This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
https://perl5.git.perl.org / perl5.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
mktables: Reduce input fractions to lowest terms
[perl5.git] / handy.h
... / ...
CommitLineData
1/* handy.h
2 *
3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
5 *
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.
8 *
9 */
10
11/* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
13
14#ifndef HANDY_H /* Guard against nested #inclusion */
15#define HANDY_H
16
17#if !defined(__STDC__)
18#ifdef NULL
19#undef NULL
20#endif
21# define NULL 0
22#endif
23
24#ifndef PERL_CORE
25# define Null(type) ((type)NULL)
26
27/*
28=head1 Handy Values
29
30=for apidoc AmU||Nullch
31Null character pointer. (No longer available when C<PERL_CORE> is
32defined.)
33
34=for apidoc AmU||Nullsv
35Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
36
37=cut
38*/
39
40# define Nullch Null(char*)
41# define Nullfp Null(PerlIO*)
42# define Nullsv Null(SV*)
43#endif
44
45#ifdef TRUE
46#undef TRUE
47#endif
48#ifdef FALSE
49#undef FALSE
50#endif
51#define TRUE (1)
52#define FALSE (0)
53
54/* The MUTABLE_*() macros cast pointers to the types shown, in such a way
55 * (compiler permitting) that casting away const-ness will give a warning;
56 * e.g.:
57 *
58 * const SV *sv = ...;
59 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
60 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
61 */
62
63#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
64# define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
65#else
66# define MUTABLE_PTR(p) ((void *) (p))
67#endif
68
69#define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
70#define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
71#define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
72#define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
73#define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
74#define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
75
76#if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
77# include <stdbool.h>
78# ifndef HAS_BOOL
79# define HAS_BOOL 1
80# endif
81#endif
82
83/* bool is built-in for g++-2.6.3 and later, which might be used
84 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
85 be sure _G_config.h will be included before this file. _G_config.h
86 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
87 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
88 g++ can be identified by __GNUG__.
89 Andy Dougherty February 2000
90*/
91#ifdef __GNUG__ /* GNU g++ has bool built-in */
92# ifndef PERL_BOOL_AS_CHAR
93# ifndef HAS_BOOL
94# define HAS_BOOL 1
95# endif
96# endif
97#endif
98
99#ifndef HAS_BOOL
100# ifdef bool
101# undef bool
102# endif
103# define bool char
104# define HAS_BOOL 1
105#endif
106
107/* cast-to-bool. A simple (bool) cast may not do the right thing: if bool is
108 * defined as char for example, then the cast from int is
109 * implementation-defined (bool)!!(cbool) in a ternary triggers a bug in xlc on
110 * AIX */
111#define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
112
113/* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
114 * XXX Should really be a Configure probe, with HAS__FUNCTION__
115 * and FUNCTION__ as results.
116 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
117#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
118# define FUNCTION__ __func__
119#else
120# if (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
121 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
122# define FUNCTION__ ""
123# else
124# define FUNCTION__ __FUNCTION__ /* Common extension. */
125# endif
126#endif
127
128/* XXX A note on the perl source internal type system. The
129 original intent was that I32 be *exactly* 32 bits.
130
131 Currently, we only guarantee that I32 is *at least* 32 bits.
132 Specifically, if int is 64 bits, then so is I32. (This is the case
133 for the Cray.) This has the advantage of meshing nicely with
134 standard library calls (where we pass an I32 and the library is
135 expecting an int), but the disadvantage that an I32 is not 32 bits.
136 Andy Dougherty August 1996
137
138 There is no guarantee that there is *any* integral type with
139 exactly 32 bits. It is perfectly legal for a system to have
140 sizeof(short) == sizeof(int) == sizeof(long) == 8.
141
142 Similarly, there is no guarantee that I16 and U16 have exactly 16
143 bits.
144
145 For dealing with issues that may arise from various 32/64-bit
146 systems, we will ask Configure to check out
147
148 SHORTSIZE == sizeof(short)
149 INTSIZE == sizeof(int)
150 LONGSIZE == sizeof(long)
151 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
152 PTRSIZE == sizeof(void *)
153 DOUBLESIZE == sizeof(double)
154 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
155
156*/
157
158#ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
159# include <inttypes.h>
160# ifdef INT32_MIN_BROKEN
161# undef INT32_MIN
162# define INT32_MIN (-2147483647-1)
163# endif
164# ifdef INT64_MIN_BROKEN
165# undef INT64_MIN
166# define INT64_MIN (-9223372036854775807LL-1)
167# endif
168#endif
169
170typedef I8TYPE I8;
171typedef U8TYPE U8;
172typedef I16TYPE I16;
173typedef U16TYPE U16;
174typedef I32TYPE I32;
175typedef U32TYPE U32;
176#ifdef PERL_CORE
177# ifdef HAS_QUAD
178typedef I64TYPE I64;
179typedef U64TYPE U64;
180# endif
181#endif /* PERL_CORE */
182
183/* INT64_C/UINT64_C are C99 from <stdint.h> (so they will not be
184 * available in strict C89 mode), but they are nice, so let's define
185 * them if necessary. */
186#if defined(HAS_QUAD)
187# undef PeRl_INT64_C
188# undef PeRl_UINT64_C
189/* Prefer the native integer types (int and long) over long long
190 * (which is not C89) and Win32-specific __int64. */
191# if QUADKIND == QUAD_IS_INT && INTSIZE == 8
192# define PeRl_INT64_C(c) (c)
193# define PeRl_UINT64_C(c) CAT2(c,U)
194# endif
195# if QUADKIND == QUAD_IS_LONG && LONGSIZE == 8
196# define PeRl_INT64_C(c) CAT2(c,L)
197# define PeRl_UINT64_C(c) CAT2(c,UL)
198# endif
199# if QUADKIND == QUAD_IS_LONG_LONG && defined(HAS_LONG_LONG)
200# define PeRl_INT64_C(c) CAT2(c,LL)
201# define PeRl_UINT64_C(c) CAT2(c,ULL)
202# endif
203# if QUADKIND == QUAD_IS___INT64
204# define PeRl_INT64_C(c) CAT2(c,I64)
205# define PeRl_UINT64_C(c) CAT2(c,UI64)
206# endif
207# ifndef PeRl_INT64_C
208# define PeRl_INT64_C(c) ((I64TYPE)(c)) /* last resort */
209# define PeRl_UINT64_C(c) ((U64TYPE)(c))
210# endif
211/* In OS X the INT64_C/UINT64_C are defined with LL/ULL, which will
212 * not fly with C89-pedantic gcc, so let's undefine them first so that
213 * we can redefine them with our native integer preferring versions. */
214# if defined(PERL_DARWIN) && defined(PERL_GCC_PEDANTIC)
215# undef INT64_C
216# undef UINT64_C
217# endif
218# ifndef INT64_C
219# define INT64_C(c) PeRl_INT64_C(c)
220# endif
221# ifndef UINT64_C
222# define UINT64_C(c) PeRl_UINT64_C(c)
223# endif
224#endif
225
226#if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
227
228/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
229 Please search CHAR_MAX in perl.h for further details. */
230#define U8_MAX UINT8_MAX
231#define U8_MIN UINT8_MIN
232
233#define I16_MAX INT16_MAX
234#define I16_MIN INT16_MIN
235#define U16_MAX UINT16_MAX
236#define U16_MIN UINT16_MIN
237
238#define I32_MAX INT32_MAX
239#define I32_MIN INT32_MIN
240#ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
241# define U32_MAX UINT32_MAX
242#else
243# define U32_MAX 4294967295U
244#endif
245#define U32_MIN UINT32_MIN
246
247#else
248
249/* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
250 Please search CHAR_MAX in perl.h for further details. */
251#define U8_MAX PERL_UCHAR_MAX
252#define U8_MIN PERL_UCHAR_MIN
253
254#define I16_MAX PERL_SHORT_MAX
255#define I16_MIN PERL_SHORT_MIN
256#define U16_MAX PERL_USHORT_MAX
257#define U16_MIN PERL_USHORT_MIN
258
259#if LONGSIZE > 4
260# define I32_MAX PERL_INT_MAX
261# define I32_MIN PERL_INT_MIN
262# define U32_MAX PERL_UINT_MAX
263# define U32_MIN PERL_UINT_MIN
264#else
265# define I32_MAX PERL_LONG_MAX
266# define I32_MIN PERL_LONG_MIN
267# define U32_MAX PERL_ULONG_MAX
268# define U32_MIN PERL_ULONG_MIN
269#endif
270
271#endif
272
273/* log(2) is pretty close to 0.30103, just in case anyone is grepping for it */
274#define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
275#define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
276#define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
277
278/* Unused by core; should be deprecated */
279#define Ctl(ch) ((ch) & 037)
280
281/* This is a helper macro to avoid preprocessor issues, replaced by nothing
282 * unless under DEBUGGING, where it expands to an assert of its argument,
283 * followed by a comma (hence the comma operator). If we just used a straight
284 * assert(), we would get a comma with nothing before it when not DEBUGGING */
285#ifdef DEBUGGING
286# define __ASSERT_(statement) assert(statement),
287#else
288# define __ASSERT_(statement)
289#endif
290
291/*
292=head1 SV-Body Allocation
293
294=for apidoc Ama|SV*|newSVpvs|const char* s
295Like C<newSVpvn>, but takes a literal C<NUL>-terminated string instead of a
296string/length pair.
