This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
handy.h: Make uniform the defns of strFOO, memFOO
[perl5.git] / handy.h
CommitLineData
a0d0e21e 1/* handy.h
a687059c 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
da5d8dbb 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
a687059c 5 *
6e21c824
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8d063cd8 8 *
8d063cd8
LW
9 */
10
4650c663
KW
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
6a5bc5ac
KW
14#ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */
15#define PERL_HANDY_H_
9d745869 16
24792b8d
NC
17#ifndef PERL_CORE
18# define Null(type) ((type)NULL)
954c1994
GS
19
20/*
ccfc67b7 21=head1 Handy Values
954c1994 22
3609ea0d 23=for apidoc AmU||Nullch
72d33970
FC
24Null character pointer. (No longer available when C<PERL_CORE> is
25defined.)
2307c6d0 26
954c1994 27=for apidoc AmU||Nullsv
72d33970 28Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
954c1994
GS
29
30=cut
31*/
32
24792b8d
NC
33# define Nullch Null(char*)
34# define Nullfp Null(PerlIO*)
35# define Nullsv Null(SV*)
36#endif
8d063cd8 37
641d3f0b
PP
38#ifdef TRUE
39#undef TRUE
40#endif
41#ifdef FALSE
42#undef FALSE
43#endif
44#define TRUE (1)
45#define FALSE (0)
46
cf3f0ffb
DM
47/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
48 * (compiler permitting) that casting away const-ness will give a warning;
49 * e.g.:
50 *
51 * const SV *sv = ...;
52 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
53 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
54 */
55
b1bc3f34 56#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
f065df04 57# define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
b1bc3f34
NC
58#else
59# define MUTABLE_PTR(p) ((void *) (p))
60#endif
61
a062e10d 62#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
ea726b52 63#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
159b6efe 64#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
dbebbdb4 65#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
a45c7426 66#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
b1bc3f34 67#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
27d4fb96 68
f789f6a4 69#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
bd31be4b
NC
70# include <stdbool.h>
71# ifndef HAS_BOOL
72# define HAS_BOOL 1
73# endif
74#endif
75
8e84507e 76/* bool is built-in for g++-2.6.3 and later, which might be used
c1d22f6b
GS
77 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
78 be sure _G_config.h will be included before this file. _G_config.h
8e84507e 79 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
c1d22f6b
GS
80 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
81 g++ can be identified by __GNUG__.
82 Andy Dougherty February 2000
5d94fbed 83*/
3609ea0d 84#ifdef __GNUG__ /* GNU g++ has bool built-in */
f789f6a4 85# ifndef PERL_BOOL_AS_CHAR
5d94fbed 86# ifndef HAS_BOOL
c1d22f6b 87# define HAS_BOOL 1
5d94fbed 88# endif
f789f6a4 89# endif
5d94fbed
AD
90#endif
91
92#ifndef HAS_BOOL
f789f6a4
FC
93# ifdef bool
94# undef bool
95# endif
70d5cb32 96# define bool char
c1d22f6b 97# define HAS_BOOL 1
a687059c 98#endif
0d3e774c 99
236415e2
KW
100/* cast-to-bool. A simple (bool) cast may not do the right thing: if bool is
101 * defined as char for example, then the cast from int is
102 * implementation-defined (bool)!!(cbool) in a ternary triggers a bug in xlc on
103 * AIX */
18f5643b 104#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
f2338a2e 105
46c6c7e2 106/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
e352bcff
JH
107 * XXX Should really be a Configure probe, with HAS__FUNCTION__
108 * and FUNCTION__ as results.
109 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
46c6c7e2
JH
110#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
111# define FUNCTION__ __func__
07798b17
AC
112#elif (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
113 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
114# define FUNCTION__ ""
46c6c7e2 115#else
07798b17 116# define FUNCTION__ __FUNCTION__ /* Common extension. */
46c6c7e2
JH
117#endif
118
27d4fb96
PP
119/* XXX A note on the perl source internal type system. The
120 original intent was that I32 be *exactly* 32 bits.
121
122 Currently, we only guarantee that I32 is *at least* 32 bits.
123 Specifically, if int is 64 bits, then so is I32. (This is the case
124 for the Cray.) This has the advantage of meshing nicely with
125 standard library calls (where we pass an I32 and the library is
126 expecting an int), but the disadvantage that an I32 is not 32 bits.
127 Andy Dougherty August 1996
24fef2a7 128
dc45a647
MB
129 There is no guarantee that there is *any* integral type with
130 exactly 32 bits. It is perfectly legal for a system to have
131 sizeof(short) == sizeof(int) == sizeof(long) == 8.
693762b4 132
dc45a647
MB
133 Similarly, there is no guarantee that I16 and U16 have exactly 16
134 bits.
693762b4 135
8e84507e
NIS
136 For dealing with issues that may arise from various 32/64-bit
137 systems, we will ask Configure to check out
8175356b 138
3609ea0d
JH
139 SHORTSIZE == sizeof(short)
140 INTSIZE == sizeof(int)
141 LONGSIZE == sizeof(long)
dc45a647 142 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
3609ea0d 143 PTRSIZE == sizeof(void *)
dc45a647
MB
144 DOUBLESIZE == sizeof(double)
145 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
8175356b 146
27d4fb96
PP
147*/
148
69512466
JH
149#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
150# include <inttypes.h>
dd0eed91
JH
151# ifdef INT32_MIN_BROKEN
152# undef INT32_MIN
153# define INT32_MIN (-2147483647-1)
154# endif
155# ifdef INT64_MIN_BROKEN
156# undef INT64_MIN
157# define INT64_MIN (-9223372036854775807LL-1)
158# endif
69512466
JH
159#endif
160
8175356b
JH
161typedef I8TYPE I8;
162typedef U8TYPE U8;
163typedef I16TYPE I16;
164typedef U16TYPE U16;
165typedef I32TYPE I32;
166typedef U32TYPE U32;
16d89be8
JH
167
168#ifdef HAS_QUAD
8175356b
JH
169typedef I64TYPE I64;
170typedef U64TYPE U64;
16d89be8 171#endif
8175356b 172
9c39b21c
JH
173/* INT64_C/UINT64_C are C99 from <stdint.h> (so they will not be
174 * available in strict C89 mode), but they are nice, so let's define
175 * them if necessary. */
0d3bd0a9 176#if defined(HAS_QUAD)
9c39b21c
JH
177# undef PeRl_INT64_C
178# undef PeRl_UINT64_C
179/* Prefer the native integer types (int and long) over long long
180 * (which is not C89) and Win32-specific __int64. */
181# if QUADKIND == QUAD_IS_INT && INTSIZE == 8
182# define PeRl_INT64_C(c) (c)
183# define PeRl_UINT64_C(c) CAT2(c,U)
184# endif
185# if QUADKIND == QUAD_IS_LONG && LONGSIZE == 8
186# define PeRl_INT64_C(c) CAT2(c,L)
187# define PeRl_UINT64_C(c) CAT2(c,UL)
188# endif
189# if QUADKIND == QUAD_IS_LONG_LONG && defined(HAS_LONG_LONG)
190# define PeRl_INT64_C(c) CAT2(c,LL)
191# define PeRl_UINT64_C(c) CAT2(c,ULL)
192# endif
193# if QUADKIND == QUAD_IS___INT64
194# define PeRl_INT64_C(c) CAT2(c,I64)
195# define PeRl_UINT64_C(c) CAT2(c,UI64)
196# endif
197# ifndef PeRl_INT64_C
702c92eb
JH
198# define PeRl_INT64_C(c) ((I64)(c)) /* last resort */
199# define PeRl_UINT64_C(c) ((U64)(c))
9c39b21c
JH
200# endif
201/* In OS X the INT64_C/UINT64_C are defined with LL/ULL, which will
202 * not fly with C89-pedantic gcc, so let's undefine them first so that
203 * we can redefine them with our native integer preferring versions. */
8eafc6eb 204# if defined(PERL_DARWIN) && defined(PERL_GCC_PEDANTIC)
9c39b21c
JH
205# undef INT64_C
206# undef UINT64_C
207# endif
208# ifndef INT64_C
209# define INT64_C(c) PeRl_INT64_C(c)
210# endif
211# ifndef UINT64_C
212# define UINT64_C(c) PeRl_UINT64_C(c)
213# endif
e8c95190 214#endif
dc750f2e 215
d8668976 216#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
5ff3f7a4 217
5ff3f7a4
GS
218/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
219 Please search CHAR_MAX in perl.h for further details. */
220#define U8_MAX UINT8_MAX
221#define U8_MIN UINT8_MIN
222
5ff3f7a4
GS
223#define I16_MAX INT16_MAX
224#define I16_MIN INT16_MIN
225#define U16_MAX UINT16_MAX
226#define U16_MIN UINT16_MIN
227
5ff3f7a4
GS
228#define I32_MAX INT32_MAX
229#define I32_MIN INT32_MIN
0e983133
GS
230#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
231# define U32_MAX UINT32_MAX
232#else
233# define U32_MAX 4294967295U
234#endif
5ff3f7a4
GS
235#define U32_MIN UINT32_MIN
236
237#else
238
5c9fa16e
KA
239/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
240 Please search CHAR_MAX in perl.h for further details. */
27d4fb96
PP
241#define U8_MAX PERL_UCHAR_MAX
242#define U8_MIN PERL_UCHAR_MIN
79072805 243
27d4fb96
PP
244#define I16_MAX PERL_SHORT_MAX
245#define I16_MIN PERL_SHORT_MIN
246#define U16_MAX PERL_USHORT_MAX
247#define U16_MIN PERL_USHORT_MIN
79072805 248
c4f23d77 249#if LONGSIZE > 4
27d4fb96
PP
250# define I32_MAX PERL_INT_MAX
251# define I32_MIN PERL_INT_MIN
252# define U32_MAX PERL_UINT_MAX
253# define U32_MIN PERL_UINT_MIN
79072805 254#else
27d4fb96
PP
255# define I32_MAX PERL_LONG_MAX
256# define I32_MIN PERL_LONG_MIN
257# define U32_MAX PERL_ULONG_MAX
258# define U32_MIN PERL_ULONG_MIN
79072805
LW
259#endif
260
5ff3f7a4
GS
261#endif
262
58a9a5d5 263/* log(2) is pretty close to 0.30103, just in case anyone is grepping for it */
fc36a67e
PP
264#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
265#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
266#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
267
88794300 268/* Unused by core; should be deprecated */
ff68c719 269#define Ctl(ch) ((ch) & 037)
8d063cd8 270
98fce2a4
KW
271#if defined(PERL_CORE) || defined(PERL_EXT)
272# ifndef MIN
273# define MIN(a,b) ((a) < (b) ? (a) : (b))
274# endif
275# ifndef MAX
276# define MAX(a,b) ((a) > (b) ? (a) : (b))
277# endif
278#endif
279
4650c663
KW
280/* This is a helper macro to avoid preprocessor issues, replaced by nothing
281 * unless under DEBUGGING, where it expands to an assert of its argument,
282 * followed by a comma (hence the comma operator). If we just used a straight
3e94db23
JH
283 * assert(), we would get a comma with nothing before it when not DEBUGGING.
284 *
285 * We also use empty definition under Coverity since the __ASSERT__
286 * checks often check for things that Really Cannot Happen, and Coverity
287 * detects that and gets all excited. */
288
289#if defined(DEBUGGING) && !defined(__COVERITY__)
0f092d08
KW
290# define __ASSERT_(statement) assert(statement),
291#else
292# define __ASSERT_(statement)
293#endif
294
3fe05580
MHM
295/*
296=head1 SV-Body Allocation
297
298=for apidoc Ama|SV*|newSVpvs|const char* s
0c395ea5 299Like C<newSVpvn>, but takes a C<NUL>-terminated literal string instead of a
30a15352 300string/length pair.
3fe05580 301
84bafc02 302=for apidoc Ama|SV*|newSVpvs_flags|const char* s|U32 flags
0c395ea5 303Like C<newSVpvn_flags>, but takes a C<NUL>-terminated literal string instead of
30a15352 304a string/length pair.
84bafc02 305
3fe05580 306=for apidoc Ama|SV*|newSVpvs_share|const char* s
0c395ea5 307Like C<newSVpvn_share>, but takes a C<NUL>-terminated literal string instead of
30a15352 308a string/length pair and omits the hash parameter.
3fe05580 309
9dcc53ea 310=for apidoc Am|void|sv_catpvs_flags|SV* sv|const char* s|I32 flags
0c395ea5 311Like C<sv_catpvn_flags>, but takes a C<NUL>-terminated literal string instead
30a15352 312of a string/length pair.
9dcc53ea
Z
313
314=for apidoc Am|void|sv_catpvs_nomg|SV* sv|const char* s
0c395ea5
KW
315Like C<sv_catpvn_nomg>, but takes a C<NUL>-terminated literal string instead of
316a string/length pair.
9dcc53ea 317
7b4e7469 318=for apidoc Am|void|sv_catpvs|SV* sv|const char* s
0c395ea5
KW
319Like C<sv_catpvn>, but takes a C<NUL>-terminated literal string instead of a
320string/length pair.
