870point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
871that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
872bytes since the titlecase version may be longer than the original character.
873
874The first code point of the titlecased version is returned
875(but note, as explained just above, that there may be more.)
876
877=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
878Converts the UTF-8 encoded character at C<p> to its titlecase version, and
879stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
880that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
881bytes since the titlecase version may be longer than the original character.
882
883The first code point of the titlecased version is returned
884(but note, as explained just above, that there may be more.)
885
886The input character at C<p> is assumed to be well-formed.
887
954c1994 888=cut
353c9b6f 889
d0da05db 890XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
25200305 891really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
243effed 892
8a58bdcf 893Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
894patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
895
954c1994
GS
896*/
897
7c062697
KW
898/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
899 * is known only in the perl core, and this macro can be called from outside
900 * that */
901#ifdef HAS_QUAD
902# define WIDEST_UTYPE U64TYPE
903#else
904# define WIDEST_UTYPE U32
905#endif
906
3912bc88
KW
907/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
908 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
909 * bits of information are lost even on a 64-bit machine, but to get the
910 * compiler to optimize it out if possible. This is because Configure makes
911 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
912 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
913 *
914 * For Coverity, be always true, because otherwise Coverity thinks
915 * it finds several expressions that are always true, independent
916 * of operands. Well, they are, but that is kind of the point.
220c71bf 917 */
7e75d1a1 918#ifndef __COVERITY__
6c5b02ac
KW
919 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
920 * pointer) */
921#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 922 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
923#else
924#define FITS_IN_8_BITS(c) (1)
925#endif
cf301eb7 926
41f43cc2 927#ifdef EBCDIC
b6340bd0 928# ifndef _ALL_SOURCE
0852beac
KW
929 /* The native libc isascii() et.al. functions return the wrong results
930 * on at least z/OS unless this is defined. */
b6340bd0
KW
931# error _ALL_SOURCE should probably be defined
932# endif
41f43cc2 933#else
0852beac
KW
934 /* There is a simple definition of ASCII for ASCII platforms. But the
935 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 936 * other macros below.
3f3c579d
KW
937 *
938 * The cast here is used instead of '(c) >= 0', because some compilers emit
939 * a warning that that test is always true when the parameter is an
940 * unsigned type. khw supposes that it could be written as
941 * && ((c) == '\0' || (c) > 0)
942 * to avoid the message, but the cast will likely avoid extra branches even
943 * with stupid compilers.
944 *
945 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
946 * a pointer) */
9c903d59 947# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
948#endif
949
38694112
KW
950/* Take the eight possible bit patterns of the lower 3 bits and you get the
951 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
952 * can be ignored. If the rest match '0', we have an octal */
953#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 954
9fb1bf9d 955#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 956
a500dc72
KW
957/* Character class numbers. For internal core Perl use only. The ones less
958 * than 32 are used in PL_charclass[] and the ones up through the one that
959 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
960 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
961 * their actual definitions are here. If that file has a name not used here,
962 * it won't compile.
1709d539
KW
963 *
964 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
965 * frequency of their use. This gives a slight edge to exiting a loop earlier
966 * (in reginclass() in regexec.c) */
1709d539
KW
967# define _CC_WORDCHAR 0 /* \w and [:word:] */
968# define _CC_DIGIT 1 /* \d and [:digit:] */
969# define _CC_ALPHA 2 /* [:alpha:] */
970# define _CC_LOWER 3 /* [:lower:] */
971# define _CC_UPPER 4 /* [:upper:] */
972# define _CC_PUNCT 5 /* [:punct:] */
973# define _CC_PRINT 6 /* [:print:] */
15861f94 974# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 975# define _CC_GRAPH 8 /* [:graph:] */
359b005e 976# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1709d539 977
b0d691b2 978#define _FIRST_NON_SWASH_CC 10
1709d539
KW
979/* The character classes above are implemented with swashes. The second group
980 * (just below) contains the ones implemented without. These are also sorted
981 * in rough order of the frequency of their use, except that \v should be last,
982 * as it isn't a real Posix character class, and some (small) inefficiencies in
983 * regular expression handling would be introduced by putting it in the middle
984 * of those that are. Also, cntrl and ascii come after the others as it may be
985 * useful to group these which have no members that match above Latin1, (or
986 * above ASCII in the latter case) */
987
779cf272 988# define _CC_SPACE 10 /* \s, [:space:] */
b0d691b2
KW
989# define _CC_BLANK 11 /* [:blank:] */
990# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
991# define _CC_CNTRL 13 /* [:cntrl:] */
992# define _CC_ASCII 14 /* [:ascii:] */
993# define _CC_VERTSPACE 15 /* \v */
1709d539 994
a0947d7b
KW
995# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
996
1709d539 997/* The members of the third group below do not need to be coordinated with data
3ffc8c70 998 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
999# define _CC_IDFIRST 16
1000# define _CC_CHARNAME_CONT 17
1001# define _CC_NONLATIN1_FOLD 18
1002# define _CC_NONLATIN1_SIMPLE_FOLD 19
1003# define _CC_QUOTEMETA 20
1004# define _CC_NON_FINAL_FOLD 21
1005# define _CC_IS_IN_SOME_FOLD 22
1006# define _CC_MNEMONIC_CNTRL 23
073c22b3
KW
1007
1008/* This next group is only used on EBCDIC platforms, so theoretically could be
1009 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
1010# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1011# define _CC_UTF8_IS_START 29
1012# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1013# define _CC_UTF8_IS_CONTINUATION 31
1014/* Unused: 24-27
f4cdb42c
KW
1015 * If more bits are needed, one could add a second word for non-64bit
1016 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1017 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1018 * is used only for optimization (as of this writing), and differs in the
1019 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1020 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1021 * would just cause /i regexes which match them to run less efficiently.
