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
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.
11 /* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
14 #ifndef HANDY_H /* Guard against nested #inclusion */
17 #if !defined(__STDC__)
25 # define Null(type) ((type)NULL)
30 =for apidoc AmU||Nullch
31 Null character pointer. (No longer available when C<PERL_CORE> is
34 =for apidoc AmU||Nullsv
35 Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
40 # define Nullch Null(char*)
41 # define Nullfp Null(PerlIO*)
42 # define Nullsv Null(SV*)
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;
59 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
60 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
63 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
64 # define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
66 # define MUTABLE_PTR(p) ((void *) (p))
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))
76 #if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
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
91 #ifdef __GNUG__ /* GNU g++ has bool built-in */
92 # ifndef PERL_BOOL_AS_CHAR
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
111 #define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
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__
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__ ""
124 # define FUNCTION__ __FUNCTION__ /* Common extension. */
128 /* XXX A note on the perl source internal type system. The
129 original intent was that I32 be *exactly* 32 bits.
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
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.
142 Similarly, there is no guarantee that I16 and U16 have exactly 16
145 For dealing with issues that may arise from various 32/64-bit
146 systems, we will ask Configure to check out
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).
158 #ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
159 # include <inttypes.h>
160 # ifdef INT32_MIN_BROKEN
162 # define INT32_MIN (-2147483647-1)
164 # ifdef INT64_MIN_BROKEN
166 # define INT64_MIN (-9223372036854775807LL-1)
182 /* INT64_C/UINT64_C are C99 from <stdint.h> (so they will not be
183 * available in strict C89 mode), but they are nice, so let's define
184 * them if necessary. */
185 #if defined(HAS_QUAD)
187 # undef PeRl_UINT64_C
188 /* Prefer the native integer types (int and long) over long long
189 * (which is not C89) and Win32-specific __int64. */
190 # if QUADKIND == QUAD_IS_INT && INTSIZE == 8
191 # define PeRl_INT64_C(c) (c)
192 # define PeRl_UINT64_C(c) CAT2(c,U)
194 # if QUADKIND == QUAD_IS_LONG && LONGSIZE == 8
195 # define PeRl_INT64_C(c) CAT2(c,L)
196 # define PeRl_UINT64_C(c) CAT2(c,UL)
198 # if QUADKIND == QUAD_IS_LONG_LONG && defined(HAS_LONG_LONG)
199 # define PeRl_INT64_C(c) CAT2(c,LL)
200 # define PeRl_UINT64_C(c) CAT2(c,ULL)
202 # if QUADKIND == QUAD_IS___INT64
203 # define PeRl_INT64_C(c) CAT2(c,I64)
204 # define PeRl_UINT64_C(c) CAT2(c,UI64)
206 # ifndef PeRl_INT64_C
207 # define PeRl_INT64_C(c) ((I64)(c)) /* last resort */
208 # define PeRl_UINT64_C(c) ((U64)(c))
210 /* In OS X the INT64_C/UINT64_C are defined with LL/ULL, which will
211 * not fly with C89-pedantic gcc, so let's undefine them first so that
212 * we can redefine them with our native integer preferring versions. */
213 # if defined(PERL_DARWIN) && defined(PERL_GCC_PEDANTIC)
218 # define INT64_C(c) PeRl_INT64_C(c)
221 # define UINT64_C(c) PeRl_UINT64_C(c)
225 #if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
227 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
228 Please search CHAR_MAX in perl.h for further details. */
229 #define U8_MAX UINT8_MAX
230 #define U8_MIN UINT8_MIN
232 #define I16_MAX INT16_MAX
233 #define I16_MIN INT16_MIN
234 #define U16_MAX UINT16_MAX
235 #define U16_MIN UINT16_MIN
237 #define I32_MAX INT32_MAX
238 #define I32_MIN INT32_MIN
239 #ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
240 # define U32_MAX UINT32_MAX
242 # define U32_MAX 4294967295U
244 #define U32_MIN UINT32_MIN
248 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
249 Please search CHAR_MAX in perl.h for further details. */
250 #define U8_MAX PERL_UCHAR_MAX
251 #define U8_MIN PERL_UCHAR_MIN
253 #define I16_MAX PERL_SHORT_MAX
254 #define I16_MIN PERL_SHORT_MIN
255 #define U16_MAX PERL_USHORT_MAX
256 #define U16_MIN PERL_USHORT_MIN
259 # define I32_MAX PERL_INT_MAX
260 # define I32_MIN PERL_INT_MIN
261 # define U32_MAX PERL_UINT_MAX
262 # define U32_MIN PERL_UINT_MIN
264 # define I32_MAX PERL_LONG_MAX
265 # define I32_MIN PERL_LONG_MIN
266 # define U32_MAX PERL_ULONG_MAX
267 # define U32_MIN PERL_ULONG_MIN
272 /* log(2) is pretty close to 0.30103, just in case anyone is grepping for it */
273 #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
274 #define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
275 #define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
277 /* Unused by core; should be deprecated */
278 #define Ctl(ch) ((ch) & 037)
280 #if defined(PERL_CORE) || defined(PERL_EXT)
282 # define MIN(a,b) ((a) < (b) ? (a) : (b))
285 # define MAX(a,b) ((a) > (b) ? (a) : (b))
289 /* This is a helper macro to avoid preprocessor issues, replaced by nothing
290 * unless under DEBUGGING, where it expands to an assert of its argument,
291 * followed by a comma (hence the comma operator). If we just used a straight
292 * assert(), we would get a comma with nothing before it when not DEBUGGING.