3fe05580 321
9dcc53ea 322=for apidoc Am|void|sv_catpvs_mg|SV* sv|const char* s
0c395ea5 323Like C<sv_catpvn_mg>, but takes a C<NUL>-terminated literal string instead of a
9dcc53ea
Z
324string/length pair.
325
7b4e7469 326=for apidoc Am|void|sv_setpvs|SV* sv|const char* s
0c395ea5
KW
327Like C<sv_setpvn>, but takes a C<NUL>-terminated literal string instead of a
328string/length pair.
3fe05580 329
9dcc53ea 330=for apidoc Am|void|sv_setpvs_mg|SV* sv|const char* s
0c395ea5 331Like C<sv_setpvn_mg>, but takes a C<NUL>-terminated literal string instead of a
9dcc53ea
Z
332string/length pair.
333
334=for apidoc Am|SV *|sv_setref_pvs|const char* s
0c395ea5
KW
335Like C<sv_setref_pvn>, but takes a C<NUL>-terminated literal string instead of
336a string/length pair.
9dcc53ea 337
3fe05580
MHM
338=head1 Memory Management
339
340=for apidoc Ama|char*|savepvs|const char* s
0c395ea5 341Like C<savepvn>, but takes a C<NUL>-terminated literal string instead of a
30a15352 342string/length pair.
3fe05580 343
9dcc53ea
Z
344=for apidoc Ama|char*|savesharedpvs|const char* s
345A version of C<savepvs()> which allocates the duplicate string in memory
346which is shared between threads.
347
3fe05580
MHM
348=head1 GV Functions
349
350=for apidoc Am|HV*|gv_stashpvs|const char* name|I32 create
0c395ea5
KW
351Like C<gv_stashpvn>, but takes a C<NUL>-terminated literal string instead of a
352string/length pair.
3fe05580
MHM
353
354=head1 Hash Manipulation Functions
355
356=for apidoc Am|SV**|hv_fetchs|HV* tb|const char* key|I32 lval
0c395ea5
KW
357Like C<hv_fetch>, but takes a C<NUL>-terminated literal string instead of a
358string/length pair.
3fe05580
MHM
359
360=for apidoc Am|SV**|hv_stores|HV* tb|const char* key|NULLOK SV* val
0c395ea5
KW
361Like C<hv_store>, but takes a C<NUL>-terminated literal string instead of a
362string/length pair
3fe05580
MHM
363and omits the hash parameter.
364
510966aa
Z
365=head1 Lexer interface
366
367=for apidoc Amx|void|lex_stuff_pvs|const char *pv|U32 flags
368
0c395ea5
KW
369Like L</lex_stuff_pvn>, but takes a C<NUL>-terminated literal string instead of
370a string/length pair.
510966aa 371
3fe05580
MHM
372=cut
373*/
374
926bb54c
KW
375/* concatenating with "" ensures that only literal strings are accepted as
376 * argument */
26bfd924 377#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
2efa8cc7 378
926bb54c
KW
379/* note that STR_WITH_LEN() can't be used as argument to macros or functions
380 * that under some configurations might be macros, which means that it requires
381 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
ba3a79e7
GA
382 */
383
384/* STR_WITH_LEN() shortcuts */
385#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
84bafc02
NC
386#define newSVpvs_flags(str,flags) \
387 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
ba3a79e7 388#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
9dcc53ea
Z
389#define sv_catpvs_flags(sv, str, flags) \
390 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
391#define sv_catpvs_nomg(sv, str) \
392 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
393#define sv_catpvs(sv, str) \
394 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
395#define sv_catpvs_mg(sv, str) \
396 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
3fe05580 397#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
9dcc53ea
Z
398#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
399#define sv_setref_pvs(rv, classname, str) \
400 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
ba3a79e7 401#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
9dcc53ea
Z
402#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
403#define gv_stashpvs(str, create) \
404 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
405#define gv_fetchpvs(namebeg, add, sv_type) \
406 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
407#define gv_fetchpvn(namebeg, len, add, sv_type) \
408 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
409#define sv_catxmlpvs(dsv, str, utf8) \
410 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
4ac46235 411
ba3a79e7 412
510966aa
Z
413#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
414
b96d8cd9
NC
415#define get_cvs(str, flags) \
416 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
5c1737d1 417
954c1994 418/*
ccfc67b7
JH
419=head1 Miscellaneous Functions
420
954c1994 421=for apidoc Am|bool|strNE|char* s1|char* s2
dc6b0978
KW
422Test two C<NUL>-terminated strings to see if they are different. Returns true
423or false.
954c1994
GS
424
425=for apidoc Am|bool|strEQ|char* s1|char* s2
dc6b0978
KW
426Test two C<NUL>-terminated strings to see if they are equal. Returns true or
427false.
954c1994
GS
428
429=for apidoc Am|bool|strLT|char* s1|char* s2
dc6b0978
KW
430Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
431second, C<s2>. Returns true or false.
954c1994
GS
432
433=for apidoc Am|bool|strLE|char* s1|char* s2
dc6b0978
KW
434Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
435equal to the second, C<s2>. Returns true or false.
954c1994
GS
436
437=for apidoc Am|bool|strGT|char* s1|char* s2
dc6b0978
KW
438Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
439the second, C<s2>. Returns true or false.
954c1994
GS
440
441=for apidoc Am|bool|strGE|char* s1|char* s2
dc6b0978
KW
442Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
443or equal to the second, C<s2>. Returns true or false.
954c1994
GS
444
445=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
dc6b0978
KW
446Test two C<NUL>-terminated strings to see if they are different. The C<len>
447parameter indicates the number of bytes to compare. Returns true or false. (A
954c1994
GS
448wrapper for C<strncmp>).
449
450=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
dc6b0978
KW
451Test two C<NUL>-terminated strings to see if they are equal. The C<len>
452parameter indicates the number of bytes to compare. Returns true or false. (A
453wrapper for C<strncmp>).
954c1994 454
bd18bd40
KW
455=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
456Test two buffers (which may contain embedded C<NUL> characters, to see if they
457are equal. The C<len> parameter indicates the number of bytes to compare.
458Returns zero if equal, or non-zero if non-equal.
459
460=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
461Test two buffers (which may contain embedded C<NUL> characters, to see if they
462are not equal. The C<len> parameter indicates the number of bytes to compare.
463Returns zero if non-equal, or non-zero if equal.
464
954c1994
GS
465=cut
466*/
467
62946e08 468
75400963
KW
469#define strNE(s1,s2) (strcmp(s1,s2) != 0)
470#define strEQ(s1,s2) (strcmp(s1,s2) == 0)
8d063cd8
LW
471#define strLT(s1,s2) (strcmp(s1,s2) < 0)
472#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
473#define strGT(s1,s2) (strcmp(s1,s2) > 0)
474#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
62946e08 475
75400963
KW
476#define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
477#define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
378cc40b 478
c8b388b0 479#define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
62946e08 480
75400963
KW
481#define memNE(s1,s2,l) (memcmp(s1,s2,l) != 0)
482#define memEQ(s1,s2,l) (memcmp(s1,s2,l) == 0)
36477c24 483
085b7534 484/* memEQ and memNE where second comparand is a string constant */
568a785a 485#define memEQs(s1, l, s2) \
777fa2cb 486 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
5f50c6c9 487#define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
568a785a 488
bdb7e3f0
KW
489#define memBEGINs(s1, l, s2) \
490 ( (l) >= sizeof(s2) - 1 \
491 && memEQ(s1, "" s2 "", sizeof(s2)-1))
de627158
KW
492#define memBEGINPs(s1, l, s2) \
493 ( (l) > sizeof(s2) - 1 \
494 && memEQ(s1, "" s2 "", sizeof(s2)-1))
bdb7e3f0
KW
495#define memENDs(s1, l, s2) \
496 ( (l) >= sizeof(s2) - 1 \
497 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
b80f8424
KW
498#define memENDPs(s1, l, s2) \
499 ( (l) > sizeof(s2) \
500 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
bdb7e3f0 501
062b6850
KW
502#define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
503#define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
504#define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
505#define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
506
bbce6d69
PP
507/*
508 * Character classes.
509 *
510 * Unfortunately, the introduction of locales means that we
511 * can't trust isupper(), etc. to tell the truth. And when
512 * it comes to /\w+/ with tainting enabled, we *must* be able
513 * to trust our character classes.
514 *
515 * Therefore, the default tests in the text of Perl will be
516 * independent of locale. Any code that wants to depend on
517 * the current locale will use the tests that begin with "lc".
518 */
519
2304df62
AD
520#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
521# ifndef CTYPE256
522# define CTYPE256
523# endif
524#endif
525
954c1994 526/*
ccfc67b7 527
dcccc8ff 528=head1 Character classification
243effed
KW
529This section is about functions (really macros) that classify characters
530into types, such as punctuation versus alphabetic, etc. Most of these are
531analogous to regular expression character classes. (See
532L<perlrecharclass/POSIX Character Classes>.) There are several variants for
533each class. (Not all macros have all variants; each item below lists the
534ones valid for it.) None are affected by C<use bytes>, and only the ones
535with C<LC> in the name are affected by the current locale.
536
537The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
538C<U8>) as input and returns a boolean as to whether or not the character
6aff1f14
KW
539represented by that octet is (or on non-ASCII platforms, corresponds to) an
540ASCII character in the named class based on platform, Unicode, and Perl rules.
541If the input is a number that doesn't fit in an octet, FALSE is returned.
243effed 542
c98722a4 543Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
550da823
KW
544with no suffix C<"_A">. This variant is used to emphasize by its name that
545only ASCII-range characters can return TRUE.
4b9734bf 546
d60679e1 547Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
4b9734bf
KW
548onto the platform. That is, the code points that are ASCII are unaffected,
549since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
550as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
551true when called with the code point 0xDF, which is a word character in both
4650c663 552ASCII and EBCDIC (though it represents different characters in each).
243effed 553
c98722a4 554Variant C<isI<FOO>_uvchr> is like the C<isI<FOO>_L1> variant, but accepts any UV code
243effed
KW
555point as input. If the code point is larger than 255, Unicode rules are used
556to determine if it is in the character class. For example,
d0da05db 557C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
6aff1f14 558WITH MACRON in Unicode, and is a word character.
243effed 559
c98722a4 560Variant C<isI<FOO>_utf8_safe> is like C<isI<FOO>_uvchr>, but is used for UTF-8
da8c1a98
KW
561encoded strings. Each call classifies one character, even if the string
562contains many. This variant takes two parameters. The first, C<p>, is a
563pointer to the first byte of the character to be classified. (Recall that it
564may take more than one byte to represent a character in UTF-8 strings.) The
565second parameter, C<e>, points to anywhere in the string beyond the first
566character, up to one byte past the end of the entire string. The suffix
567C<_safe> in the function's name indicates that it will not attempt to read
568beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
569is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
570character is malformed in some way, the program may croak, or the function may
571return FALSE, at the discretion of the implementation, and subject to change in
572future releases.
573
c98722a4 574Variant C<isI<FOO>_utf8> is like C<isI<FOO>_utf8_safe>, but takes just a single
da8c1a98 575parameter, C<p>, which has the same meaning as the corresponding parameter does
c98722a4 576in C<isI<FOO>_utf8_safe>. The function therefore can't check if it is reading
34aeb2e9 577beyond the end of the string. Starting in Perl v5.30, it will take a second
c98722a4 578parameter, becoming a synonym for C<isI<FOO>_utf8_safe>. At that time every
34aeb2e9 579program that uses it will have to be changed to successfully compile. In the
c98722a4 580meantime, the first runtime call to C<isI<FOO>_utf8> from each call point in the
34aeb2e9 581program will raise a deprecation warning, enabled by default. You can convert
c98722a4 582your program now to use C<isI<FOO>_utf8_safe>, and avoid the warnings, and get an
34aeb2e9
KW
583extra measure of protection, or you can wait until v5.30, when you'll be forced
584to add the C<e> parameter.
243effed 585
c98722a4 586Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants, but the
31f05a37
KW
587result is based on the current locale, which is what C<LC> in the name stands
588for. If Perl can determine that the current locale is a UTF-8 locale, it uses
589the published Unicode rules; otherwise, it uses the C library function that
590gives the named classification. For example, C<isDIGIT_LC()> when not in a
591UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
1a83413c
KW
592returned if the input won't fit into an octet. On some platforms where the C
593library function is known to be defective, Perl changes its result to follow
594the POSIX standard's rules.
243effed 595
c98722a4
KW
596Variant C<isI<FOO>_LC_uvchr> is like C<isI<FOO>_LC>, but is defined on any UV. It
597returns the same as C<isI<FOO>_LC> for input code points less than 256, and
243effed
KW
598returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
599
c98722a4 600Variant C<isI<FOO>_LC_utf8_safe> is like C<isI<FOO>_LC_uvchr>, but is used for UTF-8
da8c1a98
KW
601encoded strings. Each call classifies one character, even if the string
602contains many. This variant takes two parameters. The first, C<p>, is a
603pointer to the first byte of the character to be classified. (Recall that it
604may take more than one byte to represent a character in UTF-8 strings.) The
605second parameter, C<e>, points to anywhere in the string beyond the first
606character, up to one byte past the end of the entire string. The suffix
607C<_safe> in the function's name indicates that it will not attempt to read
608beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
609is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
610character is malformed in some way, the program may croak, or the function may
611return FALSE, at the discretion of the implementation, and subject to change in
612future releases.