1022 * Similarly the EBCDIC-only bits are used just for speed, and could be
1023 * replaced by other means */
96ac0975 1024
3a371f2f
KW
1025#if defined(PERL_CORE) || defined(PERL_EXT)
1026/* An enum version of the character class numbers, to help compilers
1027 * optimize */
1028typedef enum {
3a371f2f 1029 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1030 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1031 _CC_ENUM_ASCII = _CC_ASCII,
1032 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1033 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1034 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1035 _CC_ENUM_DIGIT = _CC_DIGIT,
1036 _CC_ENUM_GRAPH = _CC_GRAPH,
1037 _CC_ENUM_LOWER = _CC_LOWER,
1038 _CC_ENUM_PRINT = _CC_PRINT,
1039 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1040 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1041 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1042 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1043 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1044 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1045} _char_class_number;
1046#endif
1047
63c61c3f 1048#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1049#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1050
b7d90381
KW
1051#if defined(PERL_IN_UTF8_C) \
1052 || defined(PERL_IN_REGCOMP_C) \
1053 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1054# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1055 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1056 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1057 #error Need to adjust order of swash_property_names[]
1058# endif
1059
b44daeeb
KW
1060/* This is declared static in each of the few files that this is #defined for
1061 * to keep them from being publicly accessible. Hence there is a small amount
1062 * of wasted space */
1063
922e8cb4
KW
1064static const char* const swash_property_names[] = {
1065 "XPosixWord",
1066 "XPosixDigit",
1067 "XPosixAlpha",
1068 "XPosixLower",
1069 "XPosixUpper",
1070 "XPosixPunct",
1071 "XPosixPrint",
1072 "XPosixAlnum",
b0d691b2
KW
1073 "XPosixGraph",
1074 "Cased"
922e8cb4
KW
1075};
1076#endif
1077
6635f04f 1078START_EXTERN_C
96ac0975
NC
1079# ifdef DOINIT
1080EXTCONST U32 PL_charclass[] = {
1081# include "l1_char_class_tab.h"
1082};
1083
1084# else /* ! DOINIT */
1085EXTCONST U32 PL_charclass[];
1086# endif
6635f04f 1087END_EXTERN_C
96ac0975 1088
265c1f46 1089 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1090# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1091
1092 /* For internal core Perl use only: the base macro for defining macros like
1093 * isALPHA */
ff7ecfc3 1094# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1095 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1096
f4cdb42c
KW
1097 /* The mask for the _A versions of the macros; it just adds in the bit for
1098 * ASCII. */
1099# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1100
4650c663
KW
1101 /* For internal core Perl use only: the base macro for defining macros like
1102 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1103 * the ASCII bit are present */
b7d90381
KW
1104# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1105 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1106 == _CC_mask_A(classnum)))
f4cdb42c 1107
f4cdb42c 1108# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1109# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1110# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1111# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1112# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1113# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1114# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1115# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1116# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1117# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1118# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1119# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1120# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1121 */
d95f8b6a 1122# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1123# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1124# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1125# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1126
1127 /* continuation character for legal NAME in \N{NAME} */
1128# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1129
1130# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1131# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1132# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1133# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1134# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1135# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1136# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1137# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1138# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1139# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1140
0852beac
KW
1141# ifdef EBCDIC
1142# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1143# endif
1144
f12c0118
KW
1145 /* Participates in a single-character fold with a character above 255 */
1146# 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)))
1147
1148 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1149# 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 1150
4eeeb416 1151# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1152# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1153 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1154# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1155 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1156# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1157 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1158#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1159
1160 /* If we don't have perl.h, we are compiling a utility program. Below we
1161 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1162 * to it. Most are coded based on first principles. These are written to
74665a89 1163 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
0852beac
KW
1164# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1165# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1166# define isSPACE_A(c) (isBLANK_A(c) \
1167 || (c) == '\n' \
1168 || (c) == '\r' \
1169 || (c) == '\v' \
0852beac 1170 || (c) == '\f')
74665a89
KW
1171 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1172 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1173 * performance (if optimization isn't on), and allow the same code to be
1174 * used for both platform types */
1175# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1176 && ( (c) <= 'i' \
1177 || ((c) >= 'j' && (c) <= 'r') \
1178 || (c) >= 's'))
1179# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1180 && ( (c) <= 'I' \
1181 || ((c) >= 'J' && (c) <= 'R') \
1182 || (c) >= 'S'))
a4d7a999
KW
1183# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1184# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1185# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1186# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
74665a89
KW
1187# define isXDIGIT_A(c) (isDIGIT_A(c) \