294 * We also use empty definition under Coverity since the __ASSERT__
295 * checks often check for things that Really Cannot Happen, and Coverity
296 * detects that and gets all excited. */
298 #if defined(DEBUGGING) && !defined(__COVERITY__)
299 # define __ASSERT_(statement) assert(statement),
301 # define __ASSERT_(statement)
305 =head1 SV-Body Allocation
307 =for apidoc Ama|SV*|newSVpvs|const char* s
308 Like C<newSVpvn>, but takes a C<NUL>-terminated literal string instead of a
311 =for apidoc Ama|SV*|newSVpvs_flags|const char* s|U32 flags
312 Like C<newSVpvn_flags>, but takes a C<NUL>-terminated literal string instead of
313 a string/length pair.
315 =for apidoc Ama|SV*|newSVpvs_share|const char* s
316 Like C<newSVpvn_share>, but takes a C<NUL>-terminated literal string instead of
317 a string/length pair and omits the hash parameter.
319 =for apidoc Am|void|sv_catpvs_flags|SV* sv|const char* s|I32 flags
320 Like C<sv_catpvn_flags>, but takes a C<NUL>-terminated literal string instead
321 of a string/length pair.
323 =for apidoc Am|void|sv_catpvs_nomg|SV* sv|const char* s
324 Like C<sv_catpvn_nomg>, but takes a C<NUL>-terminated literal string instead of
325 a string/length pair.
327 =for apidoc Am|void|sv_catpvs|SV* sv|const char* s
328 Like C<sv_catpvn>, but takes a C<NUL>-terminated literal string instead of a
331 =for apidoc Am|void|sv_catpvs_mg|SV* sv|const char* s
332 Like C<sv_catpvn_mg>, but takes a C<NUL>-terminated literal string instead of a
335 =for apidoc Am|void|sv_setpvs|SV* sv|const char* s
336 Like C<sv_setpvn>, but takes a C<NUL>-terminated literal string instead of a
339 =for apidoc Am|void|sv_setpvs_mg|SV* sv|const char* s
340 Like C<sv_setpvn_mg>, but takes a C<NUL>-terminated literal string instead of a
343 =for apidoc Am|SV *|sv_setref_pvs|const char* s
344 Like C<sv_setref_pvn>, but takes a C<NUL>-terminated literal string instead of
345 a string/length pair.
347 =head1 Memory Management
349 =for apidoc Ama|char*|savepvs|const char* s
350 Like C<savepvn>, but takes a C<NUL>-terminated literal string instead of a
353 =for apidoc Ama|char*|savesharedpvs|const char* s
354 A version of C<savepvs()> which allocates the duplicate string in memory
355 which is shared between threads.
359 =for apidoc Am|HV*|gv_stashpvs|const char* name|I32 create
360 Like C<gv_stashpvn>, but takes a C<NUL>-terminated literal string instead of a
363 =head1 Hash Manipulation Functions
365 =for apidoc Am|SV**|hv_fetchs|HV* tb|const char* key|I32 lval
366 Like C<hv_fetch>, but takes a C<NUL>-terminated literal string instead of a
369 =for apidoc Am|SV**|hv_stores|HV* tb|const char* key|NULLOK SV* val
370 Like C<hv_store>, but takes a C<NUL>-terminated literal string instead of a
372 and omits the hash parameter.