613
c98722a4 614Variant C<isI<FOO>_LC_utf8> is like C<isI<FOO>_LC_utf8_safe>, but takes just a single
da8c1a98 615parameter, C<p>, which has the same meaning as the corresponding parameter does
c98722a4 616in C<isI<FOO>_LC_utf8_safe>. The function therefore can't check if it is reading
34aeb2e9 617beyond the end of the string. Starting in Perl v5.30, it will take a second
c98722a4 618parameter, becoming a synonym for C<isI<FOO>_LC_utf8_safe>. At that time every
34aeb2e9 619program that uses it will have to be changed to successfully compile. In the
c98722a4 620meantime, the first runtime call to C<isI<FOO>_LC_utf8> from each call point in
34aeb2e9 621the program will raise a deprecation warning, enabled by default. You can
c98722a4 622convert your program now to use C<isI<FOO>_LC_utf8_safe>, and avoid the warnings,
34aeb2e9
KW
623and get an extra measure of protection, or you can wait until v5.30, when
624you'll be forced to add the C<e> parameter.
ccfc67b7 625
954c1994 626=for apidoc Am|bool|isALPHA|char ch
2787a470 627Returns a boolean indicating whether the specified character is an
6aff1f14 628alphabetic character, analogous to C<m/[[:alpha:]]/>.
dcccc8ff
KW
629See the L<top of this section|/Character classification> for an explanation of
630variants
da8c1a98
KW
631C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8_safe>,
632C<isALPHA_LC>, C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8_safe>.
8a58bdcf 633
15861f94
KW
634=for apidoc Am|bool|isALPHANUMERIC|char ch
635Returns a boolean indicating whether the specified character is a either an
6aff1f14 636alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
dcccc8ff
KW
637See the L<top of this section|/Character classification> for an explanation of
638variants
d0da05db 639C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
da8c1a98
KW
640C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>,
641and C<isALPHANUMERIC_LC_utf8_safe>.
15861f94 642
8a58bdcf
KW
643=for apidoc Am|bool|isASCII|char ch
644Returns a boolean indicating whether the specified character is one of the 128
243effed 645characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
e5ad6aba 646On non-ASCII platforms, it returns TRUE iff this
8a58bdcf
KW
647character corresponds to an ASCII character. Variants C<isASCII_A()> and
648C<isASCII_L1()> are identical to C<isASCII()>.
dcccc8ff
KW
649See the L<top of this section|/Character classification> for an explanation of
650variants
da8c1a98
KW
651C<isASCII_uvchr>, C<isASCII_utf8_safe>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
652C<isASCII_LC_utf8_safe>. Note, however, that some platforms do not have the C
243effed
KW
653library routine C<isascii()>. In these cases, the variants whose names contain
654C<LC> are the same as the corresponding ones without.
655
d98532ea
KW
656Also note, that because all ASCII characters are UTF-8 invariant (meaning they
657have the exact same representation (always a single byte) whether encoded in
658UTF-8 or not), C<isASCII> will give the correct results when called with any
da8c1a98
KW
659byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8_safe>
660will work properly on any string encoded or not in UTF-8.
d98532ea 661
243effed
KW
662=for apidoc Am|bool|isBLANK|char ch
663Returns a boolean indicating whether the specified character is a
6aff1f14 664character considered to be a blank, analogous to C<m/[[:blank:]]/>.
dcccc8ff
KW
665See the L<top of this section|/Character classification> for an explanation of
666variants
da8c1a98
KW
667C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8_safe>,
668C<isBLANK_LC>, C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8_safe>. Note,
669however, that some platforms do not have the C library routine
670C<isblank()>. In these cases, the variants whose names contain C<LC> are
671the same as the corresponding ones without.
243effed
KW
672
673=for apidoc Am|bool|isCNTRL|char ch
674Returns a boolean indicating whether the specified character is a
6aff1f14 675control character, analogous to C<m/[[:cntrl:]]/>.
dcccc8ff
KW
676See the L<top of this section|/Character classification> for an explanation of
677variants
da8c1a98
KW
678C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8_safe>,
679C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8_safe> On EBCDIC
680platforms, you almost always want to use the C<isCNTRL_L1> variant.
954c1994 681
954c1994 682=for apidoc Am|bool|isDIGIT|char ch
2787a470 683Returns a boolean indicating whether the specified character is a
6aff1f14 684digit, analogous to C<m/[[:digit:]]/>.
8a58bdcf 685Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
dcccc8ff
KW
686See the L<top of this section|/Character classification> for an explanation of
687variants
da8c1a98
KW
688C<isDIGIT_uvchr>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
689C<isDIGIT_LC_utf8_safe>.
243effed
KW
690
691=for apidoc Am|bool|isGRAPH|char ch
692Returns a boolean indicating whether the specified character is a
6aff1f14 693graphic character, analogous to C<m/[[:graph:]]/>.
dcccc8ff 694See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
695variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8_safe>,
696C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8_safe>.
954c1994 697
0c82b6df 698=for apidoc Am|bool|isLOWER|char ch
2787a470 699Returns a boolean indicating whether the specified character is a
6aff1f14 700lowercase character, analogous to C<m/[[:lower:]]/>.
dcccc8ff
KW
701See the L<top of this section|/Character classification> for an explanation of
702variants
da8c1a98
KW
703C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8_safe>,
704C<isLOWER_LC>, C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8_safe>.
0c82b6df 705
c99e91e9 706=for apidoc Am|bool|isOCTAL|char ch
2787a470 707Returns a boolean indicating whether the specified character is an
6aff1f14 708octal digit, [0-7].
243effed
KW
709The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
710C<isOCTAL>.
711
712=for apidoc Am|bool|isPUNCT|char ch
713Returns a boolean indicating whether the specified character is a
6aff1f14
KW
714punctuation character, analogous to C<m/[[:punct:]]/>.
715Note that the definition of what is punctuation isn't as
243effed
KW
716straightforward as one might desire. See L<perlrecharclass/POSIX Character
717Classes> for details.
dcccc8ff 718See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
719variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8_safe>,
720C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8_safe>.
c99e91e9 721
0c82b6df 722=for apidoc Am|bool|isSPACE|char ch
2787a470 723Returns a boolean indicating whether the specified character is a
6aff1f14 724whitespace character. This is analogous
398d098a 725to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
779cf272 726this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
398d098a
KW
727locale forms of this macro (the ones with C<LC> in their names) matched
728precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
729in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
730(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
dcccc8ff
KW
731See the L<top of this section|/Character classification> for an explanation of
732variants
da8c1a98
KW
733C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8_safe>,
734C<isSPACE_LC>, C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8_safe>.
0c82b6df 735
398d098a
KW
736=for apidoc Am|bool|isPSXSPC|char ch
737(short for Posix Space)
779cf272
KW
738Starting in 5.18, this is identical in all its forms to the
739corresponding C<isSPACE()> macros.
398d098a
KW
740The locale forms of this macro are identical to their corresponding
741C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
742non-locale forms differ from their C<isSPACE()> forms only in that the
743C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
744Otherwise they are identical. Thus this macro is analogous to what
745C<m/[[:space:]]/> matches in a regular expression.
dcccc8ff 746See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
747variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8_safe>,
748C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8_safe>.
398d098a 749
954c1994 750=for apidoc Am|bool|isUPPER|char ch
2787a470 751Returns a boolean indicating whether the specified character is an
6aff1f14 752uppercase character, analogous to C<m/[[:upper:]]/>.
dcccc8ff 753See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
754variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8_safe>,
755C<isUPPER_LC>, C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8_safe>.
954c1994 756
243effed 757=for apidoc Am|bool|isPRINT|char ch
8eea39dd 758Returns a boolean indicating whether the specified character is a
6aff1f14 759printable character, analogous to C<m/[[:print:]]/>.
dcccc8ff
KW
760See the L<top of this section|/Character classification> for an explanation of
761variants
da8c1a98
KW
762C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8_safe>,
763C<isPRINT_LC>, C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8_safe>.
243effed
KW
764
765=for apidoc Am|bool|isWORDCHAR|char ch
766Returns a boolean indicating whether the specified character is a character
767that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
768in a regular expression. A word character is an alphabetic character, a
769decimal digit, a connecting punctuation character (such as an underscore), or
770a "mark" character that attaches to one of those (like some sort of accent).
771C<isALNUM()> is a synonym provided for backward compatibility, even though a
772word character includes more than the standard C language meaning of
773alphanumeric.
dcccc8ff 774See the L<top of this section|/Character classification> for an explanation of
da8c1a98
KW
775variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and
776C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and
777C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
778include the platform's native underscore.
8a58bdcf
KW
779
780=for apidoc Am|bool|isXDIGIT|char ch
781Returns a boolean indicating whether the specified character is a hexadecimal
243effed
KW
782digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
783and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
dcccc8ff
KW
784See the L<top of this section|/Character classification> for an explanation of
785variants
da8c1a98
KW
786C<isXDIGIT_uvchr>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>,
787and C<isXDIGIT_LC_utf8_safe>.
243effed 788
3c3ecf18
KW
789=for apidoc Am|bool|isIDFIRST|char ch
790Returns a boolean indicating whether the specified character can be the first
791character of an identifier. This is very close to, but not quite the same as
792the official Unicode property C<XID_Start>. The difference is that this
793returns true only if the input character also matches L</isWORDCHAR>.
dcccc8ff
KW
794See the L<top of this section|/Character classification> for an explanation of
795variants
da8c1a98
KW
796C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8_safe>,
797C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8_safe>.
3c3ecf18
KW
798
799=for apidoc Am|bool|isIDCONT|char ch
800Returns a boolean indicating whether the specified character can be the
801second or succeeding character of an identifier. This is very close to, but
802not quite the same as the official Unicode property C<XID_Continue>. The
803difference is that this returns true only if the input character also matches
dcccc8ff
KW
804L</isWORDCHAR>. See the L<top of this section|/Character classification> for
805an
d0da05db 806explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
da8c1a98
KW
807C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
808C<isIDCONT_LC_utf8_safe>.
3c3ecf18 809
243effed 810=head1 Miscellaneous Functions
8eea39dd 811
95a59cab 812=for apidoc Am|U8|READ_XDIGIT|char str*
243effed 813Returns the value of an ASCII-range hex digit and advances the string pointer.
95a59cab
YO
814Behaviour is only well defined when isXDIGIT(*str) is true.
815
e7c1e6c1 816=head1 Character case changing
21da7284
KW
817Perl uses "full" Unicode case mappings. This means that converting a single
818character to another case may result in a sequence of more than one character.
819For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
820character sequence C<SS>. This presents some complications The lowercase of
821all characters in the range 0..255 is a single character, and thus
822C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
823return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
824an API that does allow every possible legal result to be returned.) Likewise
825no other function that is crippled by not being able to give the correct
826results for the full range of possible inputs has been implemented here.
e7c1e6c1 827
1f607577
KW
828=for apidoc Am|U8|toUPPER|U8 ch
829Converts the specified character to uppercase. If the input is anything but an
830ASCII lowercase character, that input character itself is returned. Variant
c753c8d3 831C<toUPPER_A> is equivalent.
954c1994 832
d0da05db
KW
833=for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
834Converts the code point C<cp> to its uppercase version, and
835stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
836point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
837that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
838bytes since the uppercase version may be longer than the original character.
839
840The first code point of the uppercased version is returned
21da7284
KW
841(but note, as explained at L<the top of this section|/Character case
842changing>, that there may be more.)
1f607577 843
a239b1e2
KW
844=for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
845Converts the first UTF-8 encoded character in the sequence starting at C<p> and
846extending no further than S<C<e - 1>> to its uppercase version, and
1f607577
KW
847stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
848that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
849bytes since the uppercase version may be longer than the original character.
850
851The first code point of the uppercased version is returned
21da7284
KW
852(but note, as explained at L<the top of this section|/Character case
853changing>, that there may be more).
1f607577 854
a239b1e2
KW
855The suffix C<_safe> in the function's name indicates that it will not attempt
856to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
857true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
858input character is malformed in some way, the program may croak, or the
859function may return the REPLACEMENT CHARACTER, at the discretion of the
860implementation, and subject to change in future releases.
861
862=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
863This is like C<L</toUPPER_utf8_safe>>, but doesn't have the C<e>
864parameter The function therefore can't check if it is reading
607313a1
KW
865beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
866parameter, becoming a synonym for C<toUPPER_utf8_safe>. At that time every
867program that uses it will have to be changed to successfully compile. In the
868meantime, the first runtime call to C<toUPPER_utf8> from each call point in the
869program will raise a deprecation warning, enabled by default. You can convert
870your program now to use C<toUPPER_utf8_safe>, and avoid the warnings, and get an
871extra measure of protection, or you can wait until v5.30, when you'll be forced
872to add the C<e> parameter.
1f607577 873
25200305
KW
874=for apidoc Am|U8|toFOLD|U8 ch
875Converts the specified character to foldcase. If the input is anything but an
876ASCII uppercase character, that input character itself is returned. Variant
877C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
d0da05db 878Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
25200305 879
d0da05db
KW
880=for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
881Converts the code point C<cp> to its foldcase version, and
882stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
883point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
884that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
885bytes since the foldcase version may be longer than the original character.