1188 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1189 || ((c) <= 'F' && (c) >= 'A'))
74665a89
KW
1190# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1191 || (c) == '#' || (c) == '$' || (c) == '%' \
1192 || (c) == '&' || (c) == '\'' || (c) == '(' \
1193 || (c) == ')' || (c) == '*' || (c) == '+' \
1194 || (c) == ',' || (c) == '.' || (c) == '/' \
1195 || (c) == ':' || (c) == ';' || (c) == '<' \
1196 || (c) == '=' || (c) == '>' || (c) == '?' \
1197 || (c) == '@' || (c) == '[' || (c) == '\\' \
1198 || (c) == ']' || (c) == '^' || (c) == '_' \
1199 || (c) == '`' || (c) == '{' || (c) == '|' \
1200 || (c) == '}' || (c) == '~')
1201# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1202# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1203
0852beac 1204# ifdef EBCDIC
74665a89
KW
1205 /* The below is accurate for the 3 EBCDIC code pages traditionally
1206 * supported by perl. The only difference between them in the controls
1207 * is the position of \n, and that is represented symbolically below */
1208# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1209 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1210 || (c) == '\t' || (c) == '\v' \
1211 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1212 || (c) == 7 /* U+7F DEL */ \
1213 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1214 /* DLE, DC[1-3] */ \
1215 || (c) == 0x18 /* U+18 CAN */ \
1216 || (c) == 0x19 /* U+19 EOM */ \
1217 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1218 || (c) == 0x26 /* U+17 ETB */ \
1219 || (c) == 0x27 /* U+1B ESC */ \
1220 || (c) == 0x2D /* U+05 ENQ */ \
1221 || (c) == 0x2E /* U+06 ACK */ \
1222 || (c) == 0x32 /* U+16 SYN */ \
1223 || (c) == 0x37 /* U+04 EOT */ \
1224 || (c) == 0x3C /* U+14 DC4 */ \
1225 || (c) == 0x3D /* U+15 NAK */ \
1226 || (c) == 0x3F)/* U+1A SUB */
0852beac 1227# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1228# else /* isASCII is already defined for ASCII platforms, so can use that to
1229 define isCNTRL */
1230# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1231# endif
1232
3ffc8c70 1233 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1234 * during debugging, and it seems best to keep them. We may be called
1235 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1236 * do anything anyway, so make it not a problem */
1237# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1238# define NATIVE_TO_LATIN1(ch) (ch)
1239# endif
3ded5eb0
KW
1240# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1241# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1242# define isBLANK_L1(c) (isBLANK_A(c) \
1243 || (FITS_IN_8_BITS(c) \
1244 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1245# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1246# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1247# define isLOWER_L1(c) (isLOWER_A(c) \
1248 || (FITS_IN_8_BITS(c) \
1249 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1250 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1251 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1252 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1253 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1254# define isPRINT_L1(c) (isPRINT_A(c) \
1255 || (FITS_IN_8_BITS(c) \
1256 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1257# define isPUNCT_L1(c) (isPUNCT_A(c) \
1258 || (FITS_IN_8_BITS(c) \
1259 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1260 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1261 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1262 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1263 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1264 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1265 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1266# define isSPACE_L1(c) (isSPACE_A(c) \
1267 || (FITS_IN_8_BITS(c) \
1268 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1269 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1270# define isUPPER_L1(c) (isUPPER_A(c) \
1271 || (FITS_IN_8_BITS(c) \
1272 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1273 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1274 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1275# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1276# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1277# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1278 || isBLANK_L1(c) \
1279 || (c) == '-' \
1280 || (c) == '(' \
1281 || (c) == ')')
1282 /* The following are not fully accurate in the above-ASCII range. I (khw)
1283 * don't think it's necessary to be so for the purposes where this gets
1284 * compiled */
1285# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1286# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1287
1288 /* And these aren't accurate at all. They are useful only for above
1289 * Latin1, which utilities and bootstrapping don't deal with */
1290# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1291# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1292# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1293
1294 /* Many of the macros later in this file are defined in terms of these. By
1295 * implementing them with a function, which converts the class number into
1296 * a call to the desired macro, all of the later ones work. However, that
1297 * function won't be actually defined when building a utility program (no
1298 * perl.h), and so a compiler error will be generated if one is attempted
1299 * to be used. And the above-Latin1 code points require Unicode tables to
1300 * be present, something unlikely to be the case when bootstrapping */
1301# define _generic_isCC(c, classnum) \
1302 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1303# define _generic_isCC_A(c, classnum) \
1304 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1305#endif /* End of no perl.h H_PERL */
8a58bdcf 1306
e66b99e9
KW
1307#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1308#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1309#define isASCII_A(c) isASCII(c)
1310#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1311#define isBLANK(c) isBLANK_A(c)
1312#define isCNTRL(c) isCNTRL_A(c)
1313#define isDIGIT(c) isDIGIT_A(c)
1314#define isGRAPH(c) isGRAPH_A(c)
1315#define isIDFIRST(c) isIDFIRST_A(c)
1316#define isLOWER(c) isLOWER_A(c)
1317#define isPRINT(c) isPRINT_A(c)
779cf272 1318#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1319#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1320#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1321#define isPUNCT(c) isPUNCT_A(c)
1322#define isSPACE(c) isSPACE_A(c)
1323#define isUPPER(c) isUPPER_A(c)
1324#define isWORDCHAR(c) isWORDCHAR_A(c)
1325#define isXDIGIT(c) isXDIGIT_A(c)
1326
1327/* ASCII casing. These could also be written as
1328 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1329 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1330 which uses table lookup and mask instead of subtraction. (This would
1331 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1332#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1333#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1334
25200305
KW
1335/* In the ASCII range, these are equivalent to what they're here defined to be.