374 =head1 Lexer interface
376 =for apidoc Amx|void|lex_stuff_pvs|const char *pv|U32 flags
378 Like L</lex_stuff_pvn>, but takes a C<NUL>-terminated literal string instead of
379 a string/length pair.
384 /* concatenating with "" ensures that only literal strings are accepted as
386 #define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
388 /* note that STR_WITH_LEN() can't be used as argument to macros or functions
389 * that under some configurations might be macros, which means that it requires
390 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
393 /* STR_WITH_LEN() shortcuts */
394 #define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
395 #define newSVpvs_flags(str,flags) \
396 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
397 #define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
398 #define sv_catpvs_flags(sv, str, flags) \
399 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
400 #define sv_catpvs_nomg(sv, str) \
401 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
402 #define sv_catpvs(sv, str) \
403 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
404 #define sv_catpvs_mg(sv, str) \
405 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
406 #define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
407 #define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
408 #define sv_setref_pvs(rv, classname, str) \
409 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
410 #define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
411 #define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
412 #define gv_stashpvs(str, create) \
413 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
414 #define gv_fetchpvs(namebeg, add, sv_type) \
415 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
416 #define gv_fetchpvn(namebeg, len, add, sv_type) \
417 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
418 #define sv_catxmlpvs(dsv, str, utf8) \
419 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
422 #define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
424 #define get_cvs(str, flags) \
425 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
428 =head1 Miscellaneous Functions
430 =for apidoc Am|bool|strNE|char* s1|char* s2
431 Test two C<NUL>-terminated strings to see if they are different. Returns true
434 =for apidoc Am|bool|strEQ|char* s1|char* s2
435 Test two C<NUL>-terminated strings to see if they are equal. Returns true or
438 =for apidoc Am|bool|strLT|char* s1|char* s2
439 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
440 second, C<s2>. Returns true or false.
442 =for apidoc Am|bool|strLE|char* s1|char* s2
443 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
444 equal to the second, C<s2>. Returns true or false.
446 =for apidoc Am|bool|strGT|char* s1|char* s2
447 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
448 the second, C<s2>. Returns true or false.
450 =for apidoc Am|bool|strGE|char* s1|char* s2
451 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
452 or equal to the second, C<s2>. Returns true or false.
454 =for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
455 Test two C<NUL>-terminated strings to see if they are different. The C<len>
456 parameter indicates the number of bytes to compare. Returns true or false. (A
457 wrapper for C<strncmp>).
459 =for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
460 Test two C<NUL>-terminated strings to see if they are equal. The C<len>
461 parameter indicates the number of bytes to compare. Returns true or false. (A
462 wrapper for C<strncmp>).
464 =for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
465 Test two buffers (which may contain embedded C<NUL> characters, to see if they
466 are equal. The C<len> parameter indicates the number of bytes to compare.
467 Returns zero if equal, or non-zero if non-equal.
469 =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
470 Test two buffers (which may contain embedded C<NUL> characters, to see if they
471 are not equal. The C<len> parameter indicates the number of bytes to compare.
472 Returns zero if non-equal, or non-zero if equal.
478 #define strNE(s1,s2) (strcmp(s1,s2))
479 #define strEQ(s1,s2) (!strcmp(s1,s2))
480 #define strLT(s1,s2) (strcmp(s1,s2) < 0)
481 #define strLE(s1,s2) (strcmp(s1,s2) <= 0)
482 #define strGT(s1,s2) (strcmp(s1,s2) > 0)
483 #define strGE(s1,s2) (strcmp(s1,s2) >= 0)
485 #define strnNE(s1,s2,l) (strncmp(s1,s2,l))
486 #define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
488 #define strNEs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1))
489 #define strEQs(s1,s2) (!strncmp(s1,"" s2 "", sizeof(s2)-1))
492 # define memNE(s1,s2,l) (memcmp(s1,s2,l))
493 # define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
495 # define memNE(s1,s2,l) (bcmp(s1,s2,l))
496 # define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
499 /* memEQ and memNE where second comparand is a string constant */
500 #define memEQs(s1, l, s2) \
501 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
502 #define memNEs(s1, l, s2) !memEQs(s1, l, s2)
504 /* memEQ and memNE where second comparand is a string constant
505 * and we can assume the length of s1 is at least that of the string */
506 #define _memEQs(s1, s2) \
507 (memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
508 #define _memNEs(s1, s2) (memNE((s1),("" s2 ""),(sizeof(s2)-1)))
510 #define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
511 #define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
512 #define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
513 #define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
518 * Unfortunately, the introduction of locales means that we
519 * can't trust isupper(), etc. to tell the truth. And when
520 * it comes to /\w+/ with tainting enabled, we *must* be able
521 * to trust our character classes.