886
887The first code point of the foldcased version is returned
21da7284
KW
888(but note, as explained at L<the top of this section|/Character case
889changing>, that there may be more).
1f607577 890
a239b1e2
KW
891=for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
892Converts the first UTF-8 encoded character in the sequence starting at C<p> and
893extending no further than S<C<e - 1>> to its foldcase version, and
1f607577
KW
894stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
895that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
896bytes since the foldcase version may be longer than the original character.
897
898The first code point of the foldcased version is returned
21da7284
KW
899(but note, as explained at L<the top of this section|/Character case
900changing>, that there may be more).
1f607577 901
a239b1e2
KW
902The suffix C<_safe> in the function's name indicates that it will not attempt
903to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
904true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
905input character is malformed in some way, the program may croak, or the
906function may return the REPLACEMENT CHARACTER, at the discretion of the
907implementation, and subject to change in future releases.
908
909=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
910This is like C<L</toFOLD_utf8_safe>>, but doesn't have the C<e>
911parameter The function therefore can't check if it is reading
607313a1
KW
912beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
913parameter, becoming a synonym for C<toFOLD_utf8_safe>. At that time every
914program that uses it will have to be changed to successfully compile. In the
915meantime, the first runtime call to C<toFOLD_utf8> from each call point in the
916program will raise a deprecation warning, enabled by default. You can convert
917your program now to use C<toFOLD_utf8_safe>, and avoid the warnings, and get an
918extra measure of protection, or you can wait until v5.30, when you'll be forced
919to add the C<e> parameter.
1f607577
KW
920
921=for apidoc Am|U8|toLOWER|U8 ch
922Converts the specified character to lowercase. If the input is anything but an
923ASCII uppercase character, that input character itself is returned. Variant
c753c8d3 924C<toLOWER_A> is equivalent.
954c1994 925
1f607577 926=for apidoc Am|U8|toLOWER_L1|U8 ch
b7d90381
KW
927Converts the specified Latin1 character to lowercase. The results are
928undefined if the input doesn't fit in a byte.
1f607577
KW
929
930=for apidoc Am|U8|toLOWER_LC|U8 ch
931Converts the specified character to lowercase using the current locale's rules,
932if possible; otherwise returns the input character itself.
933
d0da05db
KW
934=for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
935Converts the code point C<cp> to its lowercase version, and
936stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
937point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
938that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
939bytes since the lowercase version may be longer than the original character.
940
941The first code point of the lowercased version is returned
21da7284
KW
942(but note, as explained at L<the top of this section|/Character case
943changing>, that there may be more).
1f607577 944
a239b1e2
KW
945
946=for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
947Converts the first UTF-8 encoded character in the sequence starting at C<p> and
948extending no further than S<C<e - 1>> to its lowercase version, and
1f607577
KW
949stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
950that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
951bytes since the lowercase version may be longer than the original character.
952
953The first code point of the lowercased version is returned
21da7284
KW
954(but note, as explained at L<the top of this section|/Character case
955changing>, that there may be more).
1f607577 956
a239b1e2
KW
957The suffix C<_safe> in the function's name indicates that it will not attempt
958to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
959true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
960input character is malformed in some way, the program may croak, or the
961function may return the REPLACEMENT CHARACTER, at the discretion of the
962implementation, and subject to change in future releases.
963
964=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
965This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
966parameter The function therefore can't check if it is reading
607313a1
KW
967beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
968parameter, becoming a synonym for C<toLOWER_utf8_safe>. At that time every
969program that uses it will have to be changed to successfully compile. In the
970meantime, the first runtime call to C<toLOWER_utf8> from each call point in the
971program will raise a deprecation warning, enabled by default. You can convert
972your program now to use C<toLOWER_utf8_safe>, and avoid the warnings, and get an
973extra measure of protection, or you can wait until v5.30, when you'll be forced
974to add the C<e> parameter.
1f607577 975
25200305
KW
976=for apidoc Am|U8|toTITLE|U8 ch
977Converts the specified character to titlecase. If the input is anything but an
978ASCII lowercase character, that input character itself is returned. Variant
b7d90381 979C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
d0da05db 980range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
b7d90381 981not a concept used in locale handling, so there is no functionality for that.)
25200305 982
d0da05db
KW
983=for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
984Converts the code point C<cp> to its titlecase version, and
985stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
986point is interpreted as native if less than 256; otherwise as Unicode. Note
1f607577
KW
987that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
988bytes since the titlecase version may be longer than the original character.
989
990The first code point of the titlecased version is returned
21da7284
KW
991(but note, as explained at L<the top of this section|/Character case
992changing>, that there may be more).
1f607577 993
a239b1e2
KW
994=for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
995Converts the first UTF-8 encoded character in the sequence starting at C<p> and
996extending no further than S<C<e - 1>> to its titlecase version, and
1f607577
KW
997stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
998that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
999bytes since the titlecase version may be longer than the original character.
1000
1001The first code point of the titlecased version is returned
21da7284
KW
1002(but note, as explained at L<the top of this section|/Character case
1003changing>, that there may be more).
1f607577 1004
a239b1e2
KW
1005The suffix C<_safe> in the function's name indicates that it will not attempt
1006to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1007true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1008input character is malformed in some way, the program may croak, or the
1009function may return the REPLACEMENT CHARACTER, at the discretion of the
1010implementation, and subject to change in future releases.
1011
1012=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
1013This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1014parameter The function therefore can't check if it is reading
607313a1
KW
1015beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
1016parameter, becoming a synonym for C<toTITLE_utf8_safe>. At that time every
1017program that uses it will have to be changed to successfully compile. In the
1018meantime, the first runtime call to C<toTITLE_utf8> from each call point in the
1019program will raise a deprecation warning, enabled by default. You can convert
1020your program now to use C<toTITLE_utf8_safe>, and avoid the warnings, and get an
1021extra measure of protection, or you can wait until v5.30, when you'll be forced
1022to add the C<e> parameter.
1f607577 1023
954c1994 1024=cut
353c9b6f 1025
d0da05db 1026XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1e222e4f
KW
1027really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1028and aren't general purpose as they don't work on U+DF, and assert against that.
243effed 1029
8a58bdcf 1030Note that these macros are repeated in Devel::PPPort, so should also be
62fa66b6
KW
1031patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1032
954c1994
GS
1033*/
1034
7c062697
KW
1035/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
1036 * is known only in the perl core, and this macro can be called from outside
1037 * that */
1038#ifdef HAS_QUAD
1039# define WIDEST_UTYPE U64TYPE
1040#else
1041# define WIDEST_UTYPE U32
1042#endif
1043
3912bc88
KW
1044/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1045 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1046 * bits of information are lost even on a 64-bit machine, but to get the
1047 * compiler to optimize it out if possible. This is because Configure makes
1048 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1049 * sizeof() guarantees that this evaluates to a constant true at compile time.
7e75d1a1
JH
1050 *
1051 * For Coverity, be always true, because otherwise Coverity thinks
1052 * it finds several expressions that are always true, independent
1053 * of operands. Well, they are, but that is kind of the point.
220c71bf 1054 */
7e75d1a1 1055#ifndef __COVERITY__
6c5b02ac
KW
1056 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1057 * pointer) */
1058#define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
ace3ad0f 1059 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
7e75d1a1
JH
1060#else
1061#define FITS_IN_8_BITS(c) (1)
1062#endif
cf301eb7 1063
41f43cc2 1064#ifdef EBCDIC
b6340bd0 1065# ifndef _ALL_SOURCE
0852beac
KW
1066 /* The native libc isascii() et.al. functions return the wrong results
1067 * on at least z/OS unless this is defined. */
b6340bd0
KW
1068# error _ALL_SOURCE should probably be defined
1069# endif
41f43cc2 1070#else
0852beac
KW
1071 /* There is a simple definition of ASCII for ASCII platforms. But the
1072 * EBCDIC one isn't so simple, so is defined using table look-up like the
9c903d59 1073 * other macros below.
3f3c579d
KW
1074 *
1075 * The cast here is used instead of '(c) >= 0', because some compilers emit
1076 * a warning that that test is always true when the parameter is an
1077 * unsigned type. khw supposes that it could be written as
1078 * && ((c) == '\0' || (c) > 0)
1079 * to avoid the message, but the cast will likely avoid extra branches even
1080 * with stupid compilers.
1081 *
1082 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1083 * a pointer) */
9c903d59 1084# define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
41f43cc2
KW
1085#endif
1086
38694112
KW
1087/* Take the eight possible bit patterns of the lower 3 bits and you get the
1088 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1089 * can be ignored. If the rest match '0', we have an octal */
1090#define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
c2da0b36 1091
9fb1bf9d 1092#ifdef H_PERL /* If have access to perl.h, lookup in its table */
f4cdb42c 1093
a500dc72
KW
1094/* Character class numbers. For internal core Perl use only. The ones less
1095 * than 32 are used in PL_charclass[] and the ones up through the one that
1096 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1097 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1098 * their actual definitions are here. If that file has a name not used here,
1099 * it won't compile.
1709d539
KW
1100 *
1101 * The first group of these is ordered in what I (khw) estimate to be the
31c7f561
KW
1102 * frequency of their use. This gives a slight edge to exiting a loop earlier
1103 * (in reginclass() in regexec.c) */
1709d539
KW
1104# define _CC_WORDCHAR 0 /* \w and [:word:] */
1105# define _CC_DIGIT 1 /* \d and [:digit:] */
1106# define _CC_ALPHA 2 /* [:alpha:] */
1107# define _CC_LOWER 3 /* [:lower:] */
1108# define _CC_UPPER 4 /* [:upper:] */
1109# define _CC_PUNCT 5 /* [:punct:] */
1110# define _CC_PRINT 6 /* [:print:] */
15861f94 1111# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1709d539 1112# define _CC_GRAPH 8 /* [:graph:] */
359b005e 1113# define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1709d539 1114
b0d691b2 1115#define _FIRST_NON_SWASH_CC 10
1709d539
KW
1116/* The character classes above are implemented with swashes. The second group
1117 * (just below) contains the ones implemented without. These are also sorted
1118 * in rough order of the frequency of their use, except that \v should be last,
1119 * as it isn't a real Posix character class, and some (small) inefficiencies in
1120 * regular expression handling would be introduced by putting it in the middle
1121 * of those that are. Also, cntrl and ascii come after the others as it may be
1122 * useful to group these which have no members that match above Latin1, (or
1123 * above ASCII in the latter case) */
1124
779cf272 1125# define _CC_SPACE 10 /* \s, [:space:] */
34aeb2e9
KW
1126# define _CC_PSXSPC _CC_SPACE /* XXX Temporary, can be removed
1127 when the deprecated isFOO_utf8()
1128 functions are removed */
b0d691b2
KW
1129# define _CC_BLANK 11 /* [:blank:] */
1130# define _CC_XDIGIT 12 /* [:xdigit:] */
779cf272
KW
1131# define _CC_CNTRL 13 /* [:cntrl:] */
1132# define _CC_ASCII 14 /* [:ascii:] */
1133# define _CC_VERTSPACE 15 /* \v */
1709d539 1134
a0947d7b
KW
1135# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1136
1709d539 1137/* The members of the third group below do not need to be coordinated with data
3ffc8c70 1138 * structures in regcomp.[ch] and regexec.c. */
779cf272
KW
1139# define _CC_IDFIRST 16
1140# define _CC_CHARNAME_CONT 17
1141# define _CC_NONLATIN1_FOLD 18
1142# define _CC_NONLATIN1_SIMPLE_FOLD 19
1143# define _CC_QUOTEMETA 20
1144# define _CC_NON_FINAL_FOLD 21
1145# define _CC_IS_IN_SOME_FOLD 22
1146# define _CC_MNEMONIC_CNTRL 23
073c22b3 1147
34aeb2e9
KW
1148# define _CC_IDCONT 24 /* XXX Temporary, can be removed when the deprecated
1149 isFOO_utf8() functions are removed */
1150
073c22b3
KW
1151/* This next group is only used on EBCDIC platforms, so theoretically could be
1152 * shared with something entirely different that's only on ASCII platforms */
5d5376e2
KW
1153# define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1154# define _CC_UTF8_IS_START 29
1155# define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1156# define _CC_UTF8_IS_CONTINUATION 31
1157/* Unused: 24-27
f4cdb42c
KW
1158 * If more bits are needed, one could add a second word for non-64bit
1159 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
37ede926
KW
1160 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1161 * is used only for optimization (as of this writing), and differs in the
1162 * Latin1 range from the ALPHA bit only in two relatively unimportant
a500dc72 1163 * characters: the masculine and feminine ordinal indicators, so removing it
073c22b3
KW
1164 * would just cause /i regexes which match them to run less efficiently.