297
298=for apidoc Ama|SV*|newSVpvs_flags|const char* s|U32 flags
299Like C<newSVpvn_flags>, but takes a literal C<NUL>-terminated string instead of
300a string/length pair.
301
302=for apidoc Ama|SV*|newSVpvs_share|const char* s
303Like C<newSVpvn_share>, but takes a literal C<NUL>-terminated string instead of
304a string/length pair and omits the hash parameter.
305
306=for apidoc Am|void|sv_catpvs_flags|SV* sv|const char* s|I32 flags
307Like C<sv_catpvn_flags>, but takes a literal C<NUL>-terminated string instead
308of a string/length pair.
309
310=for apidoc Am|void|sv_catpvs_nomg|SV* sv|const char* s
311Like C<sv_catpvn_nomg>, but takes a literal string instead of a
312string/length pair.
313
314=for apidoc Am|void|sv_catpvs|SV* sv|const char* s
315Like C<sv_catpvn>, but takes a literal string instead of a string/length pair.
316
317=for apidoc Am|void|sv_catpvs_mg|SV* sv|const char* s
318Like C<sv_catpvn_mg>, but takes a literal string instead of a
319string/length pair.
320
321=for apidoc Am|void|sv_setpvs|SV* sv|const char* s
322Like C<sv_setpvn>, but takes a literal string instead of a string/length pair.
323
324=for apidoc Am|void|sv_setpvs_mg|SV* sv|const char* s
325Like C<sv_setpvn_mg>, but takes a literal string instead of a
326string/length pair.
327
328=for apidoc Am|SV *|sv_setref_pvs|const char* s
329Like C<sv_setref_pvn>, but takes a literal string instead of a
330string/length pair.
331
332=head1 Memory Management
333
334=for apidoc Ama|char*|savepvs|const char* s
335Like C<savepvn>, but takes a literal C<NUL>-terminated string instead of a
336string/length pair.
337
338=for apidoc Ama|char*|savesharedpvs|const char* s
339A version of C<savepvs()> which allocates the duplicate string in memory
340which is shared between threads.
341
342=head1 GV Functions
343
344=for apidoc Am|HV*|gv_stashpvs|const char* name|I32 create
345Like C<gv_stashpvn>, but takes a literal string instead of a string/length pair.
346
347=head1 Hash Manipulation Functions
348
349=for apidoc Am|SV**|hv_fetchs|HV* tb|const char* key|I32 lval
350Like C<hv_fetch>, but takes a literal string instead of a string/length pair.
351
352=for apidoc Am|SV**|hv_stores|HV* tb|const char* key|NULLOK SV* val
353Like C<hv_store>, but takes a literal string instead of a string/length pair
354and omits the hash parameter.
355
356=head1 Lexer interface
357
358=for apidoc Amx|void|lex_stuff_pvs|const char *pv|U32 flags
359
360Like L</lex_stuff_pvn>, but takes a literal string instead of a
361string/length pair.
362
363=cut
364*/
365
366/* concatenating with "" ensures that only literal strings are accepted as
367 * argument */
368#define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
369
370/* note that STR_WITH_LEN() can't be used as argument to macros or functions
371 * that under some configurations might be macros, which means that it requires
372 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
373 */
374
375/* STR_WITH_LEN() shortcuts */
376#define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
377#define newSVpvs_flags(str,flags) \
378 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
379#define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
380#define sv_catpvs_flags(sv, str, flags) \
381 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
382#define sv_catpvs_nomg(sv, str) \
383 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
384#define sv_catpvs(sv, str) \
385 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
386#define sv_catpvs_mg(sv, str) \
387 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
388#define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
389#define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
390#define sv_setref_pvs(rv, classname, str) \
391 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
392#define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
393#define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
394#define gv_stashpvs(str, create) \
395 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
396#define gv_fetchpvs(namebeg, add, sv_type) \
397 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
398#define gv_fetchpvn(namebeg, len, add, sv_type) \
399 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
400#define sv_catxmlpvs(dsv, str, utf8) \
401 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
402#define hv_fetchs(hv,key,lval) \
403 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
404 (lval) ? (HV_FETCH_JUST_SV | HV_FETCH_LVALUE) \
405 : HV_FETCH_JUST_SV, NULL, 0))
406
407#define hv_stores(hv,key,val) \
408 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
409 (HV_FETCH_ISSTORE|HV_FETCH_JUST_SV), (val), 0))
410
411#define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
412
413#define get_cvs(str, flags) \
414 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
415
416/*
417=head1 Miscellaneous Functions
418
419=for apidoc Am|bool|strNE|char* s1|char* s2
420Test two strings to see if they are different. Returns true or
421false.
422
423=for apidoc Am|bool|strEQ|char* s1|char* s2
424Test two strings to see if they are equal. Returns true or false.
425
426=for apidoc Am|bool|strLT|char* s1|char* s2
427Test two strings to see if the first, C<s1>, is less than the second,
428C<s2>. Returns true or false.
429
430=for apidoc Am|bool|strLE|char* s1|char* s2
431Test two strings to see if the first, C<s1>, is less than or equal to the
432second, C<s2>. Returns true or false.
433
434=for apidoc Am|bool|strGT|char* s1|char* s2
435Test two strings to see if the first, C<s1>, is greater than the second,
436C<s2>. Returns true or false.
437
438=for apidoc Am|bool|strGE|char* s1|char* s2
439Test two strings to see if the first, C<s1>, is greater than or equal to
440the second, C<s2>. Returns true or false.
441
442=for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
443Test two strings to see if they are different. The C<len> parameter
444indicates the number of bytes to compare. Returns true or false. (A
445wrapper for C<strncmp>).
446
447=for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
448Test two strings to see if they are equal. The C<len> parameter indicates
449the number of bytes to compare. Returns true or false. (A wrapper for
450C<strncmp>).
451
452=for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
453Test two buffers (which may contain embedded C<NUL> characters, to see if they
454are equal. The C<len> parameter indicates the number of bytes to compare.
455Returns zero if equal, or non-zero if non-equal.
456
457=for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
458Test two buffers (which may contain embedded C<NUL> characters, to see if they
459are not equal. The C<len> parameter indicates the number of bytes to compare.
460Returns zero if non-equal, or non-zero if equal.
461
462=cut
463*/
464
465#define strNE(s1,s2) (strcmp(s1,s2))
466#define strEQ(s1,s2) (!strcmp(s1,s2))
467#define strLT(s1,s2) (strcmp(s1,s2) < 0)
468#define strLE(s1,s2) (strcmp(s1,s2) <= 0)
469#define strGT(s1,s2) (strcmp(s1,s2) > 0)
470#define strGE(s1,s2) (strcmp(s1,s2) >= 0)
471#define strnNE(s1,s2,l) (strncmp(s1,s2,l))
472#define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
473
474#ifdef HAS_MEMCMP
475# define memNE(s1,s2,l) (memcmp(s1,s2,l))
476# define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
477#else
478# define memNE(s1,s2,l) (bcmp(s1,s2,l))
479# define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
480#endif
481
482#define memEQs(s1, l, s2) \
483 (sizeof(s2)-1 == l && memEQ(s1, ("" s2 ""), (sizeof(s2)-1)))
484#define memNEs(s1, l, s2) !memEQs(s1, l, s2)
485
486/*
487 * Character classes.
488 *
489 * Unfortunately, the introduction of locales means that we
490 * can't trust isupper(), etc. to tell the truth. And when
491 * it comes to /\w+/ with tainting enabled, we *must* be able
492 * to trust our character classes.
493 *
494 * Therefore, the default tests in the text of Perl will be
495 * independent of locale. Any code that wants to depend on
496 * the current locale will use the tests that begin with "lc".
497 */
498
499#ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
500# ifndef CTYPE256
501# define CTYPE256
502# endif
503#endif
504
505/*
506
507=head1 Character classification
508This section is about functions (really macros) that classify characters
509into types, such as punctuation versus alphabetic, etc. Most of these are
510analogous to regular expression character classes. (See
511L<perlrecharclass/POSIX Character Classes>.) There are several variants for
512each class. (Not all macros have all variants; each item below lists the
513ones valid for it.) None are affected by C<use bytes>, and only the ones
514with C<LC> in the name are affected by the current locale.
515
516The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
517C<U8>) as input and returns a boolean as to whether or not the character
518represented by that octet is (or on non-ASCII platforms, corresponds to) an
519ASCII character in the named class based on platform, Unicode, and Perl rules.
520If the input is a number that doesn't fit in an octet, FALSE is returned.
521
522Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
523with no suffix C<"_A">.
524
525Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
526onto the platform. That is, the code points that are ASCII are unaffected,
527since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
528as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
529true when called with the code point 0xDF, which is a word character in both
530ASCII and EBCDIC (though it represents different characters in each).
531
532Variant C<isFOO_uni> is like the C<isFOO_L1> variant, but accepts any UV code
533point as input. If the code point is larger than 255, Unicode rules are used
534to determine if it is in the character class. For example,
535C<isWORDCHAR_uni(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
536WITH MACRON in Unicode, and is a word character.
537
538Variant C<isFOO_utf8> is like C<isFOO_uni>, but the input is a pointer to a
539(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
540classification of just the first (possibly multi-byte) character in the string
541is tested.
542
543Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
544result is based on the current locale, which is what C<LC> in the name stands
545for. If Perl can determine that the current locale is a UTF-8 locale, it uses
546the published Unicode rules; otherwise, it uses the C library function that
547gives the named classification. For example, C<isDIGIT_LC()> when not in a
548UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
549returned if the input won't fit into an octet. On some platforms where the C
550library function is known to be defective, Perl changes its result to follow
551the POSIX standard's rules.