1336 * But by creating these definitions, other code doesn't have to be aware of
1337 * this detail */
1338#define toFOLD(c) toLOWER(c)
25200305
KW
1339#define toTITLE(c) toUPPER(c)
1340
c753c8d3
KW
1341#define toLOWER_A(c) toLOWER(c)
1342#define toUPPER_A(c) toUPPER(c)
25200305
KW
1343#define toFOLD_A(c) toFOLD(c)
1344#define toTITLE_A(c) toTITLE(c)
1a0901db 1345
4650c663 1346/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1347#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1348 ? (c) \
f4cd282c 1349 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1350#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1351
1a0901db 1352/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1353 * all mapped to one of them, and these need special handling; returns the
1354 * input itself if is out-of-range */
b2bf251f 1355#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1356 ? (c) \
f4cd282c 1357 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1358#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1359
beab9ebe
KW
1360/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1361
1362/* For internal core Perl use only: the base macro for defining macros like
1363 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1364 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1365 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1366 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1367 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1368 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1369 * code points under locale rules, use the macros based on _generic_LC_uvchr
1370 * instead of this. */
beab9ebe
KW
1371#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1372 (! FITS_IN_8_BITS(c) \
1373 ? 0 \
31f05a37
KW
1374 : IN_UTF8_CTYPE_LOCALE \
1375 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1376 : cBOOL(non_utf8))
1377
1378/* For internal core Perl use only: a helper macro for defining macros like
1379 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1380 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1381 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1382 * _generic_LC_base for more info */
1383#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1384 _generic_LC_base(c,utf8_locale_classnum, \
1385 non_utf8_func( (_LC_CAST) (c)))
1386
1387/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1388 * 'c' is the platform's native underscore character */
1389#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1390 _generic_LC_base(c, utf8_locale_classnum, \
1391 (non_utf8_func( (_LC_CAST) (c)) \
1392 || (char)(c) == '_'))
1393
1394/* These next three are also for internal core Perl use only: case-change
1395 * helper macros */
1396#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1397 ? (c) \
31f05a37
KW
1398 : (IN_UTF8_CTYPE_LOCALE) \
1399 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1400 : (cast)function((cast)(c)))
beab9ebe 1401
31f05a37
KW
1402/* Note that the result can be larger than a byte in a UTF-8 locale. It
1403 * returns a single value, so can't adequately return the upper case of LATIN
1404 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1405 * values "SS"); instead it asserts against that under DEBUGGING, and
1406 * otherwise returns its input */
beab9ebe
KW
1407#define _generic_toUPPER_LC(c, function, cast) \
1408 (! FITS_IN_8_BITS(c) \
1409 ? (c) \
31f05a37 1410 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1411 ? (cast)function((cast)(c)) \
31f05a37
KW
1412 : ((((U8)(c)) == MICRO_SIGN) \
1413 ? GREEK_CAPITAL_LETTER_MU \
1414 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1415 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1416 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1417 ? (__ASSERT_(0) (c)) \
1418 : PL_mod_latin1_uc[ (U8) (c) ])))))
1419
1420/* Note that the result can be larger than a byte in a UTF-8 locale. It
1421 * returns a single value, so can't adequately return the fold case of LATIN
1422 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1423 * values "ss"); instead it asserts against that under DEBUGGING, and
1424 * otherwise returns its input */
beab9ebe 1425#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1426 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1427 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1428 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1429 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1430 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1431
84061b6a 1432/* Use the libc versions for these if available. */
f05550c0 1433#if defined(HAS_ISASCII)
84061b6a
KW
1434# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1435#else
1436# define isASCII_LC(c) isASCII(c)
1437#endif
1438
f05550c0 1439#if defined(HAS_ISBLANK)
84061b6a 1440# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1441#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1442# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1443#endif
1444
f05550c0 1445#define _LC_CAST U8
bbce6d69 1446
f05550c0 1447#ifdef WIN32
375f5f06
KW
1448 /* The Windows functions don't bother to follow the POSIX standard, which
1449 * for example says that something can't both be a printable and a control.
1450 * But Windows treats the \t control as a printable, and does such things
1451 * as making superscripts into both digits and punctuation. This tames
1452 * these flaws by assuming that the definitions of both controls and space
1453 * are correct, and then making sure that other definitions don't have
1454 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1455 * Not all possible weirdnesses are checked for, just the ones that were
1456 * detected on actual Microsoft code pages */
1457
b7d90381
KW
1458# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1459# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1460
1461# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1462 && isALPHANUMERIC_LC(c))
1463# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1464 ! isPUNCT_LC(c))
1465# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1466 isALPHANUMERIC_LC(c))
1467# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1468# define isIDFIRST_LC(c) (((c) == '_') \
1469 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1470# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1471# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1472# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1473# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1474# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1475# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1476 && isALPHANUMERIC_LC(c))
f05550c0
BF
1477
1478# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1479# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1480# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1481
1482#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1483 /* For most other platforms */
beab9ebe 1484
f05550c0
BF
1485# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1486# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1487# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1488# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1489# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1490# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1491# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1492# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1493# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1494# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1495# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1496# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1497# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1498
1499
1500# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1501# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1502# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1503
1504#else /* The final fallback position */
1505
b7d90381
KW
1506# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1507# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1508# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1509# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1510# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1511# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1512# define isLOWER_LC(c) (isascii(c) && islower(c))
1513# define isPRINT_LC(c) (isascii(c) && isprint(c))
1514# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1515# define isSPACE_LC(c) (isascii(c) && isspace(c))
1516# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1517# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1518# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1519
1520# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1521# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1522# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1523
f05550c0 1524#endif
55204971 1525
eba68aa0
KW
1526#define isIDCONT(c) isWORDCHAR(c)
1527#define isIDCONT_A(c) isWORDCHAR_A(c)
1528#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1529#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1530#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1531
4650c663 1532/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1533 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1534 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1535 * classes where there is a macro or function 'above_latin1' that takes the
1536 * single argument 'c' and returns the desired value. These exist for those
1537 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1538 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1539 * for classes where that overhead is faster than a direct lookup.