1165 * Similarly the EBCDIC-only bits are used just for speed, and could be
1166 * replaced by other means */
96ac0975 1167
3a371f2f
KW
1168#if defined(PERL_CORE) || defined(PERL_EXT)
1169/* An enum version of the character class numbers, to help compilers
1170 * optimize */
1171typedef enum {
3a371f2f 1172 _CC_ENUM_ALPHA = _CC_ALPHA,
e8d596e0
KW
1173 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1174 _CC_ENUM_ASCII = _CC_ASCII,
1175 _CC_ENUM_BLANK = _CC_BLANK,
b0d691b2 1176 _CC_ENUM_CASED = _CC_CASED,
e8d596e0 1177 _CC_ENUM_CNTRL = _CC_CNTRL,
3a371f2f
KW
1178 _CC_ENUM_DIGIT = _CC_DIGIT,
1179 _CC_ENUM_GRAPH = _CC_GRAPH,
1180 _CC_ENUM_LOWER = _CC_LOWER,
1181 _CC_ENUM_PRINT = _CC_PRINT,
1182 _CC_ENUM_PUNCT = _CC_PUNCT,
e8d596e0 1183 _CC_ENUM_SPACE = _CC_SPACE,
3a371f2f 1184 _CC_ENUM_UPPER = _CC_UPPER,
e8d596e0 1185 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
3a371f2f 1186 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
e8d596e0 1187 _CC_ENUM_XDIGIT = _CC_XDIGIT
3a371f2f
KW
1188} _char_class_number;
1189#endif
1190
63c61c3f 1191#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
86f72d56 1192#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
63c61c3f 1193
b7d90381
KW
1194#if defined(PERL_IN_UTF8_C) \
1195 || defined(PERL_IN_REGCOMP_C) \
1196 || defined(PERL_IN_REGEXEC_C)
922e8cb4 1197# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
b0d691b2
KW
1198 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1199 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
922e8cb4
KW
1200 #error Need to adjust order of swash_property_names[]
1201# endif
1202
b44daeeb
KW
1203/* This is declared static in each of the few files that this is #defined for
1204 * to keep them from being publicly accessible. Hence there is a small amount
1205 * of wasted space */
1206
922e8cb4
KW
1207static const char* const swash_property_names[] = {
1208 "XPosixWord",
1209 "XPosixDigit",
1210 "XPosixAlpha",
1211 "XPosixLower",
1212 "XPosixUpper",
1213 "XPosixPunct",
1214 "XPosixPrint",
1215 "XPosixAlnum",
b0d691b2
KW
1216 "XPosixGraph",
1217 "Cased"
922e8cb4
KW
1218};
1219#endif
1220
6635f04f 1221START_EXTERN_C
96ac0975
NC
1222# ifdef DOINIT
1223EXTCONST U32 PL_charclass[] = {
1224# include "l1_char_class_tab.h"
1225};
1226
1227# else /* ! DOINIT */
1228EXTCONST U32 PL_charclass[];
1229# endif
6635f04f 1230END_EXTERN_C
96ac0975 1231
265c1f46 1232 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
430b7c70 1233# define _CC_mask(classnum) (1U << (classnum))
4650c663
KW
1234
1235 /* For internal core Perl use only: the base macro for defining macros like
1236 * isALPHA */
ff7ecfc3 1237# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
f4cd282c 1238 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
4eeeb416 1239
f4cdb42c
KW
1240 /* The mask for the _A versions of the macros; it just adds in the bit for
1241 * ASCII. */
1242# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1243
4650c663
KW
1244 /* For internal core Perl use only: the base macro for defining macros like
1245 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1246 * the ASCII bit are present */
b7d90381
KW
1247# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1248 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1249 == _CC_mask_A(classnum)))
f4cdb42c 1250
f4cdb42c 1251# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
15861f94 1252# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
f4cdb42c
KW
1253# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1254# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
96ca48da 1255# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
f4cdb42c
KW
1256# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1257# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1258# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
f4cdb42c
KW
1259# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1260# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1261# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1262# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
b7d90381
KW
1263# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1264 */
d95f8b6a 1265# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
3ded5eb0
KW
1266# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1267# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1268# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1269
1270 /* continuation character for legal NAME in \N{NAME} */
1271# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1272
1273# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1274# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1275# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1276# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
b7d90381 1277# define isPSXSPC_L1(c) isSPACE_L1(c)
3ded5eb0
KW
1278# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1279# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1280# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1281# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1282# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
f4cdb42c 1283
0852beac
KW
1284# ifdef EBCDIC
1285# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1286# endif
1287
f12c0118
KW
1288 /* Participates in a single-character fold with a character above 255 */
1289# 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)))
1290
1291 /* Like the above, but also can be part of a multi-char fold */
f4cd282c 1292# 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 1293
4eeeb416 1294# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
26faadbd 1295# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1296 _generic_isCC(c, _CC_NON_FINAL_FOLD)
37ede926 1297# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
b7d90381 1298 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
8e35b056
KW
1299# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1300 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
687c8d01 1301#else /* else we don't have perl.h H_PERL */
3ded5eb0
KW
1302
1303 /* If we don't have perl.h, we are compiling a utility program. Below we
1304 * hard-code various macro definitions that wouldn't otherwise be available
fc273927 1305 * to it. Most are coded based on first principles. These are written to
74665a89 1306 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
0852beac
KW
1307# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1308# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
74665a89
KW
1309# define isSPACE_A(c) (isBLANK_A(c) \
1310 || (c) == '\n' \
1311 || (c) == '\r' \
1312 || (c) == '\v' \
0852beac 1313 || (c) == '\f')
74665a89
KW
1314 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1315 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1316 * performance (if optimization isn't on), and allow the same code to be
1317 * used for both platform types */
1318# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1319 && ( (c) <= 'i' \
1320 || ((c) >= 'j' && (c) <= 'r') \
1321 || (c) >= 's'))
1322# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1323 && ( (c) <= 'I' \
1324 || ((c) >= 'J' && (c) <= 'R') \
1325 || (c) >= 'S'))
a4d7a999
KW
1326# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1327# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
3ded5eb0 1328# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
0852beac 1329# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
74665a89
KW
1330# define isXDIGIT_A(c) (isDIGIT_A(c) \
1331 || ((c) >= 'a' && (c) <= 'f') \
a4d7a999 1332 || ((c) <= 'F' && (c) >= 'A'))
74665a89
KW
1333# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1334 || (c) == '#' || (c) == '$' || (c) == '%' \
1335 || (c) == '&' || (c) == '\'' || (c) == '(' \
1336 || (c) == ')' || (c) == '*' || (c) == '+' \
1337 || (c) == ',' || (c) == '.' || (c) == '/' \
1338 || (c) == ':' || (c) == ';' || (c) == '<' \
1339 || (c) == '=' || (c) == '>' || (c) == '?' \
1340 || (c) == '@' || (c) == '[' || (c) == '\\' \
1341 || (c) == ']' || (c) == '^' || (c) == '_' \
1342 || (c) == '`' || (c) == '{' || (c) == '|' \
1343 || (c) == '}' || (c) == '~')
1344# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1345# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
3ded5eb0 1346
0852beac 1347# ifdef EBCDIC
74665a89
KW
1348 /* The below is accurate for the 3 EBCDIC code pages traditionally
1349 * supported by perl. The only difference between them in the controls
1350 * is the position of \n, and that is represented symbolically below */
1351# define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1352 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1353 || (c) == '\t' || (c) == '\v' \
1354 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1355 || (c) == 7 /* U+7F DEL */ \
1356 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1357 /* DLE, DC[1-3] */ \
1358 || (c) == 0x18 /* U+18 CAN */ \
1359 || (c) == 0x19 /* U+19 EOM */ \
1360 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1361 || (c) == 0x26 /* U+17 ETB */ \
1362 || (c) == 0x27 /* U+1B ESC */ \
1363 || (c) == 0x2D /* U+05 ENQ */ \
1364 || (c) == 0x2E /* U+06 ACK */ \
1365 || (c) == 0x32 /* U+16 SYN */ \
1366 || (c) == 0x37 /* U+04 EOT */ \
1367 || (c) == 0x3C /* U+14 DC4 */ \
1368 || (c) == 0x3D /* U+15 NAK */ \
1369 || (c) == 0x3F)/* U+1A SUB */
0852beac 1370# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
74665a89
KW
1371# else /* isASCII is already defined for ASCII platforms, so can use that to
1372 define isCNTRL */
1373# define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
0852beac
KW
1374# endif
1375
3ffc8c70 1376 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
caa94d35
KW
1377 * during debugging, and it seems best to keep them. We may be called
1378 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1379 * do anything anyway, so make it not a problem */
1380# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1381# define NATIVE_TO_LATIN1(ch) (ch)
1382# endif
3ded5eb0
KW
1383# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1384# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1385# define isBLANK_L1(c) (isBLANK_A(c) \
1386 || (FITS_IN_8_BITS(c) \
1387 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1388# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1389# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1390# define isLOWER_L1(c) (isLOWER_A(c) \
1391 || (FITS_IN_8_BITS(c) \
ae683a5f 1392 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
3ded5eb0
KW
1393 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1394 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1395 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1396 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1397# define isPRINT_L1(c) (isPRINT_A(c) \
1398 || (FITS_IN_8_BITS(c) \
1399 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
3ded5eb0
KW
1400# define isPUNCT_L1(c) (isPUNCT_A(c) \
1401 || (FITS_IN_8_BITS(c) \
ae683a5f 1402 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
3ded5eb0
KW
1403 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1404 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1405 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1406 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1407 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1408 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1409# define isSPACE_L1(c) (isSPACE_A(c) \
1410 || (FITS_IN_8_BITS(c) \
ae683a5f 1411 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
3ded5eb0
KW
1412 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1413# define isUPPER_L1(c) (isUPPER_A(c) \
1414 || (FITS_IN_8_BITS(c) \
ae683a5f 1415 && ( NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
3ded5eb0
KW
1416 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1417 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1418# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1419# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1420# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1421 || isBLANK_L1(c) \
1422 || (c) == '-' \
1423 || (c) == '(' \
1424 || (c) == ')')
1425 /* The following are not fully accurate in the above-ASCII range. I (khw)
1426 * don't think it's necessary to be so for the purposes where this gets
1427 * compiled */
1428# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1429# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1430
1431 /* And these aren't accurate at all. They are useful only for above
1432 * Latin1, which utilities and bootstrapping don't deal with */
1433# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
6838b41e 1434# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
3ded5eb0
KW
1435# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1436
1437 /* Many of the macros later in this file are defined in terms of these. By
1438 * implementing them with a function, which converts the class number into
1439 * a call to the desired macro, all of the later ones work. However, that
1440 * function won't be actually defined when building a utility program (no
1441 * perl.h), and so a compiler error will be generated if one is attempted
1442 * to be used. And the above-Latin1 code points require Unicode tables to
1443 * be present, something unlikely to be the case when bootstrapping */
1444# define _generic_isCC(c, classnum) \
1445 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1446# define _generic_isCC_A(c, classnum) \
1447 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
687c8d01 1448#endif /* End of no perl.h H_PERL */
8a58bdcf 1449
e66b99e9
KW
1450#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1451#define isALPHA(c) isALPHA_A(c)
0852beac
KW
1452#define isASCII_A(c) isASCII(c)
1453#define isASCII_L1(c) isASCII(c)
e66b99e9
KW
1454#define isBLANK(c) isBLANK_A(c)
1455#define isCNTRL(c) isCNTRL_A(c)
1456#define isDIGIT(c) isDIGIT_A(c)
1457#define isGRAPH(c) isGRAPH_A(c)
1458#define isIDFIRST(c) isIDFIRST_A(c)
1459#define isLOWER(c) isLOWER_A(c)
1460#define isPRINT(c) isPRINT_A(c)
779cf272 1461#define isPSXSPC_A(c) isSPACE_A(c)
e66b99e9 1462#define isPSXSPC(c) isPSXSPC_A(c)
779cf272 1463#define isPSXSPC_L1(c) isSPACE_L1(c)
e66b99e9
KW
1464#define isPUNCT(c) isPUNCT_A(c)
1465#define isSPACE(c) isSPACE_A(c)
1466#define isUPPER(c) isUPPER_A(c)
1467#define isWORDCHAR(c) isWORDCHAR_A(c)
1468#define isXDIGIT(c) isXDIGIT_A(c)
1469
1470/* ASCII casing. These could also be written as
1471 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1472 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1473 which uses table lookup and mask instead of subtraction. (This would
1474 work because the _MOD does not apply in the ASCII range) */
68067e4e
DM
1475#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1476#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
bbce6d69 1477
25200305
KW
1478/* In the ASCII range, these are equivalent to what they're here defined to be.