523 * Therefore, the default tests in the text of Perl will be
524 * independent of locale. Any code that wants to depend on
525 * the current locale will use the tests that begin with "lc".
528 #ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
536 =head1 Character classification
537 This section is about functions (really macros) that classify characters
538 into types, such as punctuation versus alphabetic, etc. Most of these are
539 analogous to regular expression character classes. (See
540 L<perlrecharclass/POSIX Character Classes>.) There are several variants for
541 each class. (Not all macros have all variants; each item below lists the
542 ones valid for it.) None are affected by C<use bytes>, and only the ones
543 with C<LC> in the name are affected by the current locale.
545 The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
546 C<U8>) as input and returns a boolean as to whether or not the character
547 represented by that octet is (or on non-ASCII platforms, corresponds to) an
548 ASCII character in the named class based on platform, Unicode, and Perl rules.
549 If the input is a number that doesn't fit in an octet, FALSE is returned.
551 Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
552 with no suffix C<"_A">. This variant is used to emphasize by its name that
553 only ASCII-range characters can return TRUE.
555 Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
556 onto the platform. That is, the code points that are ASCII are unaffected,
557 since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
558 as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
559 true when called with the code point 0xDF, which is a word character in both
560 ASCII and EBCDIC (though it represents different characters in each).
562 Variant C<isFOO_uvchr> is like the C<isFOO_L1> variant, but accepts any UV code
563 point as input. If the code point is larger than 255, Unicode rules are used
564 to determine if it is in the character class. For example,
565 C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
566 WITH MACRON in Unicode, and is a word character.
568 Variant C<isFOO_utf8> is like C<isFOO_uvchr>, but the input is a pointer to a
569 (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
570 possibly containing embedded C<NUL> characters). The classification of just
571 the first (possibly multi-byte) character in the string is tested.
573 Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
574 result is based on the current locale, which is what C<LC> in the name stands
575 for. If Perl can determine that the current locale is a UTF-8 locale, it uses
576 the published Unicode rules; otherwise, it uses the C library function that
577 gives the named classification. For example, C<isDIGIT_LC()> when not in a
578 UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
579 returned if the input won't fit into an octet. On some platforms where the C
580 library function is known to be defective, Perl changes its result to follow
581 the POSIX standard's rules.
583 Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
584 returns the same as C<isFOO_LC> for input code points less than 256, and
585 returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
587 Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer
588 to a (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>, and
589 possibly containing embedded C<NUL> characters). The classification of just
590 the first (possibly multi-byte) character in the string is tested.
592 =for apidoc Am|bool|isALPHA|char ch
593 Returns a boolean indicating whether the specified character is an
594 alphabetic character, analogous to C<m/[[:alpha:]]/>.
595 See the L<top of this section|/Character classification> for an explanation of
597 C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>, C<isALPHA_LC>,
598 C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
600 =for apidoc Am|bool|isALPHANUMERIC|char ch
601 Returns a boolean indicating whether the specified character is a either an
602 alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
603 See the L<top of this section|/Character classification> for an explanation of
605 C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
606 C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
607 C<isALPHANUMERIC_LC_utf8>.
609 =for apidoc Am|bool|isASCII|char ch
610 Returns a boolean indicating whether the specified character is one of the 128
611 characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
612 On non-ASCII platforms, it returns TRUE iff this
613 character corresponds to an ASCII character. Variants C<isASCII_A()> and
614 C<isASCII_L1()> are identical to C<isASCII()>.
615 See the L<top of this section|/Character classification> for an explanation of
617 C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
618 C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
619 library routine C<isascii()>. In these cases, the variants whose names contain
620 C<LC> are the same as the corresponding ones without.
622 Also note, that because all ASCII characters are UTF-8 invariant (meaning they
623 have the exact same representation (always a single byte) whether encoded in
624 UTF-8 or not), C<isASCII> will give the correct results when called with any
625 byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
626 work properly on any string encoded or not in UTF-8.