d0da05db 1540 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1541 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1542 * have duplicate checks here, so could create versions of the macros that
1543 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1544
1545/* Note that all ignore 'use bytes' */
d0da05db 1546#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
cd500f2f
KW
1547 ? _generic_isCC(c, classnum) \
1548 : above_latin1(c))
d0da05db 1549#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
cd500f2f 1550 ? _generic_isCC(c, classnum) \
922e8cb4 1551 : _is_uni_FOO(classnum, c))
d0da05db
KW
1552#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1553#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1554#define isASCII_uvchr(c) isASCII(c)
1555#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1556#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1557#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1558#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1559#define isIDCONT_uvchr(c) _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1560#define isIDFIRST_uvchr(c) _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
1561#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1562#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1563
1564#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1565#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1566#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1567
1568#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1569#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1570#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1571#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1572
1573#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1574#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1575#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1576#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1577
1578/* For backwards compatibility, even though '_uni' should mean official Unicode
1579 * code points, in Perl it means native for those below 256 */
1580#define isALPHA_uni(c) isALPHA_uvchr(c)
1581#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1582#define isASCII_uni(c) isASCII_uvchr(c)
1583#define isBLANK_uni(c) isBLANK_uvchr(c)
1584#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1585#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1586#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1587#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1588#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1589#define isLOWER_uni(c) isLOWER_uvchr(c)
1590#define isPRINT_uni(c) isPRINT_uvchr(c)
1591#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1592#define isSPACE_uni(c) isSPACE_uvchr(c)
1593#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1594#define isUPPER_uni(c) isUPPER_uvchr(c)
1595#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1596#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1597#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1598#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1599#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1600#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1601#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1602
4650c663
KW
1603/* For internal core Perl use only: the base macros for defining macros like
1604 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1605 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1606 * simple class definitions; the other for more complex. These are like
d0da05db 1607 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1608#define _generic_LC_uvchr(latin1, above_latin1, c) \
1609 (c < 256 ? latin1(c) : above_latin1(c))
1610#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1611 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1612
1613#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1614#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1615 _CC_ALPHANUMERIC, c)
b7d90381
KW
1616#define isASCII_LC_uvchr(c) isASCII_LC(c)
1617#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1618 is_HORIZWS_cp_high, c)
feeab5a9 1619#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1620#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1621#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1622#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1623 _is_uni_perl_idcont, c)
b7d90381 1624#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1625 _is_uni_perl_idstart, c)
1626#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1627#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1628#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1629#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1630#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1631 is_XPERLSPACE_cp_high, c)
cd500f2f 1632#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1633#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1634 _CC_WORDCHAR, c)
b7d90381
KW
1635#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1636 is_XDIGIT_cp_high, c)
e712593e 1637
b7d90381 1638#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1639
4650c663
KW
1640/* For internal core Perl use only: the base macros for defining macros like
1641 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1642 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1643 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1644 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1645 * same representation whether utf8 or not. Note that it assumes that the utf8
1646 * has been validated, and ignores 'use bytes' */
cd500f2f 1647#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
aea0c08c 1648 ? _generic_isCC(*(p), classnum) \
7d7a6efc 1649 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
aea0c08c 1650 ? _generic_isCC( \
a62b247b 1651 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
aea0c08c 1652 *((p)+1 )), \
94bb8c36 1653 classnum) \
252810d0 1654 : utf8)
b7d90381
KW
1655/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1656 * 'above_latin1' can be a macro */
cd500f2f
KW
1657#define _generic_func_utf8(classnum, above_latin1, p) \
1658 _generic_utf8(classnum, p, above_latin1(p))
4650c663 1659/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1660 * 'above_latin1' parameter */
922e8cb4 1661#define _generic_swash_utf8(classnum, p) \
cd500f2f 1662 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
922e8cb4 1663
cc8ab7c0 1664/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1665 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1666 * class is TRUE for. Hence it can skip the tests for this range.
1667 * 'above_latin1' should include its arguments */
3ded5eb0 1668#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
cc8ab7c0
KW
1669 (UTF8_IS_INVARIANT(*(p)) \
1670 ? _generic_isCC(*(p), classnum) \
1671 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
3ded5eb0 1672 ? above_latin1 \
cc8ab7c0
KW
1673 : 0)
1674
84238efa
KW
1675/* NOTE that some of these macros have very similar ones in regcharclass.h.