1479 * But by creating these definitions, other code doesn't have to be aware of
1480 * this detail */
1481#define toFOLD(c) toLOWER(c)
25200305
KW
1482#define toTITLE(c) toUPPER(c)
1483
c753c8d3
KW
1484#define toLOWER_A(c) toLOWER(c)
1485#define toUPPER_A(c) toUPPER(c)
25200305
KW
1486#define toFOLD_A(c) toFOLD(c)
1487#define toTITLE_A(c) toTITLE(c)
1a0901db 1488
4650c663 1489/* Use table lookup for speed; returns the input itself if is out-of-range */
b2bf251f 1490#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1491 ? (c) \
f4cd282c 1492 : PL_latin1_lc[ (U8) (c) ])
c753c8d3
KW
1493#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1494
1a0901db 1495/* Modified uc. Is correct uc except for three non-ascii chars which are
4650c663
KW
1496 * all mapped to one of them, and these need special handling; returns the
1497 * input itself if is out-of-range */
b2bf251f 1498#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
8e7c6e7d 1499 ? (c) \
f4cd282c 1500 : PL_mod_latin1_uc[ (U8) (c) ])
31f05a37 1501#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
84061b6a 1502
beab9ebe
KW
1503/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1504
1505/* For internal core Perl use only: the base macro for defining macros like
1506 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
31f05a37
KW
1507 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1508 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1509 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1510 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1511 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1512 * code points under locale rules, use the macros based on _generic_LC_uvchr
1513 * instead of this. */
beab9ebe
KW
1514#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1515 (! FITS_IN_8_BITS(c) \
1516 ? 0 \
31f05a37
KW
1517 : IN_UTF8_CTYPE_LOCALE \
1518 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
beab9ebe
KW
1519 : cBOOL(non_utf8))
1520
1521/* For internal core Perl use only: a helper macro for defining macros like
1522 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1523 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1524 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1525 * _generic_LC_base for more info */
1526#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1527 _generic_LC_base(c,utf8_locale_classnum, \
1528 non_utf8_func( (_LC_CAST) (c)))
1529
1530/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1531 * 'c' is the platform's native underscore character */
1532#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1533 _generic_LC_base(c, utf8_locale_classnum, \
1534 (non_utf8_func( (_LC_CAST) (c)) \
1535 || (char)(c) == '_'))
1536
1537/* These next three are also for internal core Perl use only: case-change
1538 * helper macros */
1539#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1540 ? (c) \
31f05a37
KW
1541 : (IN_UTF8_CTYPE_LOCALE) \
1542 ? PL_latin1_lc[ (U8) (c) ] \
d2a8cf7f 1543 : (cast)function((cast)(c)))
beab9ebe 1544
31f05a37
KW
1545/* Note that the result can be larger than a byte in a UTF-8 locale. It
1546 * returns a single value, so can't adequately return the upper case of LATIN
1547 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1548 * values "SS"); instead it asserts against that under DEBUGGING, and
1549 * otherwise returns its input */
beab9ebe
KW
1550#define _generic_toUPPER_LC(c, function, cast) \
1551 (! FITS_IN_8_BITS(c) \
1552 ? (c) \
31f05a37 1553 : ((! IN_UTF8_CTYPE_LOCALE) \
b7d90381 1554 ? (cast)function((cast)(c)) \
31f05a37
KW
1555 : ((((U8)(c)) == MICRO_SIGN) \
1556 ? GREEK_CAPITAL_LETTER_MU \
1557 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1558 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1559 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1560 ? (__ASSERT_(0) (c)) \
1561 : PL_mod_latin1_uc[ (U8) (c) ])))))
1562
1563/* Note that the result can be larger than a byte in a UTF-8 locale. It
1564 * returns a single value, so can't adequately return the fold case of LATIN
1565 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1566 * values "ss"); instead it asserts against that under DEBUGGING, and
1567 * otherwise returns its input */
beab9ebe 1568#define _generic_toFOLD_LC(c, function, cast) \
4d7db1b9 1569 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
31f05a37 1570 ? GREEK_SMALL_LETTER_MU \
4d7db1b9
KW
1571 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1572 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1573 _generic_toLOWER_LC(c, function, cast)))
beab9ebe 1574
84061b6a 1575/* Use the libc versions for these if available. */
f05550c0 1576#if defined(HAS_ISASCII)
84061b6a
KW
1577# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1578#else
1579# define isASCII_LC(c) isASCII(c)
1580#endif
1581
f05550c0 1582#if defined(HAS_ISBLANK)
84061b6a 1583# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
95f34b6f 1584#else /* Unlike isASCII, varies if in a UTF-8 locale */
278b2b56 1585# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
84061b6a
KW
1586#endif
1587
f05550c0 1588#define _LC_CAST U8
bbce6d69 1589
f05550c0 1590#ifdef WIN32
375f5f06
KW
1591 /* The Windows functions don't bother to follow the POSIX standard, which
1592 * for example says that something can't both be a printable and a control.
1593 * But Windows treats the \t control as a printable, and does such things
1594 * as making superscripts into both digits and punctuation. This tames
1595 * these flaws by assuming that the definitions of both controls and space
1596 * are correct, and then making sure that other definitions don't have
1597 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1598 * Not all possible weirdnesses are checked for, just the ones that were
1599 * detected on actual Microsoft code pages */
1600
b7d90381
KW
1601# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1602# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1603
1604# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1605 && isALPHANUMERIC_LC(c))
1606# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1607 ! isPUNCT_LC(c))
1608# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1609 isALPHANUMERIC_LC(c))
1610# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1611# define isIDFIRST_LC(c) (((c) == '_') \
1612 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1613# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1614# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1615# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1616# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
f05550c0 1617# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
b7d90381
KW
1618# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1619 && isALPHANUMERIC_LC(c))
f05550c0
BF
1620
1621# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1622# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1623# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1624
1625#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
4650c663 1626 /* For most other platforms */
beab9ebe 1627
f05550c0
BF
1628# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1629# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1630# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1631# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1632# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1633# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1634# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1635# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1636# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1637# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1638# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1639# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1640# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1641
1642
1643# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1644# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1645# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1646
1647#else /* The final fallback position */
1648
b7d90381
KW
1649# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1650# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1651# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1652# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1653# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
f05550c0 1654# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
b7d90381
KW
1655# define isLOWER_LC(c) (isascii(c) && islower(c))
1656# define isPRINT_LC(c) (isascii(c) && isprint(c))
1657# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1658# define isSPACE_LC(c) (isascii(c) && isspace(c))
1659# define isUPPER_LC(c) (isascii(c) && isupper(c))
f05550c0 1660# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
b7d90381 1661# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
f05550c0
BF
1662
1663# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1664# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1665# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
bbce6d69 1666
f05550c0 1667#endif
55204971 1668
eba68aa0
KW
1669#define isIDCONT(c) isWORDCHAR(c)
1670#define isIDCONT_A(c) isWORDCHAR_A(c)
1671#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1672#define isIDCONT_LC(c) isWORDCHAR_LC(c)
13380643 1673#define isPSXSPC_LC(c) isSPACE_LC(c)
aaa51d5e 1674
4650c663 1675/* For internal core Perl use only: the base macros for defining macros like
d0da05db
KW
1676 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1677 * number defined earlier in this file. _generic_uvchr() is used for POSIX
4650c663
KW
1678 * classes where there is a macro or function 'above_latin1' that takes the
1679 * single argument 'c' and returns the desired value. These exist for those
1680 * classes which have simple definitions, avoiding the overhead of a hash
d0da05db 1681 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
4650c663 1682 * for classes where that overhead is faster than a direct lookup.
d0da05db 1683 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
4650c663
KW
1684 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1685 * have duplicate checks here, so could create versions of the macros that
1686 * don't, but experiments show that gcc optimizes them out anyway. */
66c17564
KW
1687
1688/* Note that all ignore 'use bytes' */
1e222e4f
KW
1689#define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1690 ? _generic_isCC(c, classnum) \
cd500f2f 1691 : above_latin1(c))
1e222e4f
KW
1692#define _generic_swash_uvchr(classnum, c) ((c) < 256 \
1693 ? _generic_isCC(c, classnum) \
922e8cb4 1694 : _is_uni_FOO(classnum, c))
d0da05db
KW
1695#define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1696#define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1697#define isASCII_uvchr(c) isASCII(c)
1698#define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1699#define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1700#define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1701#define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1e222e4f
KW
1702#define isIDCONT_uvchr(c) \
1703 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1704#define isIDFIRST_uvchr(c) \
1705 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
d0da05db
KW
1706#define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1707#define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1708
1709#define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1710#define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1711#define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1712
1713#define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1714#define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1715#define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1716#define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1717
1718#define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1719#define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1720#define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1721#define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1722
1723/* For backwards compatibility, even though '_uni' should mean official Unicode
1724 * code points, in Perl it means native for those below 256 */
1725#define isALPHA_uni(c) isALPHA_uvchr(c)
1726#define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1727#define isASCII_uni(c) isASCII_uvchr(c)
1728#define isBLANK_uni(c) isBLANK_uvchr(c)
1729#define isCNTRL_uni(c) isCNTRL_uvchr(c)
1730#define isDIGIT_uni(c) isDIGIT_uvchr(c)
1731#define isGRAPH_uni(c) isGRAPH_uvchr(c)
1732#define isIDCONT_uni(c) isIDCONT_uvchr(c)
1733#define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1734#define isLOWER_uni(c) isLOWER_uvchr(c)
1735#define isPRINT_uni(c) isPRINT_uvchr(c)
1736#define isPUNCT_uni(c) isPUNCT_uvchr(c)
1737#define isSPACE_uni(c) isSPACE_uvchr(c)
1738#define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1739#define isUPPER_uni(c) isUPPER_uvchr(c)
1740#define isVERTWS_uni(c) isVERTWS_uvchr(c)
1741#define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1742#define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1743#define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1744#define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1745#define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1746#define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
a0ed51b3 1747
4650c663
KW
1748/* For internal core Perl use only: the base macros for defining macros like
1749 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
d0da05db 1750 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
4650c663 1751 * simple class definitions; the other for more complex. These are like
d0da05db 1752 * _generic_uvchr, so see it for more info. */
cd500f2f
KW
1753#define _generic_LC_uvchr(latin1, above_latin1, c) \
1754 (c < 256 ? latin1(c) : above_latin1(c))
1755#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1756 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1757
1758#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1759#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1760 _CC_ALPHANUMERIC, c)
b7d90381
KW
1761#define isASCII_LC_uvchr(c) isASCII_LC(c)
1762#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1763 is_HORIZWS_cp_high, c)
feeab5a9 1764#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
cd500f2f
KW
1765#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1766#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
b7d90381 1767#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
eba68aa0 1768 _is_uni_perl_idcont, c)
b7d90381 1769#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
cd500f2f
KW
1770 _is_uni_perl_idstart, c)
1771#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1772#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
b7d90381 1773#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
cd500f2f 1774#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
b7d90381 1775#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
509fb054 1776 is_XPERLSPACE_cp_high, c)
cd500f2f 1777#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
b7d90381 1778#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
cd500f2f 1779 _CC_WORDCHAR, c)
b7d90381
KW
1780#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1781 is_XDIGIT_cp_high, c)
e712593e 1782
b7d90381 1783#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
aaa51d5e 1784
4650c663
KW
1785/* For internal core Perl use only: the base macros for defining macros like
1786 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1787 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1788 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1789 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1790 * same representation whether utf8 or not. Note that it assumes that the utf8
1791 * has been validated, and ignores 'use bytes' */
34aeb2e9
KW
1792#define _base_generic_utf8(enum_name, name, p, use_locale ) \
1793 _is_utf8_FOO(CAT2(_CC_, enum_name), \
1794 (const U8 *) p, \
1795 "is" STRINGIFY(name) "_utf8", \
1796 "is" STRINGIFY(name) "_utf8_safe", \
1797 1, use_locale, __FILE__,__LINE__)
1798
1799#define _generic_utf8(name, p) _base_generic_utf8(name, name, p, 0)
da8c1a98
KW
1800
1801/* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
1802 * they don't otherwise go out of their way to look for malformed UTF-8. If
1803 * they can return accurate results without knowing if the input is otherwise
1804 * malformed, they do so. For example isASCII is accurate in spite of any
1805 * non-length malformations because it looks only at a single byte. Likewise
1806 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
1807 * variant ones return FALSE. But, if the input has to be well-formed in order
1808 * for the results to be accurate, the macros will test and if malformed will
1809 * call a routine to die
1810 *
1811 * Except for toke.c, the macros do assume that e > p, asserting that on
1812 * DEBUGGING builds. Much code that calls these depends on this being true,
1813 * for other reasons. toke.c is treated specially as using the regular
1814 * assertion breaks it in many ways. All strings that these operate on there
1815 * are supposed to have an extra NUL character at the end, so that *e = \0. A
1816 * bunch of code in toke.c assumes that this is true, so the assertion allows
1817 * for that */
1818#ifdef PERL_IN_TOKE_C
1819# define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
1820#else
1821# define _utf8_safe_assert(p,e) ((e) > (p))
1822#endif
1823
1824#define _generic_utf8_safe(classnum, p, e, above_latin1) \
1825 (__ASSERT_(_utf8_safe_assert(p, e)) \
1826 (UTF8_IS_INVARIANT(*(p))) \
1827 ? _generic_isCC(*(p), classnum) \
1828 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1829 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1830 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
1831 classnum) \
1832 : (_force_out_malformed_utf8_message( \
1833 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1834 : above_latin1))
b7d90381
KW
1835/* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1836 * 'above_latin1' can be a macro */
da8c1a98
KW
1837#define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
1838 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
1839#define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1840 _generic_utf8_safe(classnum, p, e, \
1841 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1842 ? (_force_out_malformed_utf8_message( \
1843 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1844 : above_latin1(p)))
4650c663 1845/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
cd500f2f 1846 * 'above_latin1' parameter */
da8c1a98
KW
1847#define _generic_swash_utf8_safe(classnum, p, e) \
1848_generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e))
922e8cb4 1849
cc8ab7c0 1850/* Like the above, but should be used only when it is known that there are no
ff7ecfc3
KW
1851 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1852 * class is TRUE for. Hence it can skip the tests for this range.