552
553Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
554returns the same as C<isFOO_LC> for input code points less than 256, and
555returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
556
557Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer to a
558(known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
559classification of just the first (possibly multi-byte) character in the string
560is tested.
561
562=for apidoc Am|bool|isALPHA|char ch
563Returns a boolean indicating whether the specified character is an
564alphabetic character, analogous to C<m/[[:alpha:]]/>.
565See the L<top of this section|/Character classification> for an explanation of
566variants
567C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uni>, C<isALPHA_utf8>, C<isALPHA_LC>,
568C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
569
570=for apidoc Am|bool|isALPHANUMERIC|char ch
571Returns a boolean indicating whether the specified character is a either an
572alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
573See the L<top of this section|/Character classification> for an explanation of
574variants
575C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uni>,
576C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
577C<isALPHANUMERIC_LC_utf8>.
578
579=for apidoc Am|bool|isASCII|char ch
580Returns a boolean indicating whether the specified character is one of the 128
581characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
582On non-ASCII platforms, it returns TRUE iff this
583character corresponds to an ASCII character. Variants C<isASCII_A()> and
584C<isASCII_L1()> are identical to C<isASCII()>.
585See the L<top of this section|/Character classification> for an explanation of
586variants
587C<isASCII_uni>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
588C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
589library routine C<isascii()>. In these cases, the variants whose names contain
590C<LC> are the same as the corresponding ones without.
591
592Also note, that because all ASCII characters are UTF-8 invariant (meaning they
593have the exact same representation (always a single byte) whether encoded in
594UTF-8 or not), C<isASCII> will give the correct results when called with any
595byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
596work properly on any string encoded or not in UTF-8.
597
598=for apidoc Am|bool|isBLANK|char ch
599Returns a boolean indicating whether the specified character is a
600character considered to be a blank, analogous to C<m/[[:blank:]]/>.
601See the L<top of this section|/Character classification> for an explanation of
602variants
603C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uni>, C<isBLANK_utf8>, C<isBLANK_LC>,
604C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
605platforms do not have the C library routine C<isblank()>. In these cases, the
606variants whose names contain C<LC> are the same as the corresponding ones
607without.
608
609=for apidoc Am|bool|isCNTRL|char ch
610Returns a boolean indicating whether the specified character is a
611control character, analogous to C<m/[[:cntrl:]]/>.
612See the L<top of this section|/Character classification> for an explanation of
613variants
614C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uni>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
615C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
616On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
617
618=for apidoc Am|bool|isDIGIT|char ch
619Returns a boolean indicating whether the specified character is a
620digit, analogous to C<m/[[:digit:]]/>.
621Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
622See the L<top of this section|/Character classification> for an explanation of
623variants
624C<isDIGIT_uni>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
625C<isDIGIT_LC_utf8>.
626
627=for apidoc Am|bool|isGRAPH|char ch
628Returns a boolean indicating whether the specified character is a
629graphic character, analogous to C<m/[[:graph:]]/>.
630See the L<top of this section|/Character classification> for an explanation of
631variants
632C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uni>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
633C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
634
635=for apidoc Am|bool|isLOWER|char ch
636Returns a boolean indicating whether the specified character is a
637lowercase character, analogous to C<m/[[:lower:]]/>.
638See the L<top of this section|/Character classification> for an explanation of
639variants
640C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uni>, C<isLOWER_utf8>, C<isLOWER_LC>,
641C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
642
643=for apidoc Am|bool|isOCTAL|char ch
644Returns a boolean indicating whether the specified character is an
645octal digit, [0-7].
646The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
647C<isOCTAL>.
648
649=for apidoc Am|bool|isPUNCT|char ch
650Returns a boolean indicating whether the specified character is a
651punctuation character, analogous to C<m/[[:punct:]]/>.
652Note that the definition of what is punctuation isn't as
653straightforward as one might desire. See L<perlrecharclass/POSIX Character
654Classes> for details.
655See the L<top of this section|/Character classification> for an explanation of
656variants
657C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uni>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
658C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
659
660=for apidoc Am|bool|isSPACE|char ch
661Returns a boolean indicating whether the specified character is a
662whitespace character. This is analogous
663to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
664this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
665locale forms of this macro (the ones with C<LC> in their names) matched
666precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
667in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
668(See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
669See the L<top of this section|/Character classification> for an explanation of
670variants
671C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uni>, C<isSPACE_utf8>, C<isSPACE_LC>,
672C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
673
674=for apidoc Am|bool|isPSXSPC|char ch
675(short for Posix Space)
676Starting in 5.18, this is identical in all its forms to the
677corresponding C<isSPACE()> macros.
678The locale forms of this macro are identical to their corresponding
679C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
680non-locale forms differ from their C<isSPACE()> forms only in that the
681C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
682Otherwise they are identical. Thus this macro is analogous to what
683C<m/[[:space:]]/> matches in a regular expression.
684See the L<top of this section|/Character classification> for an explanation of
685variants
686C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uni>, C<isPSXSPC_utf8>, C<isPSXSPC_LC>,
687C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
688
689=for apidoc Am|bool|isUPPER|char ch
690Returns a boolean indicating whether the specified character is an
691uppercase character, analogous to C<m/[[:upper:]]/>.
692See the L<top of this section|/Character classification> for an explanation of
693variants
694C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uni>, C<isUPPER_utf8>, C<isUPPER_LC>,
695C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
696
697=for apidoc Am|bool|isPRINT|char ch
698Returns a boolean indicating whether the specified character is a
699printable character, analogous to C<m/[[:print:]]/>.
700See the L<top of this section|/Character classification> for an explanation of
701variants
702C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uni>, C<isPRINT_utf8>, C<isPRINT_LC>,
703C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
704
705=for apidoc Am|bool|isWORDCHAR|char ch
706Returns a boolean indicating whether the specified character is a character
707that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
708in a regular expression. A word character is an alphabetic character, a
709decimal digit, a connecting punctuation character (such as an underscore), or
710a "mark" character that attaches to one of those (like some sort of accent).
711C<isALNUM()> is a synonym provided for backward compatibility, even though a
712word character includes more than the standard C language meaning of
713alphanumeric.
714See the L<top of this section|/Character classification> for an explanation of
715variants
716C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uni>, and C<isWORDCHAR_utf8>.
717C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
718described there, but additionally include the platform's native underscore.
719
720=for apidoc Am|bool|isXDIGIT|char ch
721Returns a boolean indicating whether the specified character is a hexadecimal
722digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
723and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
724See the L<top of this section|/Character classification> for an explanation of
725variants
726C<isXDIGIT_uni>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
727C<isXDIGIT_LC_utf8>.
728
729=for apidoc Am|bool|isIDFIRST|char ch
730Returns a boolean indicating whether the specified character can be the first
731character of an identifier. This is very close to, but not quite the same as
732the official Unicode property C<XID_Start>. The difference is that this
733returns true only if the input character also matches L</isWORDCHAR>.
734See the L<top of this section|/Character classification> for an explanation of
735variants
736C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uni>, C<isIDFIRST_utf8>,
737C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
738
739=for apidoc Am|bool|isIDCONT|char ch
740Returns a boolean indicating whether the specified character can be the
741second or succeeding character of an identifier. This is very close to, but
742not quite the same as the official Unicode property C<XID_Continue>. The
743difference is that this returns true only if the input character also matches
744L</isWORDCHAR>. See the L<top of this section|/Character classification> for
745an
746explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uni>,
747C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
748C<isIDCONT_LC_utf8>.
749
750=head1 Miscellaneous Functions
751
752=for apidoc Am|U8|READ_XDIGIT|char str*
753Returns the value of an ASCII-range hex digit and advances the string pointer.
754Behaviour is only well defined when isXDIGIT(*str) is true.
755
756=head1 Character case changing
757
758=for apidoc Am|U8|toUPPER|U8 ch
759Converts the specified character to uppercase. If the input is anything but an
760ASCII lowercase character, that input character itself is returned. Variant
761C<toUPPER_A> is equivalent.
762
763=for apidoc Am|UV|toUPPER_uni|UV cp|U8* s|STRLEN* lenp
764Converts the Unicode code point C<cp> to its uppercase version, and
765stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
766that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
767bytes since the uppercase version may be longer than the original character.
768
769The first code point of the uppercased version is returned
770(but note, as explained just above, that there may be more.)
771
772=for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
773Converts the UTF-8 encoded character at C<p> to its uppercase version, and
774stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
775that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
776bytes since the uppercase version may be longer than the original character.
777
778The first code point of the uppercased version is returned
779(but note, as explained just above, that there may be more.)
780
781The input character at C<p> is assumed to be well-formed.
782
783=for apidoc Am|U8|toFOLD|U8 ch
784Converts the specified character to foldcase. If the input is anything but an
785ASCII uppercase character, that input character itself is returned. Variant
786C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
787Latin1 range, as the full generality of L</toFOLD_uni> is needed there.)
788
789=for apidoc Am|UV|toFOLD_uni|UV cp|U8* s|STRLEN* lenp
790Converts the Unicode code point C<cp> to its foldcase version, and
791stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
792that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
793bytes since the foldcase version may be longer than the original character.
794
795The first code point of the foldcased version is returned
796(but note, as explained just above, that there may be more.)
797
798=for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
799Converts the UTF-8 encoded character at C<p> to its foldcase version, and
800stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
801that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
802bytes since the foldcase version may be longer than the original character.