628 =for apidoc Am|bool|isBLANK|char ch
629 Returns a boolean indicating whether the specified character is a
630 character considered to be a blank, analogous to C<m/[[:blank:]]/>.
631 See the L<top of this section|/Character classification> for an explanation of
633 C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>, C<isBLANK_LC>,
634 C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
635 platforms do not have the C library routine C<isblank()>. In these cases, the
636 variants whose names contain C<LC> are the same as the corresponding ones
639 =for apidoc Am|bool|isCNTRL|char ch
640 Returns a boolean indicating whether the specified character is a
641 control character, analogous to C<m/[[:cntrl:]]/>.
642 See the L<top of this section|/Character classification> for an explanation of
644 C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
645 C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
646 On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
648 =for apidoc Am|bool|isDIGIT|char ch
649 Returns a boolean indicating whether the specified character is a
650 digit, analogous to C<m/[[:digit:]]/>.
651 Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
652 See the L<top of this section|/Character classification> for an explanation of
654 C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
657 =for apidoc Am|bool|isGRAPH|char ch
658 Returns a boolean indicating whether the specified character is a
659 graphic character, analogous to C<m/[[:graph:]]/>.
660 See the L<top of this section|/Character classification> for an explanation of
662 C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
663 C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
665 =for apidoc Am|bool|isLOWER|char ch
666 Returns a boolean indicating whether the specified character is a
667 lowercase character, analogous to C<m/[[:lower:]]/>.
668 See the L<top of this section|/Character classification> for an explanation of
670 C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>, C<isLOWER_LC>,
671 C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
673 =for apidoc Am|bool|isOCTAL|char ch
674 Returns a boolean indicating whether the specified character is an
676 The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
679 =for apidoc Am|bool|isPUNCT|char ch
680 Returns a boolean indicating whether the specified character is a
681 punctuation character, analogous to C<m/[[:punct:]]/>.
682 Note that the definition of what is punctuation isn't as
683 straightforward as one might desire. See L<perlrecharclass/POSIX Character
684 Classes> for details.
685 See the L<top of this section|/Character classification> for an explanation of
687 C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
688 C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
690 =for apidoc Am|bool|isSPACE|char ch
691 Returns a boolean indicating whether the specified character is a
692 whitespace character. This is analogous
693 to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
694 this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
695 locale forms of this macro (the ones with C<LC> in their names) matched
696 precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
697 in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
698 (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
699 See the L<top of this section|/Character classification> for an explanation of
701 C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>, C<isSPACE_LC>,
702 C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
704 =for apidoc Am|bool|isPSXSPC|char ch
705 (short for Posix Space)
706 Starting in 5.18, this is identical in all its forms to the
707 corresponding C<isSPACE()> macros.
708 The locale forms of this macro are identical to their corresponding
709 C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
710 non-locale forms differ from their C<isSPACE()> forms only in that the
711 C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
712 Otherwise they are identical. Thus this macro is analogous to what
713 C<m/[[:space:]]/> matches in a regular expression.
714 See the L<top of this section|/Character classification> for an explanation of
715 variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
716 C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
718 =for apidoc Am|bool|isUPPER|char ch
719 Returns a boolean indicating whether the specified character is an
720 uppercase character, analogous to C<m/[[:upper:]]/>.
721 See the L<top of this section|/Character classification> for an explanation of
723 C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>, C<isUPPER_LC>,
724 C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
726 =for apidoc Am|bool|isPRINT|char ch
727 Returns a boolean indicating whether the specified character is a
728 printable character, analogous to C<m/[[:print:]]/>.
729 See the L<top of this section|/Character classification> for an explanation of
731 C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>, C<isPRINT_LC>,
732 C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
734 =for apidoc Am|bool|isWORDCHAR|char ch
735 Returns a boolean indicating whether the specified character is a character
736 that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
737 in a regular expression. A word character is an alphabetic character, a
738 decimal digit, a connecting punctuation character (such as an underscore), or
739 a "mark" character that attaches to one of those (like some sort of accent).
740 C<isALNUM()> is a synonym provided for backward compatibility, even though a
741 word character includes more than the standard C language meaning of
743 See the L<top of this section|/Character classification> for an explanation of
745 C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and C<isWORDCHAR_utf8>.
746 C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
747 described there, but additionally include the platform's native underscore.