1853 * 'above_latin1' should include its arguments */
da8c1a98
KW
1854#define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
1855 (__ASSERT_(_utf8_safe_assert(p, e)) \
1856 (UTF8_IS_INVARIANT(*(p))) \
1857 ? _generic_isCC(*(p), classnum) \
1858 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1859 ? 0 /* Note that doesn't check validity for latin1 */ \
1860 : above_latin1)
1861
84238efa
KW
1862/* NOTE that some of these macros have very similar ones in regcharclass.h.
1863 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1864 * there, differing in name only by an underscore from the one here
1865 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1866 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1867 * points; the regcharclass.h ones are implemented as a series of
1868 * "if-else-if-else ..." */
1869
34aeb2e9
KW
1870#define isALPHA_utf8(p) _generic_utf8(ALPHA, p)
1871#define isALPHANUMERIC_utf8(p) _generic_utf8(ALPHANUMERIC, p)
1872#define isASCII_utf8(p) _generic_utf8(ASCII, p)
1873#define isBLANK_utf8(p) _generic_utf8(BLANK, p)
1874#define isCNTRL_utf8(p) _generic_utf8(CNTRL, p)
1875#define isDIGIT_utf8(p) _generic_utf8(DIGIT, p)
1876#define isGRAPH_utf8(p) _generic_utf8(GRAPH, p)
1877#define isIDCONT_utf8(p) _generic_utf8(IDCONT, p)
1878#define isIDFIRST_utf8(p) _generic_utf8(IDFIRST, p)
1879#define isLOWER_utf8(p) _generic_utf8(LOWER, p)
1880#define isPRINT_utf8(p) _generic_utf8(PRINT, p)
1881#define isPSXSPC_utf8(p) _generic_utf8(PSXSPC, p)
1882#define isPUNCT_utf8(p) _generic_utf8(PUNCT, p)
1883#define isSPACE_utf8(p) _generic_utf8(SPACE, p)
1884#define isUPPER_utf8(p) _generic_utf8(UPPER, p)
1885#define isVERTWS_utf8(p) _generic_utf8(VERTSPACE, p)
1886#define isWORDCHAR_utf8(p) _generic_utf8(WORDCHAR, p)
1887#define isXDIGIT_utf8(p) _generic_utf8(XDIGIT, p)
e8fa43e2 1888
da8c1a98
KW
1889#define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e)
1890#define isALPHANUMERIC_utf8_safe(p, e) \
1891 _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e)
1892#define isASCII_utf8_safe(p, e) \
1893 /* Because ASCII is invariant under utf8, the non-utf8 macro \
1894 * works */ \
1895 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
1896#define isBLANK_utf8_safe(p, e) \
1897 _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
1898
e8fa43e2
KW
1899#ifdef EBCDIC
1900 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1901 * more efficient macro instead of the more general one */
da8c1a98 1902# define isCNTRL_utf8_safe(p, e) \
56d02b8c 1903 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
e8fa43e2 1904#else
da8c1a98 1905# define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
e8fa43e2
KW
1906#endif
1907
da8c1a98
KW
1908#define isDIGIT_utf8_safe(p, e) \
1909 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
1910 _is_utf8_FOO_with_len(_CC_DIGIT, p, e))
1911#define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e)
1912#define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
1913 _is_utf8_perl_idcont_with_len, p, e)
e5dcd934 1914
c11ff943
KW
1915/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1916 * IDFIRST is an alnum. See
c8362b00 1917 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
f91dcd13
KW
1918 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1919 * This used to be not the XID version, but we decided to go with the more
1920 * modern Unicode definition */
da8c1a98
KW
1921#define isIDFIRST_utf8_safe(p, e) \
1922 _generic_func_utf8_safe(_CC_IDFIRST, \
1923 _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e))
1924
1925#define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e)
1926#define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e)
1927#define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
1928#define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e)
1929#define isSPACE_utf8_safe(p, e) \
1930 _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
1931#define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e)
1932#define isVERTWS_utf8_safe(p, e) \
1933 _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
1934#define isWORDCHAR_utf8_safe(p, e) \
1935 _generic_swash_utf8_safe(_CC_WORDCHAR, p, e)
1936#define isXDIGIT_utf8_safe(p, e) \
1937 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
1938 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1939 ? (_force_out_malformed_utf8_message( \
1940 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1941 : is_XDIGIT_high(p)))
a0ed51b3 1942
7d2b61dd 1943#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
509fb054
KW
1944#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1945#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1946#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2e8adce6 1947
567b353c 1948/* For internal core use only, subject to change */
607313a1
KW
1949#define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f, "", 0)
1950#define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f, "", 0)
1951#define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f, "", 0)
1952#define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f, "", 0)
a1a5ec35
KW
1953
1954#define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
1955#define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
1956#define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
1957#define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
567b353c 1958
4650c663
KW
1959/* For internal core Perl use only: the base macros for defining macros like
1960 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1961 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1962 * 'macro' parameter */
34aeb2e9
KW
1963#define _generic_LC_utf8(name, p) _base_generic_utf8(name, name, p, 1)
1964
1965#define isALPHA_LC_utf8(p) _generic_LC_utf8(ALPHA, p)
1966#define isALPHANUMERIC_LC_utf8(p) _generic_LC_utf8(ALPHANUMERIC, p)
1967#define isASCII_LC_utf8(p) _generic_LC_utf8(ASCII, p)
1968#define isBLANK_LC_utf8(p) _generic_LC_utf8(BLANK, p)
1969#define isCNTRL_LC_utf8(p) _generic_LC_utf8(CNTRL, p)
1970#define isDIGIT_LC_utf8(p) _generic_LC_utf8(DIGIT, p)
1971#define isGRAPH_LC_utf8(p) _generic_LC_utf8(GRAPH, p)
1972#define isIDCONT_LC_utf8(p) _generic_LC_utf8(IDCONT, p)
1973#define isIDFIRST_LC_utf8(p) _generic_LC_utf8(IDFIRST, p)
1974#define isLOWER_LC_utf8(p) _generic_LC_utf8(LOWER, p)
1975#define isPRINT_LC_utf8(p) _generic_LC_utf8(PRINT, p)
1976#define isPSXSPC_LC_utf8(p) _generic_LC_utf8(PSXSPC, p)
1977#define isPUNCT_LC_utf8(p) _generic_LC_utf8(PUNCT, p)
1978#define isSPACE_LC_utf8(p) _generic_LC_utf8(SPACE, p)
1979#define isUPPER_LC_utf8(p) _generic_LC_utf8(UPPER, p)
1980#define isWORDCHAR_LC_utf8(p) _generic_LC_utf8(WORDCHAR, p)
1981#define isXDIGIT_LC_utf8(p) _generic_LC_utf8(XDIGIT, p)
1982
da8c1a98
KW
1983/* For internal core Perl use only: the base macros for defining macros like
1984 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
1985 * point in 'p' is within the 0-255 range, it uses locale rules from the
1986 * passed-in 'macro' parameter */
1987#define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
1988 (__ASSERT_(_utf8_safe_assert(p, e)) \
1989 (UTF8_IS_INVARIANT(*(p))) \
1990 ? macro(*(p)) \
1991 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1992 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1993 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
1994 : (_force_out_malformed_utf8_message( \
1995 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1996 : above_latin1))
1997
1998#define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \
1999 _generic_LC_utf8_safe(macro, p, e, \
2000 _is_utf8_FOO_with_len(classnum, p, e))
2001
2002#define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2003 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2004
2005#define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \
2006 _generic_LC_utf8_safe(classnum, p, e, \
2007 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2008 ? (_force_out_malformed_utf8_message( \
2009 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2010 : above_latin1(p)))
2011
2012#define isALPHANUMERIC_LC_utf8_safe(p, e) \
2013 _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \
2014 _CC_ALPHANUMERIC, p, e)
2015#define isALPHA_LC_utf8_safe(p, e) \
2016 _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2017#define isASCII_LC_utf8_safe(p, e) \
2018 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2019#define isBLANK_LC_utf8_safe(p, e) \
2020 _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2021#define isCNTRL_LC_utf8_safe(p, e) \
2022 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2023#define isDIGIT_LC_utf8_safe(p, e) \
2024 _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2025#define isGRAPH_LC_utf8_safe(p, e) \
2026 _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2027#define isIDCONT_LC_utf8_safe(p, e) \
2028 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2029 _is_utf8_perl_idcont_with_len, p, e)
2030#define isIDFIRST_LC_utf8_safe(p, e) \
2031 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2032 _is_utf8_perl_idstart_with_len, p, e)
2033#define isLOWER_LC_utf8_safe(p, e) \
2034 _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2035#define isPRINT_LC_utf8_safe(p, e) \
2036 _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2037#define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2038#define isPUNCT_LC_utf8_safe(p, e) \
2039 _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2040#define isSPACE_LC_utf8_safe(p, e) \
2041 _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2042#define isUPPER_LC_utf8_safe(p, e) \
2043 _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2044#define isWORDCHAR_LC_utf8_safe(p, e) \
2045 _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2046#define isXDIGIT_LC_utf8_safe(p, e) \
2047 _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
aaa51d5e 2048
fbc19f27
KW
2049/* Macros for backwards compatibility and for completeness when the ASCII and
2050 * Latin1 values are identical */
b7d90381
KW
2051#define isALPHAU(c) isALPHA_L1(c)
2052#define isDIGIT_L1(c) isDIGIT_A(c)
2053#define isOCTAL(c) isOCTAL_A(c)
2054#define isOCTAL_L1(c) isOCTAL_A(c)
2055#define isXDIGIT_L1(c) isXDIGIT_A(c)
2056#define isALNUM(c) isWORDCHAR(c)
2057#define isALNUMU(c) isWORDCHAR_L1(c)
2058#define isALNUM_LC(c) isWORDCHAR_LC(c)
2059#define isALNUM_uni(c) isWORDCHAR_uni(c)
2e28f0b9 2060#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
b7d90381
KW
2061#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
2062#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
2063#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2064#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2065#define isALNUMC(c) isALPHANUMERIC(c)
2066#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2067#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
15861f94 2068#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
b7d90381 2069#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
15861f94 2070#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
fbc19f27 2071
2bd1cbf6
KW
2072/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2073 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2074 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2075 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2076 * the outlier from the block that contains the other controls, just like
2077 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2078 * block. If it weren't special cased, it would yield a non-control.
88794300
KW
2079 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2080 * etc. */
2bd1cbf6 2081#ifndef EBCDIC
75763b3a 2082# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2bd1cbf6 2083#else
75763b3a
KW
2084# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2085 ((isPRINT_A(c)) \
2086 ? (UNLIKELY((c) == '?') \
2087 ? QUESTION_MARK_CTRL \
2088 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2089 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2090 ? '?' \
2091 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2bd1cbf6 2092#endif
bbce6d69 2093
dea28490
JJ
2094/* Line numbers are unsigned, 32 bits. */
2095typedef U32 line_t;
e5dcd934 2096#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
378cc40b 2097
91152fc1
DG
2098/* Helpful alias for version prescan */
2099#define is_LAX_VERSION(a,b) \
2100 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2101
2102#define is_STRICT_VERSION(a,b) \
2103 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2104
2105#define BADVERSION(a,b,c) \
2106 if (b) { \
2107 *b = c; \
2108 } \
2109 return a;
8c52afec 2110
cb27eebd
KW
2111/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2112 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2113 * which is advanced. The input is validated only by an assert() in DEBUGGING
2114 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2115 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2116#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2117 ? (c) \
2118 : ((c) + 9))))
902008b8
KW
2119#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2120 ? (*(s)++) \
2121 : (*(s)++ + 9))))
95a59cab 2122
cb27eebd
KW
2123/* Converts a character known to represent an octal digit (0-7) to its numeric
2124 * value. The input is validated only by an assert() in DEBUGGING builds. In
2125 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2126#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2127
305b8651
KW
2128/* Efficiently returns a boolean as to if two native characters are equivalent
2129 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2130 * the ALPHA in the name is to remind you of that. This is asserted() in
2131 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2132 * works (on valid input) for both non- and UTF-8-encoded bytes.
2133 *
2134 * When one of the inputs is a compile-time constant and gets folded by the
2135 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2136 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2137 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
96ca48da
KW
2138 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2139 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2140 * just a single 0, in the bit position where the upper- and lowercase differ.
2141 * */
305b8651
KW
2142#define isALPHA_FOLD_EQ(c1, c2) \
2143 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2144 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2145#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2146
8e84507e 2147/*
ccfc67b7
JH
2148=head1 Memory Management
2149
a02a5408 2150=for apidoc Am|void|Newx|void* ptr|int nitems|type
954c1994
GS
2151The XSUB-writer's interface to the C C<malloc> function.
2152
596f7718 2153Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2154
c5008215
JC
2155In 5.9.3, Newx() and friends replace the older New() API, and drops
2156the first parameter, I<x>, a debug aid which allowed callers to identify
37b8b4c9 2157themselves. This aid has been superseded by a new build option,
d10b4965 2158PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
c5008215
JC
2159there for use in XS modules supporting older perls.