803
804The first code point of the foldcased version is returned
805(but note, as explained just above, that there may be more.)
806
807The input character at C<p> is assumed to be well-formed.
808
809=for apidoc Am|U8|toLOWER|U8 ch
810Converts the specified character to lowercase. If the input is anything but an
811ASCII uppercase character, that input character itself is returned. Variant
812C<toLOWER_A> is equivalent.
813
814=for apidoc Am|U8|toLOWER_L1|U8 ch
815Converts the specified Latin1 character to lowercase. The results are undefined if
816the input doesn't fit in a byte.
817
818=for apidoc Am|U8|toLOWER_LC|U8 ch
819Converts the specified character to lowercase using the current locale's rules,
820if possible; otherwise returns the input character itself.
821
822=for apidoc Am|UV|toLOWER_uni|UV cp|U8* s|STRLEN* lenp
823Converts the Unicode code point C<cp> to its lowercase version, and
824stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
825that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
826bytes since the lowercase version may be longer than the original character.
827
828The first code point of the lowercased version is returned
829(but note, as explained just above, that there may be more.)
830
831=for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
832Converts the UTF-8 encoded character at C<p> to its lowercase version, and
833stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
834that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
835bytes since the lowercase version may be longer than the original character.
836
837The first code point of the lowercased version is returned
838(but note, as explained just above, that there may be more.)
839
840The input character at C<p> is assumed to be well-formed.
841
842=for apidoc Am|U8|toTITLE|U8 ch
843Converts the specified character to titlecase. If the input is anything but an
844ASCII lowercase character, that input character itself is returned. Variant
845C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1 range,
846as the full generality of L</toTITLE_uni> is needed there. Titlecase is not a
847concept used in locale handling, so there is no functionality for that.)
848
849=for apidoc Am|UV|toTITLE_uni|UV cp|U8* s|STRLEN* lenp
850Converts the Unicode code point C<cp> to its titlecase version, and
851stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
852that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
853bytes since the titlecase version may be longer than the original character.
854
855The first code point of the titlecased version is returned
856(but note, as explained just above, that there may be more.)
857
858=for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
859Converts the UTF-8 encoded character at C<p> to its titlecase version, and
860stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
861that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
862bytes since the titlecase version may be longer than the original character.
863
864The first code point of the titlecased version is returned
865(but note, as explained just above, that there may be more.)
866
867The input character at C<p> is assumed to be well-formed.
868
869=cut
870
871XXX Still undocumented isVERTWS_uni and _utf8; it's unclear what their names
872really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
873
874Note that these macros are repeated in Devel::PPPort, so should also be
875patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
876
877*/
878
879/* Specify the widest unsigned type on the platform. Use U64TYPE because U64
880 * is known only in the perl core, and this macro can be called from outside
881 * that */
882#ifdef HAS_QUAD
883# define WIDEST_UTYPE U64TYPE
884#else
885# define WIDEST_UTYPE U32
886#endif
887
888/* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
889 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
890 * bits of information are lost even on a 64-bit machine, but to get the
891 * compiler to optimize it out if possible. This is because Configure makes
892 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
893 * sizeof() guarantees that this evaluates to a constant true at compile time.
894 */
895#define FITS_IN_8_BITS(c) ((sizeof(c) == 1) || !(((WIDEST_UTYPE)(c)) & ~0xFF))
896
897#ifdef EBCDIC
898# ifndef _ALL_SOURCE
899 /* The native libc isascii() et.al. functions return the wrong results
900 * on at least z/OS unless this is defined. */
901# error _ALL_SOURCE should probably be defined
902# endif
903#else
904 /* There is a simple definition of ASCII for ASCII platforms. But the
905 * EBCDIC one isn't so simple, so is defined using table look-up like the
906 * other macros below */
907# define isASCII(c) ((WIDEST_UTYPE)(c) < 128)
908#endif
909
910/* The lower 3 bits in both the ASCII and EBCDIC representations of '0' are 0,
911 * and the 8 possible permutations of those bits exactly comprise the 8 octal
912 * digits */
913#define isOCTAL_A(c) cBOOL(FITS_IN_8_BITS(c) && (0xF8 & (c)) == '0')
914
915#ifdef H_PERL /* If have access to perl.h, lookup in its table */
916
917/* Character class numbers. For internal core Perl use only. The ones less
918 * than 32 are used in PL_charclass[] and the ones up through the one that
919 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
920 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
921 * their actual definitions are here. If that file has a name not used here,
922 * it won't compile.
923 *
924 * The first group of these is ordered in what I (khw) estimate to be the
925 * frequency of their use. This gives a slight edge to exiting a loop earlier
926 * (in reginclass() in regexec.c) */
927# define _CC_WORDCHAR 0 /* \w and [:word:] */
928# define _CC_DIGIT 1 /* \d and [:digit:] */
929# define _CC_ALPHA 2 /* [:alpha:] */
930# define _CC_LOWER 3 /* [:lower:] */
931# define _CC_UPPER 4 /* [:upper:] */
932# define _CC_PUNCT 5 /* [:punct:] */
933# define _CC_PRINT 6 /* [:print:] */
934# define _CC_ALPHANUMERIC 7 /* [:alnum:] */
935# define _CC_GRAPH 8 /* [:graph:] */
936# define _CC_CASED 9 /* [:lower:] and [:upper:] under /i */
937
938#define _FIRST_NON_SWASH_CC 10
939/* The character classes above are implemented with swashes. The second group
940 * (just below) contains the ones implemented without. These are also sorted
941 * in rough order of the frequency of their use, except that \v should be last,
942 * as it isn't a real Posix character class, and some (small) inefficiencies in
943 * regular expression handling would be introduced by putting it in the middle
944 * of those that are. Also, cntrl and ascii come after the others as it may be
945 * useful to group these which have no members that match above Latin1, (or
946 * above ASCII in the latter case) */
947
948# define _CC_SPACE 10 /* \s, [:space:] */
949# define _CC_BLANK 11 /* [:blank:] */
950# define _CC_XDIGIT 12 /* [:xdigit:] */
951# define _CC_CNTRL 13 /* [:cntrl:] */
952# define _CC_ASCII 14 /* [:ascii:] */
953# define _CC_VERTSPACE 15 /* \v */
954
955# define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
956
957/* The members of the third group below do not need to be coordinated with data
958 * structures in regcomp.[ch] and regexec.c. */
959# define _CC_IDFIRST 16
960# define _CC_CHARNAME_CONT 17
961# define _CC_NONLATIN1_FOLD 18
962# define _CC_NONLATIN1_SIMPLE_FOLD 19
963# define _CC_QUOTEMETA 20
964# define _CC_NON_FINAL_FOLD 21
965# define _CC_IS_IN_SOME_FOLD 22
966# define _CC_MNEMONIC_CNTRL 23
967/* Unused: 24-31
968 * If more bits are needed, one could add a second word for non-64bit
969 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
970 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
971 * is used only for optimization (as of this writing), and differs in the
972 * Latin1 range from the ALPHA bit only in two relatively unimportant
973 * characters: the masculine and feminine ordinal indicators, so removing it
974 * would just cause /i regexes which match them to run less efficiently */
975
976#if defined(PERL_CORE) || defined(PERL_EXT)
977/* An enum version of the character class numbers, to help compilers
978 * optimize */
979typedef enum {
980 _CC_ENUM_ALPHA = _CC_ALPHA,
981 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
982 _CC_ENUM_ASCII = _CC_ASCII,
983 _CC_ENUM_BLANK = _CC_BLANK,
984 _CC_ENUM_CASED = _CC_CASED,
985 _CC_ENUM_CNTRL = _CC_CNTRL,
986 _CC_ENUM_DIGIT = _CC_DIGIT,
987 _CC_ENUM_GRAPH = _CC_GRAPH,
988 _CC_ENUM_LOWER = _CC_LOWER,
989 _CC_ENUM_PRINT = _CC_PRINT,
990 _CC_ENUM_PUNCT = _CC_PUNCT,
991 _CC_ENUM_SPACE = _CC_SPACE,
992 _CC_ENUM_UPPER = _CC_UPPER,
993 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
994 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
995 _CC_ENUM_XDIGIT = _CC_XDIGIT
996} _char_class_number;
997#endif
998
999#define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
1000#define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
1001
1002#if defined(PERL_IN_UTF8_C) || defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_REGEXEC_C)
1003# if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
1004 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1005 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
1006 #error Need to adjust order of swash_property_names[]
1007# endif
1008
1009/* This is declared static in each of the few files that this is #defined for
1010 * to keep them from being publicly accessible. Hence there is a small amount
1011 * of wasted space */
1012
1013static const char* const swash_property_names[] = {
1014 "XPosixWord",
1015 "XPosixDigit",
1016 "XPosixAlpha",
1017 "XPosixLower",
1018 "XPosixUpper",
1019 "XPosixPunct",
1020 "XPosixPrint",
1021 "XPosixAlnum",
1022 "XPosixGraph",
1023 "Cased"
1024};
1025#endif
1026
1027START_EXTERN_C
1028# ifdef DOINIT
1029EXTCONST U32 PL_charclass[] = {
1030# include "l1_char_class_tab.h"
1031};
1032
1033# else /* ! DOINIT */
1034EXTCONST U32 PL_charclass[];
1035# endif
1036END_EXTERN_C
1037
1038 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
1039# define _CC_mask(classnum) (1U << (classnum))
1040
1041 /* For internal core Perl use only: the base macro for defining macros like
1042 * isALPHA */
1043# define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
1044 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
1045
1046 /* The mask for the _A versions of the macros; it just adds in the bit for
1047 * ASCII. */
1048# define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1049
1050 /* For internal core Perl use only: the base macro for defining macros like
1051 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1052 * the ASCII bit are present */
1053# define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1054 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1055 == _CC_mask_A(classnum)))
1056
1057# define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1058# define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
1059# define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1060# define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
1061# define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