1676 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1677 * there, differing in name only by an underscore from the one here
1678 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1679 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1680 * points; the regcharclass.h ones are implemented as a series of
1681 * "if-else-if-else ..." */
1682
b7d90381
KW
1683#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1684#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1685#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
509fb054
KW
1686 utf8, the non-utf8 macro works
1687 */
b7d90381 1688#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
e8fa43e2
KW
1689
1690#ifdef EBCDIC
1691 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1692 * more efficient macro instead of the more general one */
88794300 1693# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
e8fa43e2
KW
1694#else
1695# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1696#endif
1697
cd500f2f
KW
1698#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1699 _is_utf8_FOO(_CC_DIGIT, p))
922e8cb4 1700#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
eba68aa0 1701#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
2f4622f0 1702 _is_utf8_perl_idcont, p)
e5dcd934 1703
c11ff943
KW
1704/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1705 * IDFIRST is an alnum. See
c8362b00 1706 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1707 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1708 * This used to be not the XID version, but we decided to go with the more
1709 * modern Unicode definition */
b7d90381 1710#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
509fb054 1711 _is_utf8_perl_idstart, p)
e5dcd934 1712
b7d90381
KW
1713#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1714#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1715#define isPSXSPC_utf8(p) isSPACE_utf8(p)
1716#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1717#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1718#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1719#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1720#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1721#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
cd500f2f 1722 is_XDIGIT_high(p))
a0ed51b3 1723
7d2b61dd 1724#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1725#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1726#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1727#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1728
4650c663
KW
1729/* For internal core Perl use only: the base macros for defining macros like
1730 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1731 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1732 * 'macro' parameter */
94bb8c36
KW
1733#define _generic_LC_utf8(macro, p, utf8) \
1734 (UTF8_IS_INVARIANT(*(p)) \
1735 ? macro(*(p)) \
1736 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
a62b247b 1737 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
9201a572
KW
1738 : utf8)
1739
cd500f2f
KW
1740#define _generic_LC_swash_utf8(macro, classnum, p) \
1741 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1742#define _generic_LC_func_utf8(macro, above_latin1, p) \
4650c663 1743 _generic_LC_utf8(macro, p, above_latin1(p))
9201a572 1744
b7d90381 1745#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
cd500f2f 1746 _CC_ALPHANUMERIC, p)
b7d90381
KW
1747#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1748#define isASCII_LC_utf8(p) isASCII_LC(*p)
1749#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1750 is_HORIZWS_high, p)
1751#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1752#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1753#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1754#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1755 _is_utf8_perl_idcont, p)
1756#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1757 _is_utf8_perl_idstart, p)
1758#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1759#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1760#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1761#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1762#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1763 is_XPERLSPACE_high, p)
1764#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
cd500f2f
KW
1765#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1766 _CC_WORDCHAR, p)
b7d90381
KW
1767#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1768 is_XDIGIT_high, p)
aaa51d5e 1769
fbc19f27
KW
1770/* Macros for backwards compatibility and for completeness when the ASCII and
1771 * Latin1 values are identical */
b7d90381
KW
1772#define isALPHAU(c) isALPHA_L1(c)
1773#define isDIGIT_L1(c) isDIGIT_A(c)
1774#define isOCTAL(c) isOCTAL_A(c)
1775#define isOCTAL_L1(c) isOCTAL_A(c)
1776#define isXDIGIT_L1(c) isXDIGIT_A(c)
1777#define isALNUM(c) isWORDCHAR(c)
1778#define isALNUMU(c) isWORDCHAR_L1(c)
1779#define isALNUM_LC(c) isWORDCHAR_LC(c)
1780#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 1781#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
1782#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1783#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1784#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1785#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1786#define isALNUMC(c) isALPHANUMERIC(c)
1787#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1788#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 1789#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 1790#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 1791#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 1792
2bd1cbf6
KW
1793/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1794 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1795 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1796 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1797 * the outlier from the block that contains the other controls, just like
1798 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1799 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
1800 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1801 * etc. */
2bd1cbf6 1802#ifndef EBCDIC
75763b3a 1803# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 1804#else
75763b3a
KW
1805# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1806 ((isPRINT_A(c)) \
1807 ? (UNLIKELY((c) == '?') \
1808 ? QUESTION_MARK_CTRL \
1809 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1810 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1811 ? '?' \
1812 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 1813#endif
bbce6d69 1814
dea28490
JJ
1815/* Line numbers are unsigned, 32 bits. */
1816typedef U32 line_t;
e5dcd934 1817#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 1818
91152fc1
DG
1819/* Helpful alias for version prescan */
1820#define is_LAX_VERSION(a,b) \
1821 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1822
1823#define is_STRICT_VERSION(a,b) \
1824 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1825
1826#define BADVERSION(a,b,c) \
1827 if (b) { \
1828 *b = c; \
1829 } \
1830 return a;
8c52afec 1831
cb27eebd
KW
1832/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1833 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1834 * which is advanced. The input is validated only by an assert() in DEBUGGING
1835 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1836 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1837#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1838 ? (c) \
1839 : ((c) + 9))))
902008b8
KW
1840#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1841 ? (*(s)++) \
1842 : (*(s)++ + 9))))
95a59cab 1843
cb27eebd
KW
1844/* Converts a character known to represent an octal digit (0-7) to its numeric
1845 * value. The input is validated only by an assert() in DEBUGGING builds. In
1846 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1847#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1848
305b8651
KW
1849/* Efficiently returns a boolean as to if two native characters are equivalent
1850 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1851 * the ALPHA in the name is to remind you of that. This is asserted() in
1852 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1853 * works (on valid input) for both non- and UTF-8-encoded bytes.
1854 *
1855 * When one of the inputs is a compile-time constant and gets folded by the
1856 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1857 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1858 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
1859 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1860 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1861 * just a single 0, in the bit position where the upper- and lowercase differ.
1862 * */
305b8651
KW
1863#define isALPHA_FOLD_EQ(c1, c2) \
1864 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1865 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1866#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1867
8e84507e 1868/*
ccfc67b7
JH
1869=head1 Memory Management
1870
a02a5408 1871=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
1872The XSUB-writer's interface to the C C<malloc> function.
1873
596f7718 1874Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1875
c5008215
JC
1876In 5.9.3, Newx() and friends replace the older New() API, and drops
1877the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 1878themselves. This aid has been superseded by a new build option,
d10b4965 1879PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
1880there for use in XS modules supporting older perls.
1881
a02a5408 1882=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 1883The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 1884cast. See also C<L</Newx>>.
954c1994 1885
596f7718 1886Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1887
a02a5408 1888=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 1889The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 1890memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 1891
596f7718 1892Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1893
954c1994
GS
1894=for apidoc Am|void|Renew|void* ptr|int nitems|type
1895The XSUB-writer's interface to the C C<realloc> function.