749 =for apidoc Am|bool|isXDIGIT|char ch
750 Returns a boolean indicating whether the specified character is a hexadecimal
751 digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
752 and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
753 See the L<top of this section|/Character classification> for an explanation of
755 C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
758 =for apidoc Am|bool|isIDFIRST|char ch
759 Returns a boolean indicating whether the specified character can be the first
760 character of an identifier. This is very close to, but not quite the same as
761 the official Unicode property C<XID_Start>. The difference is that this
762 returns true only if the input character also matches L</isWORDCHAR>.
763 See the L<top of this section|/Character classification> for an explanation of
765 C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
766 C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
768 =for apidoc Am|bool|isIDCONT|char ch
769 Returns a boolean indicating whether the specified character can be the
770 second or succeeding character of an identifier. This is very close to, but
771 not quite the same as the official Unicode property C<XID_Continue>. The
772 difference is that this returns true only if the input character also matches
773 L</isWORDCHAR>. See the L<top of this section|/Character classification> for
775 explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
776 C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
779 =head1 Miscellaneous Functions
781 =for apidoc Am|U8|READ_XDIGIT|char str*
782 Returns the value of an ASCII-range hex digit and advances the string pointer.
783 Behaviour is only well defined when isXDIGIT(*str) is true.
785 =head1 Character case changing
786 Perl uses "full" Unicode case mappings. This means that converting a single
787 character to another case may result in a sequence of more than one character.
788 For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
789 character sequence C<SS>. This presents some complications The lowercase of
790 all characters in the range 0..255 is a single character, and thus
791 C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
792 return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
793 an API that does allow every possible legal result to be returned.) Likewise
794 no other function that is crippled by not being able to give the correct
795 results for the full range of possible inputs has been implemented here.
797 =for apidoc Am|U8|toUPPER|U8 ch
798 Converts the specified character to uppercase. If the input is anything but an
799 ASCII lowercase character, that input character itself is returned. Variant
800 C<toUPPER_A> is equivalent.
802 =for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
803 Converts the code point C<cp> to its uppercase version, and
804 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
805 point is interpreted as native if less than 256; otherwise as Unicode. Note
806 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
807 bytes since the uppercase version may be longer than the original character.
809 The first code point of the uppercased version is returned
810 (but note, as explained at L<the top of this section|/Character case
811 changing>, that there may be more.)
813 =for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
814 Converts the UTF-8 encoded character at C<p> to its uppercase version, and
815 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
816 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
817 bytes since the uppercase version may be longer than the original character.
819 The first code point of the uppercased version is returned
820 (but note, as explained at L<the top of this section|/Character case
821 changing>, that there may be more).
823 The input character at C<p> is assumed to be well-formed.
825 =for apidoc Am|U8|toFOLD|U8 ch
826 Converts the specified character to foldcase. If the input is anything but an
827 ASCII uppercase character, that input character itself is returned. Variant
828 C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
829 Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
831 =for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
832 Converts the code point C<cp> to its foldcase version, and
833 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
834 point is interpreted as native if less than 256; otherwise as Unicode. Note
835 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
836 bytes since the foldcase version may be longer than the original character.
838 The first code point of the foldcased version is returned
839 (but note, as explained at L<the top of this section|/Character case
840 changing>, that there may be more).
842 =for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
843 Converts the UTF-8 encoded character at C<p> to its foldcase version, and
844 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
845 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
846 bytes since the foldcase version may be longer than the original character.
848 The first code point of the foldcased version is returned
849 (but note, as explained at L<the top of this section|/Character case
850 changing>, that there may be more).
852 The input character at C<p> is assumed to be well-formed.
854 =for apidoc Am|U8|toLOWER|U8 ch
855 Converts the specified character to lowercase. If the input is anything but an
856 ASCII uppercase character, that input character itself is returned. Variant
857 C<toLOWER_A> is equivalent.
859 =for apidoc Am|U8|toLOWER_L1|U8 ch
860 Converts the specified Latin1 character to lowercase. The results are
861 undefined if the input doesn't fit in a byte.
863 =for apidoc Am|U8|toLOWER_LC|U8 ch
864 Converts the specified character to lowercase using the current locale's rules,
865 if possible; otherwise returns the input character itself.
867 =for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
868 Converts the code point C<cp> to its lowercase version, and
869 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
870 point is interpreted as native if less than 256; otherwise as Unicode. Note
871 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
872 bytes since the lowercase version may be longer than the original character.