1062# define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1063# define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1064# define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1065# define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1066# define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1067# define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1068# define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
1069# define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits */
1070# define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
1071# define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1072# define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1073# define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1074
1075 /* continuation character for legal NAME in \N{NAME} */
1076# define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1077
1078# define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1079# define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1080# define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1081# define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1082# define isPSXSPC_L1(c) isSPACE_L1(c)
1083# define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1084# define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1085# define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1086# define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1087# define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
1088
1089# ifdef EBCDIC
1090# define isASCII(c) _generic_isCC(c, _CC_ASCII)
1091# endif
1092
1093 /* Participates in a single-character fold with a character above 255 */
1094# 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)))
1095
1096 /* Like the above, but also can be part of a multi-char fold */
1097# 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)))
1098
1099# define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
1100# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1101 _generic_isCC(c, _CC_NON_FINAL_FOLD)
1102# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1103 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1104# define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1105 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1106#else /* else we don't have perl.h H_PERL */
1107
1108 /* If we don't have perl.h, we are compiling a utility program. Below we
1109 * hard-code various macro definitions that wouldn't otherwise be available
1110 * to it. Most are coded based on first principals. First some ones common
1111 * to both ASCII and EBCDIC */
1112# define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1113# define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1114# define isSPACE_A(c) (isBLANK_A(c) \
1115 || (c) == '\n' \
1116 || (c) == '\r' \
1117 || (c) == '\v' \
1118 || (c) == '\f')
1119# ifdef EBCDIC /* There are gaps between 'i' and 'j'; 'r' and 's'. Same
1120 for uppercase. This is ordered to exclude most things
1121 early */
1122# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1123 && ((c) <= 'i' \
1124 || ((c) >= 'j' && (c) <= 'r') \
1125 || (c) >= 's'))
1126# define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1127 && ((c) <= 'I' \
1128 || ((c) >= 'J' && (c) <= 'R') \
1129 || (c) >= 'S'))
1130# else /* ASCII platform. */
1131# define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z')
1132# define isUPPER_A(c) ((c) <= 'Z' && (c) >= 'A')
1133# endif
1134
1135 /* Some more ASCII, non-ASCII common definitions */
1136# define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1137# define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
1138# define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
1139# define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1140# define isXDIGIT_A(c) (isDIGIT_A(c) \
1141 || ((c) >= 'a' && (c) <= 'f') \
1142 || ((c) <= 'F' && (c) >= 'A'))
1143
1144# ifdef EBCDIC
1145# define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1146 || (c) == '#' || (c) == '$' || (c) == '%' \
1147 || (c) == '&' || (c) == '\'' || (c) == '(' \
1148 || (c) == ')' || (c) == '*' || (c) == '+' \
1149 || (c) == ',' || (c) == '.' || (c) == '/' \
1150 || (c) == ':' || (c) == ';' || (c) == '<' \
1151 || (c) == '=' || (c) == '>' || (c) == '?' \
1152 || (c) == '@' || (c) == '[' || (c) == '\\' \
1153 || (c) == ']' || (c) == '^' || (c) == '_' \
1154 || (c) == '`' || (c) == '{' || (c) == '|' \
1155 || (c) == '}' || (c) == '~')
1156# define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1157# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1158
1159# ifdef QUESTION_MARK_CTRL
1160# define _isQMC(c) ((c) == QUESTION_MARK_CTRL)
1161# else
1162# define _isQMC(c) 0
1163# endif
1164
1165 /* I (khw) can't think of a way to define all the ASCII controls
1166 * without resorting to a libc (locale-sensitive) call. But we know
1167 * that all controls but the question-mark one are in the range 0-0x3f.
1168 * This makes sure that all the controls that have names are included,
1169 * and all controls that are also considered ASCII in the locale. This
1170 * may include more or fewer than what it actually should, but the
1171 * wrong ones are less-important controls, so likely won't impact
1172 * things (keep in mind that this is compiled only if perl.h isn't
1173 * available). The question mark control is included if available */
1174# define isCNTRL_A(c) (((c) < 0x40 && isascii(c)) \
1175 || (c) == '\0' || (c) == '\a' || (c) == '\b' \
1176 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1177 || (c) == '\t' || (c) == '\v' || _isQMC(c))
1178
1179# define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1180# else /* ASCII platform; things are simpler, and isASCII has already
1181 been defined */
1182# define isGRAPH_A(c) (((c) > ' ' && (c) < 127))
1183# define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1184# define isPUNCT_A(c) (isGRAPH_A(c) && (! isALPHANUMERIC_A(c)))
1185# define isCNTRL_A(c) (isASCII(c) && (! isPRINT_A(c)))
1186# endif
1187
1188 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
1189 * during debugging, and it seems best to keep them. We may be called
1190 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1191 * do anything anyway, so make it not a problem */
1192# if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1193# define NATIVE_TO_LATIN1(ch) (ch)
1194# endif
1195# define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1196# define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1197# define isBLANK_L1(c) (isBLANK_A(c) \
1198 || (FITS_IN_8_BITS(c) \
1199 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1200# define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1201# define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1202# define isLOWER_L1(c) (isLOWER_A(c) \
1203 || (FITS_IN_8_BITS(c) \
1204 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1205 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1206 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1207 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1208 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1209# define isPRINT_L1(c) (isPRINT_A(c) \
1210 || (FITS_IN_8_BITS(c) \
1211 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1212# define isPUNCT_L1(c) (isPUNCT_A(c) \
1213 || (FITS_IN_8_BITS(c) \
1214 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1215 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1216 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1217 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1218 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1219 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1220 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1221# define isSPACE_L1(c) (isSPACE_A(c) \
1222 || (FITS_IN_8_BITS(c) \
1223 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1224 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1225# define isUPPER_L1(c) (isUPPER_A(c) \
1226 || (FITS_IN_8_BITS(c) \
1227 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1228 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1229 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1230# define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1231# define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1232# define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1233 || isBLANK_L1(c) \
1234 || (c) == '-' \
1235 || (c) == '(' \
1236 || (c) == ')')
1237 /* The following are not fully accurate in the above-ASCII range. I (khw)
1238 * don't think it's necessary to be so for the purposes where this gets
1239 * compiled */
1240# define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1241# define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1242
1243 /* And these aren't accurate at all. They are useful only for above
1244 * Latin1, which utilities and bootstrapping don't deal with */
1245# define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
1246# define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1247# define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1248
1249 /* Many of the macros later in this file are defined in terms of these. By
1250 * implementing them with a function, which converts the class number into
1251 * a call to the desired macro, all of the later ones work. However, that
1252 * function won't be actually defined when building a utility program (no
1253 * perl.h), and so a compiler error will be generated if one is attempted
1254 * to be used. And the above-Latin1 code points require Unicode tables to
1255 * be present, something unlikely to be the case when bootstrapping */
1256# define _generic_isCC(c, classnum) \
1257 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1258# define _generic_isCC_A(c, classnum) \
1259 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
1260#endif /* End of no perl.h H_PERL */
1261
1262#define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1263#define isALPHA(c) isALPHA_A(c)
1264#define isASCII_A(c) isASCII(c)
1265#define isASCII_L1(c) isASCII(c)
1266#define isBLANK(c) isBLANK_A(c)
1267#define isCNTRL(c) isCNTRL_A(c)
1268#define isDIGIT(c) isDIGIT_A(c)
1269#define isGRAPH(c) isGRAPH_A(c)
1270#define isIDFIRST(c) isIDFIRST_A(c)
1271#define isLOWER(c) isLOWER_A(c)
1272#define isPRINT(c) isPRINT_A(c)
1273#define isPSXSPC_A(c) isSPACE_A(c)
1274#define isPSXSPC(c) isPSXSPC_A(c)
1275#define isPSXSPC_L1(c) isSPACE_L1(c)
1276#define isPUNCT(c) isPUNCT_A(c)
1277#define isSPACE(c) isSPACE_A(c)
1278#define isUPPER(c) isUPPER_A(c)
1279#define isWORDCHAR(c) isWORDCHAR_A(c)
1280#define isXDIGIT(c) isXDIGIT_A(c)
1281
1282/* ASCII casing. These could also be written as
1283 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1284 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1285 which uses table lookup and mask instead of subtraction. (This would
1286 work because the _MOD does not apply in the ASCII range) */
1287#define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1288#define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
1289
1290/* In the ASCII range, these are equivalent to what they're here defined to be.