1896
596f7718 1897Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1898
954c1994
GS
1899=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1900The XSUB-writer's interface to the C C<realloc> function, with
1901cast.
1902
596f7718 1903Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 1904
49b8b560 1905=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
1906The XSUB-writer's interface to the C C<free> function.
1907
596f7718 1908This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 1909
954c1994
GS
1910=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1911The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 1912source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1913C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 1914
e90e2364 1915=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 1916Like C<Move> but returns C<dest>. Useful
72d33970 1917for encouraging compilers to tail-call
e90e2364
NC
1918optimise.
1919
954c1994
GS
1920=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1921The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 1922source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 1923C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 1924
e90e2364
NC
1925=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1926
796b6530 1927Like C<Copy> but returns C<dest>. Useful
72d33970 1928for encouraging compilers to tail-call
e90e2364
NC
1929optimise.
1930
954c1994
GS
1931=for apidoc Am|void|Zero|void* dest|int nitems|type
1932
1933The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1934destination, C<nitems> is the number of items, and C<type> is the type.
1935
e90e2364
NC
1936=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1937
72d33970
FC
1938Like C<Zero> but returns dest. Useful
1939for encouraging compilers to tail-call
e90e2364
NC
1940optimise.
1941
da5d8dbb 1942=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 1943This is an architecture-independent macro to copy one structure to another.
954c1994 1944
7e337ee0
JH
1945=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1946
1947Fill up memory with a byte pattern (a byte repeated over and over
1948again) that hopefully catches attempts to access uninitialized memory.
1949
1950=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1951
1952PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1953
1c12ffb4 1954=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
1955
1956PoisonWith(0xEF) for catching access to freed memory.
1957
9965345d
JH
1958=for apidoc Am|void|Poison|void* dest|int nitems|type
1959
7e337ee0 1960PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
1961
1962=cut */
954c1994 1963
561b68a9
SH
1964/* Maintained for backwards-compatibility only. Use newSV() instead. */
1965#ifndef PERL_CORE
ff06c60c 1966#define NEWSV(x,len) newSV(len)
561b68a9 1967#endif
ff06c60c 1968
19a94d75
JH
1969#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1970
e6bdf523 1971
27d5b266 1972#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
1973
1974/* This expression will be constant-folded at compile time. It checks
1975 * whether or not the type of the count n is so small (e.g. U8 or U16, or
1976 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
1977 * As well as avoiding the need for a run-time check in some cases, it's
1978 * designed to avoid compiler warnings like:
1979 * comparison is always false due to limited range of data type
73e8ff00
DM
1980 * It's mathematically equivalent to
1981 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
1982 */
1983
1984# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
004073ba 1985 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
e6bdf523 1986
88f9f128 1987/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
1988 * compiler warnings. We *want* to write the expression as
1989 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
1990 * (for some compile-time constant C), but even when the LHS
1991 * constant-folds to false at compile-time, g++ insists on emitting
1992 * warnings about the RHS (e.g. "comparison is always false"), so instead
1993 * we write it as
e6bdf523 1994 *
d98e5cde 1995 * (cond ? n : X) > C
88f9f128 1996 *
d98e5cde
DM
1997 * where X is a constant with X > C always false. Choosing a value for X
1998 * is tricky. If 0, some compilers will complain about 0 > C always being
1999 * false; if 1, Coverity complains when n happens to be the constant value
2000 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2001 * for X and hope that nothing else whines.
e6bdf523
DM
2002 */
2003
2004# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2005 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2006 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2007
2008# define MEM_WRAP_CHECK(n,t) \
2009 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2010 && (croak_memory_wrap(),0))
2011
2012# define MEM_WRAP_CHECK_1(n,t,a) \
2013 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2014 && (Perl_croak_nocontext("%s",(a)),0))
2015
8b44ba4c 2016#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2017
d1decf2b 2018#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
2019#else
2020
410319be
NC
2021#define MEM_WRAP_CHECK(n,t)
2022#define MEM_WRAP_CHECK_1(n,t,a)
2023#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
2024#define MEM_WRAP_CHECK_(n,t)
2025
6fff79ce 2026#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 2027
1936d2a7 2028#endif
8b44ba4c 2029
fe4f188c 2030#ifdef PERL_MEM_LOG
46c6c7e2 2031/*
9f653bb5 2032 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2033 * go through functions, which are handy for debugging breakpoints, but
2034 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2035 * line number, and C function name if available) passed in. This info can
2036 * then be used for logging the calls, for which one gets a sample
73d1d973 2037 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2038 *
46c6c7e2 2039 * Known problems:
94e892a6 2040 * - not all memory allocs get logged, only those
46c6c7e2 2041 * that go through Newx() and derivatives (while all
94e892a6 2042 * Safefrees do get logged)
46c6c7e2
JH
2043 * - __FILE__ and __LINE__ do not work everywhere
2044 * - __func__ or __FUNCTION__ even less so
2045 * - I think more goes on after the perlio frees but
2046 * the thing is that STDERR gets closed (as do all
2047 * the file descriptors)
2048 * - no deeper calling stack than the caller of the Newx()
2049 * or the kind, but do I look like a C reflection/introspection
2050 * utility to you?
2051 * - the function prototypes for the logging functions
2052 * probably should maybe be somewhere else than handy.h
2053 * - one could consider inlining (macrofying) the logging
2054 * for speed, but I am too lazy
2055 * - one could imagine recording the allocations in a hash,
2056 * (keyed by the allocation address?), and maintain that
2057 * through reallocs and frees, but how to do that without
2058 * any News() happening...?