2160
a02a5408 2161=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
954c1994 2162The XSUB-writer's interface to the C C<malloc> function, with
fbe13c60 2163cast. See also C<L</Newx>>.
954c1994 2164
596f7718 2165Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2166
a02a5408 2167=for apidoc Am|void|Newxz|void* ptr|int nitems|type
954c1994 2168The XSUB-writer's interface to the C C<malloc> function. The allocated
fbe13c60 2169memory is zeroed with C<memzero>. See also C<L</Newx>>.
a02a5408 2170
596f7718 2171Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2172
954c1994
GS
2173=for apidoc Am|void|Renew|void* ptr|int nitems|type
2174The XSUB-writer's interface to the C C<realloc> function.
2175
596f7718 2176Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2177
954c1994
GS
2178=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2179The XSUB-writer's interface to the C C<realloc> function, with
2180cast.
2181
596f7718 2182Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
0d7b2759 2183
49b8b560 2184=for apidoc Am|void|Safefree|void* ptr
954c1994
GS
2185The XSUB-writer's interface to the C C<free> function.
2186
596f7718 2187This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
0d7b2759 2188
954c1994
GS
2189=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2190The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
926bb54c 2191source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2192C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
954c1994 2193
e90e2364 2194=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
796b6530 2195Like C<Move> but returns C<dest>. Useful
72d33970 2196for encouraging compilers to tail-call
e90e2364
NC
2197optimise.
2198
954c1994
GS
2199=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2200The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
926bb54c 2201source, C<dest> is the destination, C<nitems> is the number of items, and
fbe13c60 2202C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
954c1994 2203
e90e2364
NC
2204=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2205
796b6530 2206Like C<Copy> but returns C<dest>. Useful
72d33970 2207for encouraging compilers to tail-call
e90e2364
NC
2208optimise.
2209
954c1994
GS
2210=for apidoc Am|void|Zero|void* dest|int nitems|type
2211
2212The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2213destination, C<nitems> is the number of items, and C<type> is the type.
2214
e90e2364
NC
2215=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2216
72d33970
FC
2217Like C<Zero> but returns dest. Useful
2218for encouraging compilers to tail-call
e90e2364
NC
2219optimise.
2220
da5d8dbb 2221=for apidoc Am|void|StructCopy|type *src|type *dest|type
4375e838 2222This is an architecture-independent macro to copy one structure to another.
954c1994 2223
7e337ee0
JH
2224=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2225
2226Fill up memory with a byte pattern (a byte repeated over and over
2227again) that hopefully catches attempts to access uninitialized memory.
2228
2229=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2230
2231PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2232
1c12ffb4 2233=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
7e337ee0
JH
2234
2235PoisonWith(0xEF) for catching access to freed memory.
2236
9965345d
JH
2237=for apidoc Am|void|Poison|void* dest|int nitems|type
2238
7e337ee0 2239PoisonWith(0xEF) for catching access to freed memory.
9965345d
JH
2240
2241=cut */
954c1994 2242
561b68a9
SH
2243/* Maintained for backwards-compatibility only. Use newSV() instead. */
2244#ifndef PERL_CORE
ff06c60c 2245#define NEWSV(x,len) newSV(len)
561b68a9 2246#endif
ff06c60c 2247
19a94d75
JH
2248#define MEM_SIZE_MAX ((MEM_SIZE)~0)
2249
e6bdf523 2250
27d5b266 2251#ifdef PERL_MALLOC_WRAP
e6bdf523
DM
2252
2253/* This expression will be constant-folded at compile time. It checks
2254 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2255 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2256 * As well as avoiding the need for a run-time check in some cases, it's
2257 * designed to avoid compiler warnings like:
2258 * comparison is always false due to limited range of data type
73e8ff00
DM
2259 * It's mathematically equivalent to
2260 * max(n) * sizeof(t) > MEM_SIZE_MAX
e6bdf523
DM
2261 */
2262
2263# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
004073ba 2264 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
e6bdf523 2265
88f9f128 2266/* This is written in a slightly odd way to avoid various spurious
d98e5cde
DM
2267 * compiler warnings. We *want* to write the expression as
2268 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2269 * (for some compile-time constant C), but even when the LHS
2270 * constant-folds to false at compile-time, g++ insists on emitting
2271 * warnings about the RHS (e.g. "comparison is always false"), so instead
2272 * we write it as
e6bdf523 2273 *
d98e5cde 2274 * (cond ? n : X) > C
88f9f128 2275 *
d98e5cde
DM
2276 * where X is a constant with X > C always false. Choosing a value for X
2277 * is tricky. If 0, some compilers will complain about 0 > C always being
2278 * false; if 1, Coverity complains when n happens to be the constant value
2279 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2280 * for X and hope that nothing else whines.
e6bdf523
DM
2281 */
2282
2283# define _MEM_WRAP_WILL_WRAP(n,t) \
88f9f128
DM
2284 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2285 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
e6bdf523
DM
2286
2287# define MEM_WRAP_CHECK(n,t) \
2288 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2289 && (croak_memory_wrap(),0))
2290
2291# define MEM_WRAP_CHECK_1(n,t,a) \
2292 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2293 && (Perl_croak_nocontext("%s",(a)),0))
2294
8b44ba4c 2295#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
27d5b266 2296
d1decf2b 2297#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
2298#else
2299
410319be
NC
2300#define MEM_WRAP_CHECK(n,t)
2301#define MEM_WRAP_CHECK_1(n,t,a)
2302#define MEM_WRAP_CHECK_2(n,t,a,b)
8b44ba4c
NC
2303#define MEM_WRAP_CHECK_(n,t)
2304
6fff79ce 2305#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
27d5b266 2306
1936d2a7 2307#endif
8b44ba4c 2308
fe4f188c 2309#ifdef PERL_MEM_LOG
46c6c7e2 2310/*
9f653bb5 2311 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
46c6c7e2
JH
2312 * go through functions, which are handy for debugging breakpoints, but
2313 * which more importantly get the immediate calling environment (file and
e352bcff
JH
2314 * line number, and C function name if available) passed in. This info can
2315 * then be used for logging the calls, for which one gets a sample
73d1d973 2316 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
3609ea0d 2317 *
46c6c7e2 2318 * Known problems:
94e892a6 2319 * - not all memory allocs get logged, only those
46c6c7e2 2320 * that go through Newx() and derivatives (while all
94e892a6 2321 * Safefrees do get logged)
46c6c7e2
JH
2322 * - __FILE__ and __LINE__ do not work everywhere
2323 * - __func__ or __FUNCTION__ even less so
2324 * - I think more goes on after the perlio frees but
2325 * the thing is that STDERR gets closed (as do all
2326 * the file descriptors)
2327 * - no deeper calling stack than the caller of the Newx()
2328 * or the kind, but do I look like a C reflection/introspection
2329 * utility to you?
2330 * - the function prototypes for the logging functions
2331 * probably should maybe be somewhere else than handy.h
2332 * - one could consider inlining (macrofying) the logging
2333 * for speed, but I am too lazy
2334 * - one could imagine recording the allocations in a hash,
2335 * (keyed by the allocation address?), and maintain that
2336 * through reallocs and frees, but how to do that without
2337 * any News() happening...?
73d1d973 2338 * - lots of -Ddefines to get useful/controllable output
b953482e 2339 * - lots of ENV reads
46c6c7e2
JH
2340 */
2341
0b0ab801 2342# ifdef PERL_CORE
73d1d973 2343# ifndef PERL_MEM_LOG_NOIMPL
0b0ab801
MHM
2344enum mem_log_type {
2345 MLT_ALLOC,
2346 MLT_REALLOC,
d7a2c63c
MHM
2347 MLT_FREE,
2348 MLT_NEW_SV,
2349 MLT_DEL_SV
0b0ab801
MHM
2350};
2351# endif
12754f92 2352# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
d7a2c63c
MHM
2353void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2354void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
12754f92 2355# endif
0b0ab801
MHM
2356# endif
2357
fe4f188c
JH
2358#endif
2359
2360#ifdef PERL_MEM_LOG
d1401ee9
MHM
2361#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2362#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
46c6c7e2 2363#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
fe4f188c
JH
2364#endif
2365
2366#ifndef MEM_LOG_ALLOC
2367#define MEM_LOG_ALLOC(n,t,a) (a)
2368#endif
2369#ifndef MEM_LOG_REALLOC
2370#define MEM_LOG_REALLOC(n,t,v,a) (a)
2371#endif
2372#ifndef MEM_LOG_FREE
2373#define MEM_LOG_FREE(a) (a)
2374#endif
2375
d1401ee9
MHM
2376#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2377#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2378#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
a6f6820f
NC
2379
2380#ifndef PERL_CORE
a02a5408
JC
2381/* pre 5.9.x compatibility */
2382#define New(x,v,n,t) Newx(v,n,t)
2383#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
4541904d 2384#define Newz(x,v,n,t) Newxz(v,n,t)
a6f6820f 2385#endif
a02a5408 2386
ff68c719 2387#define Renew(v,n,t) \
d1401ee9 2388 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
ff68c719 2389#define Renewc(v,n,t,c) \
d1401ee9 2390 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
94010e71
NC
2391
2392#ifdef PERL_POISON
2393#define Safefree(d) \
06c0cc96 2394 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
94010e71 2395#else
fe4f188c 2396#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
94010e71 2397#endif
55497cff 2398
45908e4d 2399#define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
55497cff 2400
45908e4d
YO
2401
2402#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2403#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2404#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2405
2406#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2407#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
45908e4d 2408#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
e90e2364 2409
7e337ee0
JH
2410#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2411#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2412#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2413#define Poison(d,n,t) PoisonFree(d,n,t)
27d5b266 2414
caa674f3
DD
2415#ifdef PERL_POISON
2416# define PERL_POISON_EXPR(x) x
2417#else
2418# define PERL_POISON_EXPR(x)
2419#endif
2420
ff68c719 2421#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2cc61e15 2422
c3caa5c3
JH
2423/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2424 * want your zero-based indices to be less than but not equal to).
2425 *
2426 * C_ARRAY_END is one past the last: half-open/half-closed range,
2427 * not last-inclusive range. */
622913ab 2428#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
c3caa5c3 2429#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
622913ab 2430
2cc61e15
DD
2431#ifdef NEED_VA_COPY
2432# ifdef va_copy
2433# define Perl_va_copy(s, d) va_copy(d, s)
07798b17
AC
2434# elif defined(__va_copy)
2435# define Perl_va_copy(s, d) __va_copy(d, s)
2cc61e15 2436# else
07798b17 2437# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2cc61e15
DD
2438# endif
2439#endif
2440
472d47bc
SB
2441/* convenience debug macros */
2442#ifdef USE_ITHREADS
2443#define pTHX_FORMAT "Perl interpreter: 0x%p"
2444#define pTHX__FORMAT ", Perl interpreter: 0x%p"
f54cb97a
AL
2445#define pTHX_VALUE_ (void *)my_perl,
2446#define pTHX_VALUE (void *)my_perl
2447#define pTHX__VALUE_ ,(void *)my_perl,
2448#define pTHX__VALUE ,(void *)my_perl
472d47bc 2449#else
3609ea0d 2450#define pTHX_FORMAT
472d47bc 2451#define pTHX__FORMAT
3609ea0d 2452#define pTHX_VALUE_
472d47bc 2453#define pTHX_VALUE
3609ea0d 2454#define pTHX__VALUE_
472d47bc
SB
2455#define pTHX__VALUE
2456#endif /* USE_ITHREADS */
3609ea0d 2457
2acdbac1
NC
2458/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2459 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2460 CPAN::Unpack show any users outside the core. */
2461#ifdef PERL_CORE
dc6e8de0
A
2462# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2463 "Use of " s " is deprecated")
c9680906
A
2464# define deprecate_disappears_in(when,message) \
2465 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2466 message ", and will disappear in Perl " when)
ac641426
A
2467# define deprecate_fatal_in(when,message) \
2468 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2469 message ". Its use will be fatal in Perl " when)
2acdbac1
NC
2470#endif
2471
dfff4baf
BF
2472/* Internal macros to deal with gids and uids */
2473#ifdef PERL_CORE
2474
2475# if Uid_t_size > IVSIZE
2476# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2477# define SvUID(sv) SvNV(sv)
07798b17
AC
2478# elif Uid_t_sign <= 0
2479# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2480# define SvUID(sv) SvIV(sv)
dfff4baf 2481# else
07798b17
AC
2482# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2483# define SvUID(sv) SvUV(sv)
dfff4baf
BF
2484# endif /* Uid_t_size */
2485
2486# if Gid_t_size > IVSIZE
2487# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2488# define SvGID(sv) SvNV(sv)
07798b17
AC
2489# elif Gid_t_sign <= 0
2490# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2491# define SvGID(sv) SvIV(sv)
dfff4baf 2492# else
07798b17
AC
2493# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2494# define SvGID(sv) SvUV(sv)
dfff4baf
BF
2495# endif /* Gid_t_size */
2496
2497#endif
2498
6a5bc5ac 2499#endif /* PERL_HANDY_H_ */
9d745869 2500
e9a8c099 2501/*
14d04a33 2502 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 2503 */