1291 * But by creating these definitions, other code doesn't have to be aware of
1292 * this detail */
1293#define toFOLD(c) toLOWER(c)
1294#define toTITLE(c) toUPPER(c)
1295
1296#define toLOWER_A(c) toLOWER(c)
1297#define toUPPER_A(c) toUPPER(c)
1298#define toFOLD_A(c) toFOLD(c)
1299#define toTITLE_A(c) toTITLE(c)
1300
1301/* Use table lookup for speed; returns the input itself if is out-of-range */
1302#define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
1303 ? (c) \
1304 : PL_latin1_lc[ (U8) (c) ])
1305#define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1306
1307/* Modified uc. Is correct uc except for three non-ascii chars which are
1308 * all mapped to one of them, and these need special handling; returns the
1309 * input itself if is out-of-range */
1310#define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
1311 ? (c) \
1312 : PL_mod_latin1_uc[ (U8) (c) ])
1313#define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
1314
1315/* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1316
1317/* For internal core Perl use only: the base macro for defining macros like
1318 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
1319 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1320 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1321 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1322 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1323 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1324 * code points under locale rules, use the macros based on _generic_LC_uvchr
1325 * instead of this. */
1326#define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1327 (! FITS_IN_8_BITS(c) \
1328 ? 0 \
1329 : IN_UTF8_CTYPE_LOCALE \
1330 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
1331 : cBOOL(non_utf8))
1332
1333/* For internal core Perl use only: a helper macro for defining macros like
1334 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1335 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1336 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1337 * _generic_LC_base for more info */
1338#define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1339 _generic_LC_base(c,utf8_locale_classnum, \
1340 non_utf8_func( (_LC_CAST) (c)))
1341
1342/* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1343 * 'c' is the platform's native underscore character */
1344#define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1345 _generic_LC_base(c, utf8_locale_classnum, \
1346 (non_utf8_func( (_LC_CAST) (c)) \
1347 || (char)(c) == '_'))
1348
1349/* These next three are also for internal core Perl use only: case-change
1350 * helper macros */
1351#define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1352 ? (c) \
1353 : (IN_UTF8_CTYPE_LOCALE) \
1354 ? PL_latin1_lc[ (U8) (c) ] \
1355 : (cast)function((cast)(c)))
1356
1357/* Note that the result can be larger than a byte in a UTF-8 locale. It
1358 * returns a single value, so can't adequately return the upper case of LATIN
1359 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1360 * values "SS"); instead it asserts against that under DEBUGGING, and
1361 * otherwise returns its input */
1362#define _generic_toUPPER_LC(c, function, cast) \
1363 (! FITS_IN_8_BITS(c) \
1364 ? (c) \
1365 : ((! IN_UTF8_CTYPE_LOCALE) \
1366 ? (cast)function((cast)(c)) \
1367 : ((((U8)(c)) == MICRO_SIGN) \
1368 ? GREEK_CAPITAL_LETTER_MU \
1369 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1370 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1371 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1372 ? (__ASSERT_(0) (c)) \
1373 : PL_mod_latin1_uc[ (U8) (c) ])))))
1374
1375/* Note that the result can be larger than a byte in a UTF-8 locale. It
1376 * returns a single value, so can't adequately return the fold case of LATIN
1377 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1378 * values "ss"); instead it asserts against that under DEBUGGING, and
1379 * otherwise returns its input */
1380#define _generic_toFOLD_LC(c, function, cast) \
1381 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
1382 ? GREEK_SMALL_LETTER_MU \
1383 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1384 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1385 _generic_toLOWER_LC(c, function, cast)))
1386
1387/* Use the libc versions for these if available. */
1388#if defined(HAS_ISASCII)
1389# define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1390#else
1391# define isASCII_LC(c) isASCII(c)
1392#endif
1393
1394#if defined(HAS_ISBLANK)
1395# define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
1396#else /* Unlike isASCII, varies if in a UTF-8 locale */
1397# define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
1398#endif
1399
1400#define _LC_CAST U8
1401
1402#ifdef WIN32
1403 /* The Windows functions don't bother to follow the POSIX standard, which
1404 * for example says that something can't both be a printable and a control.
1405 * But Windows treats the \t control as a printable, and does such things
1406 * as making superscripts into both digits and punctuation. This tames
1407 * these flaws by assuming that the definitions of both controls and space
1408 * are correct, and then making sure that other definitions don't have
1409 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1410 * Not all possible weirdnesses are checked for, just the ones that were
1411 * detected on actual Microsoft code pages */
1412
1413# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1414# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1415
1416# define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) && isALPHANUMERIC_LC(c))
1417# define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && ! isPUNCT_LC(c))
1418# define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && isALPHANUMERIC_LC(c))
1419# define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1420# define isIDFIRST_LC(c) (((c) == '_') || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1421# define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1422# define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1423# define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1424# define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1425# define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1426# define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) && isALPHANUMERIC_LC(c))
1427
1428# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1429# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1430# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1431
1432#elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
1433 /* For most other platforms */
1434
1435# define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1436# define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1437# define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1438# define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1439# define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1440# define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1441# define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1442# define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1443# define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1444# define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1445# define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1446# define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1447# define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1448
1449
1450# define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1451# define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1452# define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1453
1454#else /* The final fallback position */
1455
1456# define isALPHA_LC(c) (isascii(c) && isalpha(c))
1457# define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1458# define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1459# define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1460# define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1461# define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1462# define isLOWER_LC(c) (isascii(c) && islower(c))
1463# define isPRINT_LC(c) (isascii(c) && isprint(c))
1464# define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1465# define isSPACE_LC(c) (isascii(c) && isspace(c))
1466# define isUPPER_LC(c) (isascii(c) && isupper(c))
1467# define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1468# define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1469
1470# define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1471# define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1472# define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
1473
1474#endif
1475
1476#define isIDCONT(c) isWORDCHAR(c)
1477#define isIDCONT_A(c) isWORDCHAR_A(c)
1478#define isIDCONT_L1(c) isWORDCHAR_L1(c)
1479#define isIDCONT_LC(c) isWORDCHAR_LC(c)
1480#define isPSXSPC_LC(c) isSPACE_LC(c)
1481
1482/* For internal core Perl use only: the base macros for defining macros like
1483 * isALPHA_uni. 'c' is the code point to check. 'classnum' is the POSIX class
1484 * number defined earlier in this file. _generic_uni() is used for POSIX
1485 * classes where there is a macro or function 'above_latin1' that takes the
1486 * single argument 'c' and returns the desired value. These exist for those
1487 * classes which have simple definitions, avoiding the overhead of a hash
1488 * lookup or inversion list binary search. _generic_swash_uni() can be used
1489 * for classes where that overhead is faster than a direct lookup.
1490 * _generic_uni() won't compile if 'c' isn't unsigned, as it won't match the
1491 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1492 * have duplicate checks here, so could create versions of the macros that
1493 * don't, but experiments show that gcc optimizes them out anyway. */
1494
1495/* Note that all ignore 'use bytes' */
1496#define _generic_uni(classnum, above_latin1, c) ((c) < 256 \
1497 ? _generic_isCC(c, classnum) \
1498 : above_latin1(c))
1499#define _generic_swash_uni(classnum, c) ((c) < 256 \
1500 ? _generic_isCC(c, classnum) \
1501 : _is_uni_FOO(classnum, c))
1502#define isALPHA_uni(c) _generic_swash_uni(_CC_ALPHA, c)
1503#define isALPHANUMERIC_uni(c) _generic_swash_uni(_CC_ALPHANUMERIC, c)
1504#define isASCII_uni(c) isASCII(c)
1505#define isBLANK_uni(c) _generic_uni(_CC_BLANK, is_HORIZWS_cp_high, c)
1506#define isCNTRL_uni(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1507#define isDIGIT_uni(c) _generic_swash_uni(_CC_DIGIT, c)
1508#define isGRAPH_uni(c) _generic_swash_uni(_CC_GRAPH, c)
1509#define isIDCONT_uni(c) _generic_uni(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1510#define isIDFIRST_uni(c) _generic_uni(_CC_IDFIRST, _is_uni_perl_idstart, c)
1511#define isLOWER_uni(c) _generic_swash_uni(_CC_LOWER, c)
1512#define isPRINT_uni(c) _generic_swash_uni(_CC_PRINT, c)
1513
1514#define isPUNCT_uni(c) _generic_swash_uni(_CC_PUNCT, c)
1515#define isSPACE_uni(c) _generic_uni(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1516#define isPSXSPC_uni(c) isSPACE_uni(c)
1517
1518#define isUPPER_uni(c) _generic_swash_uni(_CC_UPPER, c)
1519#define isVERTWS_uni(c) _generic_uni(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1520#define isWORDCHAR_uni(c) _generic_swash_uni(_CC_WORDCHAR, c)
1521#define isXDIGIT_uni(c) _generic_uni(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1522
1523#define toFOLD_uni(c,s,l) to_uni_fold(c,s,l)
1524#define toLOWER_uni(c,s,l) to_uni_lower(c,s,l)
1525#define toTITLE_uni(c,s,l) to_uni_title(c,s,l)
1526#define toUPPER_uni(c,s,l) to_uni_upper(c,s,l)
1527
1528/* For internal core Perl use only: the base macros for defining macros like
1529 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1530 * point, not just 0-255. Like _generic_uni, there are two versions, one for
1531 * simple class definitions; the other for more complex. These are like
1532 * _generic_uni, so see it for more info. */
1533#define _generic_LC_uvchr(latin1, above_latin1, c) \
1534 (c < 256 ? latin1(c) : above_latin1(c))
1535#define _generic_LC_swash_uvchr(latin1, classnum, c) \
1536 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1537
1538#define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1539#define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1540 _CC_ALPHANUMERIC, c)
1541#define isASCII_LC_uvchr(c) isASCII_LC(c)
1542#define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, is_HORIZWS_cp_high, c)
1543#define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
1544#define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1545#define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
1546#define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
1547 _is_uni_perl_idcont, c)
1548#define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
1549 _is_uni_perl_idstart, c)
1550#define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1551#define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
1552#define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
1553#define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
1554#define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
1555 is_XPERLSPACE_cp_high, c)
1556#define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
1557#define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
1558 _CC_WORDCHAR, c)
1559#define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, is_XDIGIT_cp_high, c)
1560
1561#define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
1562
1563/* For internal core Perl use only: the base macros for defining macros like
1564 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1565 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1566 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1567 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1568 * same representation whether utf8 or not. Note that it assumes that the utf8
1569 * has been validated, and ignores 'use bytes' */
1570#define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
1571 ? _generic_isCC(*(p), classnum) \
1572 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1573 ? _generic_isCC( \
1574 TWO_BYTE_UTF8_TO_NATIVE(*(p), \
1575 *((p)+1 )), \
1576 classnum) \
1577 : utf8)
1578/* Like the above, but calls 'above_latin1(p)' to get the utf8 value. 'above_latin1'
1579 * can be a macro */
1580#define _generic_func_utf8(classnum, above_latin1, p) \
1581 _generic_utf8(classnum, p, above_latin1(p))
1582/* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
1583 * 'above_latin1' parameter */
1584#define _generic_swash_utf8(classnum, p) \
1585 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
1586
1587/* Like the above, but should be used only when it is known that there are no
1588 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1589 * class is TRUE for. Hence it can skip the tests for this range.