73d1d973 2059 * - lots of -Ddefines to get useful/controllable output
b953482e 2060 * - lots of ENV reads
46c6c7e2
JH
2061 */
2062
0b0ab801 2063# ifdef PERL_CORE
73d1d973 2064# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2065enum mem_log_type {
2066 MLT_ALLOC,
2067 MLT_REALLOC,
d7a2c63c
MHM
2068 MLT_FREE,
2069 MLT_NEW_SV,
2070 MLT_DEL_SV
0b0ab801
MHM
2071};
2072# endif
12754f92 2073# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2074void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2075void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2076# endif
0b0ab801
MHM
2077# endif
2078
fe4f188c
JH
2079#endif
2080
2081#ifdef PERL_MEM_LOG
d1401ee9
MHM
2082#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2083#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2084#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2085#endif
2086
2087#ifndef MEM_LOG_ALLOC
2088#define MEM_LOG_ALLOC(n,t,a) (a)
2089#endif
2090#ifndef MEM_LOG_REALLOC
2091#define MEM_LOG_REALLOC(n,t,v,a) (a)
2092#endif
2093#ifndef MEM_LOG_FREE
2094#define MEM_LOG_FREE(a) (a)
2095#endif
2096
d1401ee9
MHM
2097#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2098#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2099#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2100
2101#ifndef PERL_CORE
a02a5408
JC
2102/* pre 5.9.x compatibility */
2103#define New(x,v,n,t) Newx(v,n,t)
2104#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2105#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2106#endif
a02a5408 2107
ff68c719 2108#define Renew(v,n,t) \
d1401ee9 2109 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2110#define Renewc(v,n,t,c) \
d1401ee9 2111 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2112
2113#ifdef PERL_POISON
2114#define Safefree(d) \
06c0cc96 2115 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2116#else
fe4f188c 2117#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2118#endif
55497cff 2119
8b44ba4c
NC
2120#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2121#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2122#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
55497cff 2123
8b44ba4c
NC
2124#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2125#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
e90e2364 2126#ifdef HAS_MEMSET
8b44ba4c 2127#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2128#else
8b44ba4c
NC
2129/* Using bzero(), which returns void. */
2130#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
e90e2364
NC
2131#endif
2132
7e337ee0
JH
2133#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2134#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2135#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2136#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2137
caa674f3
DD
2138#ifdef PERL_POISON
2139# define PERL_POISON_EXPR(x) x
2140#else
2141# define PERL_POISON_EXPR(x)
2142#endif
2143
2304df62 2144#ifdef USE_STRUCT_COPY
ff68c719 2145#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
bee1dbe2
LW
2146#else
2147#define StructCopy(s,d,t) Copy(s,d,1,t)
2148#endif
2cc61e15 2149
c3caa5c3
JH
2150/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2151 * want your zero-based indices to be less than but not equal to).
2152 *
2153 * C_ARRAY_END is one past the last: half-open/half-closed range,
2154 * not last-inclusive range. */
622913ab 2155#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2156#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2157
2cc61e15
DD
2158#ifdef NEED_VA_COPY
2159# ifdef va_copy
2160# define Perl_va_copy(s, d) va_copy(d, s)
2cc61e15 2161# else
a1866d1b
JH
2162# if defined(__va_copy)
2163# define Perl_va_copy(s, d) __va_copy(d, s)
2164# else
2165# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2166# endif
2cc61e15
DD
2167# endif
2168#endif
2169
472d47bc
SB
2170/* convenience debug macros */
2171#ifdef USE_ITHREADS
2172#define pTHX_FORMAT "Perl interpreter: 0x%p"
2173#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2174#define pTHX_VALUE_ (void *)my_perl,
2175#define pTHX_VALUE (void *)my_perl
2176#define pTHX__VALUE_ ,(void *)my_perl,
2177#define pTHX__VALUE ,(void *)my_perl
472d47bc 2178#else
3609ea0d 2179#define pTHX_FORMAT
472d47bc 2180#define pTHX__FORMAT
3609ea0d 2181#define pTHX_VALUE_
472d47bc 2182#define pTHX_VALUE
3609ea0d 2183#define pTHX__VALUE_
472d47bc
SB
2184#define pTHX__VALUE
2185#endif /* USE_ITHREADS */
3609ea0d 2186
2acdbac1
NC
2187/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2188 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2189 CPAN::Unpack show any users outside the core. */
2190#ifdef PERL_CORE
b7d90381
KW
2191# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2192 "Use of " s " is deprecated")
2acdbac1
NC
2193#endif
2194
dfff4baf
BF
2195/* Internal macros to deal with gids and uids */
2196#ifdef PERL_CORE
2197
2198# if Uid_t_size > IVSIZE
2199# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2200# define SvUID(sv) SvNV(sv)
2201# else
2202# if Uid_t_sign <= 0
2203# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2204# define SvUID(sv) SvIV(sv)
2205# else
2206# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2207# define SvUID(sv) SvUV(sv)
2208# endif
2209# endif /* Uid_t_size */
2210
2211# if Gid_t_size > IVSIZE
2212# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2213# define SvGID(sv) SvNV(sv)
2214# else
2215# if Gid_t_sign <= 0
2216# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2217# define SvGID(sv) SvIV(sv)
2218# else
2219# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2220# define SvGID(sv) SvUV(sv)
2221# endif
2222# endif /* Gid_t_size */
2223
2224#endif
2225
9d745869
KW
2226#endif /* HANDY_H */
2227
e9a8c099 2228/*
14d04a33 2229 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2230 */