1590 * 'above_latin1' should include its arguments */
1591#define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
1592 (UTF8_IS_INVARIANT(*(p)) \
1593 ? _generic_isCC(*(p), classnum) \
1594 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
1595 ? above_latin1 \
1596 : 0)
1597
1598/* NOTE that some of these macros have very similar ones in regcharclass.h.
1599 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1600 * there, differing in name only by an underscore from the one here
1601 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1602 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1603 * points; the regcharclass.h ones are implemented as a series of
1604 * "if-else-if-else ..." */
1605
1606#define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1607#define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1608#define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
1609 utf8, the non-utf8 macro works
1610 */
1611#define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
1612
1613#ifdef EBCDIC
1614 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1615 * more efficient macro instead of the more general one */
1616# define isCNTRL_utf8(p) isCNTRL_L1(*(p))
1617#else
1618# define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1619#endif
1620
1621#define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1622 _is_utf8_FOO(_CC_DIGIT, p))
1623#define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
1624#define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
1625 _is_utf8_perl_idcont, p)
1626
1627/* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1628 * IDFIRST is an alnum. See
1629 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
1630 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1631 * This used to be not the XID version, but we decided to go with the more
1632 * modern Unicode definition */
1633#define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
1634 _is_utf8_perl_idstart, p)
1635
1636#define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1637#define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1638#define isPSXSPC_utf8(p) isSPACE_utf8(p)
1639#define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1640#define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1641#define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1642#define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1643#define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1644#define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
1645 is_XDIGIT_high(p))
1646
1647#define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
1648#define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1649#define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1650#define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
1651
1652/* For internal core Perl use only: the base macros for defining macros like
1653 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1654 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1655 * 'macro' parameter */
1656#define _generic_LC_utf8(macro, p, utf8) \
1657 (UTF8_IS_INVARIANT(*(p)) \
1658 ? macro(*(p)) \
1659 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1660 ? macro(TWO_BYTE_UTF8_TO_NATIVE(*(p), *((p)+1))) \
1661 : utf8)
1662
1663#define _generic_LC_swash_utf8(macro, classnum, p) \
1664 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1665#define _generic_LC_func_utf8(macro, above_latin1, p) \
1666 _generic_LC_utf8(macro, p, above_latin1(p))
1667
1668#define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
1669 _CC_ALPHANUMERIC, p)
1670#define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1671#define isASCII_LC_utf8(p) isASCII_LC(*p)
1672#define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, is_HORIZWS_high, p)
1673#define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1674#define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1675#define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1676#define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, _is_utf8_perl_idcont, p)
1677#define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, _is_utf8_perl_idstart, p)
1678#define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1679#define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1680#define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1681#define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1682#define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, is_XPERLSPACE_high, p)
1683#define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
1684#define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1685 _CC_WORDCHAR, p)
1686#define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, is_XDIGIT_high, p)
1687
1688/* Macros for backwards compatibility and for completeness when the ASCII and
1689 * Latin1 values are identical */
1690#define isALPHAU(c) isALPHA_L1(c)
1691#define isDIGIT_L1(c) isDIGIT_A(c)
1692#define isOCTAL(c) isOCTAL_A(c)
1693#define isOCTAL_L1(c) isOCTAL_A(c)
1694#define isXDIGIT_L1(c) isXDIGIT_A(c)
1695#define isALNUM(c) isWORDCHAR(c)
1696#define isALNUMU(c) isWORDCHAR_L1(c)
1697#define isALNUM_LC(c) isWORDCHAR_LC(c)
1698#define isALNUM_uni(c) isWORDCHAR_uni(c)
1699#define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
1700#define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1701#define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1702#define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1703#define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1704#define isALNUMC(c) isALPHANUMERIC(c)
1705#define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1706#define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
1707#define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
1708#define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
1709#define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
1710
1711/* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1712 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1713 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1714 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1715 * the outlier from the block that contains the other controls, just like
1716 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1717 * block. If it weren't special cased, it would yield a non-control.
1718 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1719 * etc. */
1720#ifndef EBCDIC
1721# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
1722#else
1723# define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1724 ((isPRINT_A(c)) \
1725 ? (UNLIKELY((c) == '?') \
1726 ? QUESTION_MARK_CTRL \
1727 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1728 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1729 ? '?' \
1730 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
1731#endif
1732
1733/* Line numbers are unsigned, 32 bits. */
1734typedef U32 line_t;
1735#define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
1736
1737/* Helpful alias for version prescan */
1738#define is_LAX_VERSION(a,b) \
1739 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1740
1741#define is_STRICT_VERSION(a,b) \
1742 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1743
1744#define BADVERSION(a,b,c) \
1745 if (b) { \
1746 *b = c; \
1747 } \
1748 return a;
1749
1750/* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1751 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1752 * which is advanced. The input is validated only by an assert() in DEBUGGING
1753 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1754 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1755#define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1756 ? (c) \
1757 : ((c) + 9))))
1758#define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1759 ? (*(s)++) \
1760 : (*(s)++ + 9))))
1761
1762/* Converts a character known to represent an octal digit (0-7) to its numeric
1763 * value. The input is validated only by an assert() in DEBUGGING builds. In
1764 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1765#define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1766
1767/* Efficiently returns a boolean as to if two native characters are equivalent
1768 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1769 * the ALPHA in the name is to remind you of that. This is asserted() in
1770 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1771 * works (on valid input) for both non- and UTF-8-encoded bytes.
1772 *
1773 * When one of the inputs is a compile-time constant and gets folded by the
1774 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1775 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1776 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
1777 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1778 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1779 * just a single 0, in the bit position where the upper- and lowercase differ.
1780 * */
1781#define isALPHA_FOLD_EQ(c1, c2) \
1782 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1783 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1784#define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1785
1786/*
1787=head1 Memory Management
1788
1789=for apidoc Am|void|Newx|void* ptr|int nitems|type
1790The XSUB-writer's interface to the C C<malloc> function.
1791
1792Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1793
1794In 5.9.3, Newx() and friends replace the older New() API, and drops
1795the first parameter, I<x>, a debug aid which allowed callers to identify
1796themselves. This aid has been superseded by a new build option,
1797PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
1798there for use in XS modules supporting older perls.
1799
1800=for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
1801The XSUB-writer's interface to the C C<malloc> function, with
1802cast. See also C<Newx>.
1803
1804Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1805
1806=for apidoc Am|void|Newxz|void* ptr|int nitems|type
1807The XSUB-writer's interface to the C C<malloc> function. The allocated
1808memory is zeroed with C<memzero>. See also C<Newx>.
1809
1810Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1811
1812=for apidoc Am|void|Renew|void* ptr|int nitems|type
1813The XSUB-writer's interface to the C C<realloc> function.
1814
1815Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1816
1817=for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1818The XSUB-writer's interface to the C C<realloc> function, with
1819cast.
1820
1821Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1822
1823=for apidoc Am|void|Safefree|void* ptr
1824The XSUB-writer's interface to the C C<free> function.
1825
1826This should B<ONLY> be used on memory obtained using L<"Newx"> and friends.
1827
1828=for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1829The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
1830source, C<dest> is the destination, C<nitems> is the number of items, and
1831C<type> is the type. Can do overlapping moves. See also C<Copy>.
1832
1833=for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
1834Like C<Move> but returns dest. Useful
1835for encouraging compilers to tail-call
1836optimise.
1837
1838=for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1839The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
1840source, C<dest> is the destination, C<nitems> is the number of items, and
1841C<type> is the type. May fail on overlapping copies. See also C<Move>.
1842
1843=for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1844
1845Like C<Copy> but returns dest. Useful
1846for encouraging compilers to tail-call
1847optimise.
1848
1849=for apidoc Am|void|Zero|void* dest|int nitems|type
1850
1851The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1852destination, C<nitems> is the number of items, and C<type> is the type.
1853
1854=for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1855
1856Like C<Zero> but returns dest. Useful
1857for encouraging compilers to tail-call
1858optimise.
1859
1860=for apidoc Am|void|StructCopy|type *src|type *dest|type
1861This is an architecture-independent macro to copy one structure to another.
1862
1863=for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1864
1865Fill up memory with a byte pattern (a byte repeated over and over
1866again) that hopefully catches attempts to access uninitialized memory.
1867
1868=for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1869
1870PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1871
1872=for apidoc Am|void|PoisonFree|void* dest|int nitems|type
1873
1874PoisonWith(0xEF) for catching access to freed memory.
1875
1876=for apidoc Am|void|Poison|void* dest|int nitems|type
1877
1878PoisonWith(0xEF) for catching access to freed memory.
1879
1880=cut */
1881
1882/* Maintained for backwards-compatibility only. Use newSV() instead. */
1883#ifndef PERL_CORE
1884#define NEWSV(x,len) newSV(len)
1885#endif
1886
1887#define MEM_SIZE_MAX ((MEM_SIZE)~0)
1888
1889
1890#ifdef PERL_MALLOC_WRAP
1891
1892/* This expression will be constant-folded at compile time. It checks
1893 * whether or not the type of the count n is so small (e.g. U8 or U16, or
1894 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
1895 * As well as avoiding the need for a run-time check in some cases, it's
1896 * designed to avoid compiler warnings like:
1897 * comparison is always false due to limited range of data type
1898 */
1899
1900# define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
1901 (sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
1902
1903/* This is written in a slightly odd way to avoid various spurious
1904 * compiler warnings. We *want* to write the expression as
1905 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
1906 * (for some compile-time constant C), but even when the LHS
1907 * constant-folds to false at compile-time, g++ insists on emitting
1908 * warnings about the RHS (e.g. "comparison is always false"), so instead
1909 * we write it as
1910 *
1911 * (cond ? n : X) > C
1912 *
1913 * where X is a constant with X > C always false. Choosing a value for X
1914 * is tricky. If 0, some compilers will complain about 0 > C always being
1915 * false; if 1, Coverity complains when n happens to be the constant value
1916 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
1917 * for X and hope that nothing else whines.
1918 */
1919
1920# define _MEM_WRAP_WILL_WRAP(n,t) \
1921 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
1922 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
1923
1924# define MEM_WRAP_CHECK(n,t) \
1925 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1926 && (croak_memory_wrap(),0))
1927
1928# define MEM_WRAP_CHECK_1(n,t,a) \
1929 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1930 && (Perl_croak_nocontext("%s",(a)),0))
1931
1932#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
1933
1934#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)))
1935#else
1936
1937#define MEM_WRAP_CHECK(n,t)
1938#define MEM_WRAP_CHECK_1(n,t,a)
1939#define MEM_WRAP_CHECK_2(n,t,a,b)
1940#define MEM_WRAP_CHECK_(n,t)
1941
1942#define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
1943
1944#endif
1945
1946#ifdef PERL_MEM_LOG
1947/*
1948 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
1949 * go through functions, which are handy for debugging breakpoints, but
1950 * which more importantly get the immediate calling environment (file and
1951 * line number, and C function name if available) passed in. This info can
1952 * then be used for logging the calls, for which one gets a sample
1953 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
1954 *
1955 * Known problems:
1956 * - not all memory allocs get logged, only those
1957 * that go through Newx() and derivatives (while all
1958 * Safefrees do get logged)
1959 * - __FILE__ and __LINE__ do not work everywhere
1960 * - __func__ or __FUNCTION__ even less so
1961 * - I think more goes on after the perlio frees but
1962 * the thing is that STDERR gets closed (as do all
1963 * the file descriptors)
1964 * - no deeper calling stack than the caller of the Newx()
1965 * or the kind, but do I look like a C reflection/introspection
1966 * utility to you?
1967 * - the function prototypes for the logging functions
1968 * probably should maybe be somewhere else than handy.h
1969 * - one could consider inlining (macrofying) the logging
1970 * for speed, but I am too lazy
1971 * - one could imagine recording the allocations in a hash,
1972 * (keyed by the allocation address?), and maintain that
1973 * through reallocs and frees, but how to do that without
1974 * any News() happening...?
1975 * - lots of -Ddefines to get useful/controllable output
1976 * - lots of ENV reads
1977 */
1978
1979PERL_CALLCONV Malloc_t Perl_mem_log_alloc(const UV n, const UV typesize, const char *type_name, Malloc_t newalloc, const char *filename, const int linenumber, const char *funcname);
1980
1981PERL_CALLCONV Malloc_t Perl_mem_log_realloc(const UV n, const UV typesize, const char *type_name, Malloc_t oldalloc, Malloc_t newalloc, const char *filename, const int linenumber, const char *funcname);
1982
1983PERL_CALLCONV Malloc_t Perl_mem_log_free(Malloc_t oldalloc, const char *filename, const int linenumber, const char *funcname);
1984
1985# ifdef PERL_CORE
1986# ifndef PERL_MEM_LOG_NOIMPL
1987enum mem_log_type {
1988 MLT_ALLOC,
1989 MLT_REALLOC,
1990 MLT_FREE,
1991 MLT_NEW_SV,
1992 MLT_DEL_SV
1993};
1994# endif
1995# if defined(PERL_IN_SV_C) /* those are only used in sv.c */
1996void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
1997void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
1998# endif
1999# endif
2000
2001#endif
2002
2003#ifdef PERL_MEM_LOG
2004#define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2005#define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
2006#define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
2007#endif
2008
2009#ifndef MEM_LOG_ALLOC
2010#define MEM_LOG_ALLOC(n,t,a) (a)
2011#endif
2012#ifndef MEM_LOG_REALLOC
2013#define MEM_LOG_REALLOC(n,t,v,a) (a)
2014#endif
2015#ifndef MEM_LOG_FREE
2016#define MEM_LOG_FREE(a) (a)
2017#endif
2018
2019#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2020#define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2021#define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
2022
2023#ifndef PERL_CORE
2024/* pre 5.9.x compatibility */
2025#define New(x,v,n,t) Newx(v,n,t)
2026#define Newc(x,v,n,t,c) Newxc(v,n,t,c)
2027#define Newz(x,v,n,t) Newxz(v,n,t)
2028#endif
2029
2030#define Renew(v,n,t) \
2031 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2032#define Renewc(v,n,t,c) \
2033 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2034
2035#ifdef PERL_POISON
2036#define Safefree(d) \
2037 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
2038#else
2039#define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
2040#endif
2041
2042#define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2043#define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2044#define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
2045
2046#define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2047#define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2048#ifdef HAS_MEMSET
2049#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
2050#else
2051/* Using bzero(), which returns void. */
2052#define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
2053#endif
2054
2055#define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2056#define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2057#define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2058#define Poison(d,n,t) PoisonFree(d,n,t)
2059
2060#ifdef PERL_POISON
2061# define PERL_POISON_EXPR(x) x
2062#else
2063# define PERL_POISON_EXPR(x)
2064#endif
2065
2066#ifdef USE_STRUCT_COPY
2067#define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2068#else
2069#define StructCopy(s,d,t) Copy(s,d,1,t)
2070#endif
2071
2072/* C_ARRAY_LENGTH is the number of elements in the C array (so you
2073 * want your zero-based indices to be less than but not equal to).
2074 *
2075 * C_ARRAY_END is one past the last: half-open/half-closed range,
2076 * not last-inclusive range. */
2077#define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
2078#define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
2079
2080#ifdef NEED_VA_COPY
2081# ifdef va_copy
2082# define Perl_va_copy(s, d) va_copy(d, s)
2083# else
2084# if defined(__va_copy)
2085# define Perl_va_copy(s, d) __va_copy(d, s)
2086# else
2087# define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2088# endif
2089# endif
2090#endif
2091
2092/* convenience debug macros */
2093#ifdef USE_ITHREADS
2094#define pTHX_FORMAT "Perl interpreter: 0x%p"
2095#define pTHX__FORMAT ", Perl interpreter: 0x%p"
2096#define pTHX_VALUE_ (void *)my_perl,
2097#define pTHX_VALUE (void *)my_perl
2098#define pTHX__VALUE_ ,(void *)my_perl,
2099#define pTHX__VALUE ,(void *)my_perl
2100#else
2101#define pTHX_FORMAT
2102#define pTHX__FORMAT
2103#define pTHX_VALUE_
2104#define pTHX_VALUE
2105#define pTHX__VALUE_
2106#define pTHX__VALUE
2107#endif /* USE_ITHREADS */
2108
2109/* Perl_deprecate was not part of the public API, and did not have a deprecate()
2110 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2111 CPAN::Unpack show any users outside the core. */
2112#ifdef PERL_CORE
2113# define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), "Use of " s " is deprecated")
2114#endif
2115
2116/* Internal macros to deal with gids and uids */
2117#ifdef PERL_CORE
2118
2119# if Uid_t_size > IVSIZE
2120# define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2121# define SvUID(sv) SvNV(sv)
2122# else
2123# if Uid_t_sign <= 0
2124# define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2125# define SvUID(sv) SvIV(sv)
2126# else
2127# define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2128# define SvUID(sv) SvUV(sv)
2129# endif
2130# endif /* Uid_t_size */
2131
2132# if Gid_t_size > IVSIZE
2133# define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2134# define SvGID(sv) SvNV(sv)
2135# else
2136# if Gid_t_sign <= 0
2137# define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2138# define SvGID(sv) SvIV(sv)
2139# else
2140# define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2141# define SvGID(sv) SvUV(sv)
2142# endif
2143# endif /* Gid_t_size */
2144
2145#endif
2146
2147#endif /* HANDY_H */
2148
2149/*
2150 * ex: set ts=8 sts=4 sw=4 et:
2151 */
Unnamed repository; edit this file 'description' to name the repository.