874 The first code point of the lowercased version is returned
875 (but note, as explained at L<the top of this section|/Character case
876 changing>, that there may be more).
878 =for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
879 Converts the UTF-8 encoded character at C<p> to its lowercase version, and
880 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
881 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
882 bytes since the lowercase version may be longer than the original character.
884 The first code point of the lowercased version is returned
885 (but note, as explained at L<the top of this section|/Character case
886 changing>, that there may be more).
888 The input character at C<p> is assumed to be well-formed.
890 =for apidoc Am|U8|toTITLE|U8 ch
891 Converts the specified character to titlecase. If the input is anything but an
892 ASCII lowercase character, that input character itself is returned. Variant
893 C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
894 range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
895 not a concept used in locale handling, so there is no functionality for that.)
897 =for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
898 Converts the code point C<cp> to its titlecase version, and
899 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
900 point is interpreted as native if less than 256; otherwise as Unicode. Note
901 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
902 bytes since the titlecase version may be longer than the original character.
904 The first code point of the titlecased version is returned
905 (but note, as explained at L<the top of this section|/Character case
906 changing>, that there may be more).
908 =for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
909 Converts the UTF-8 encoded character at C<p> to its titlecase version, and
910 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
911 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
912 bytes since the titlecase version may be longer than the original character.
914 The first code point of the titlecased version is returned
915 (but note, as explained at L<the top of this section|/Character case
916 changing>, that there may be more).
918 The input character at C<p> is assumed to be well-formed.
922 XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
923 really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
925 Note that these macros are repeated in Devel::PPPort, so should also be
926 patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
930 /* Specify the widest unsigned type on the platform. Use U64TYPE because U64
931 * is known only in the perl core, and this macro can be called from outside
934 # define WIDEST_UTYPE U64TYPE
936 # define WIDEST_UTYPE U32
939 /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
940 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
941 * bits of information are lost even on a 64-bit machine, but to get the
942 * compiler to optimize it out if possible. This is because Configure makes
943 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
944 * sizeof() guarantees that this evaluates to a constant true at compile time.
946 * For Coverity, be always true, because otherwise Coverity thinks
947 * it finds several expressions that are always true, independent
948 * of operands. Well, they are, but that is kind of the point.
951 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
953 #define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
954 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
956 #define FITS_IN_8_BITS(c) (1)
961 /* The native libc isascii() et.al. functions return the wrong results
962 * on at least z/OS unless this is defined. */
963 # error _ALL_SOURCE should probably be defined
966 /* There is a simple definition of ASCII for ASCII platforms. But the
967 * EBCDIC one isn't so simple, so is defined using table look-up like the
968 * other macros below.
970 * The cast here is used instead of '(c) >= 0', because some compilers emit
971 * a warning that that test is always true when the parameter is an
972 * unsigned type. khw supposes that it could be written as
973 * && ((c) == '\0' || (c) > 0)
974 * to avoid the message, but the cast will likely avoid extra branches even
975 * with stupid compilers.
977 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
979 # define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
982 /* Take the eight possible bit patterns of the lower 3 bits and you get the
983 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
984 * can be ignored. If the rest match '0', we have an octal */
985 #define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
987 #ifdef H_PERL /* If have access to perl.h, lookup in its table */
989 /* Character class numbers. For internal core Perl use only. The ones less
990 * than 32 are used in PL_charclass[] and the ones up through the one that
991 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
992 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
993 * their actual definitions are here. If that file has a name not used here,
996 * The first group of these is ordered in what I (khw) estimate to be the
997 * frequency of their use. This gives a slight edge to exiting a loop earlier
998 * (in reginclass() in regexec.c) */
999 # define _CC_WORDCHAR 0 /* \w and [:word:] */
1000 # define _CC_DIGIT 1 /* \d and [:digit:] */
1001 # define _CC_ALPHA 2 /* [:alpha:] */
1002 # define _CC_LOWER 3 /* [:lower:] */
1003 # define _CC_UPPER 4 /* [:upper:] */
1004 # define _CC_PUNCT 5 /* [:punct:] */
1005 # define _CC_PRINT 6 /* [:print:] */
1006 # define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1007 # define _CC_GRAPH 8 /* [:graph:] */
1008 # define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1010 #define _FIRST_NON_SWASH_CC 10
1011 /* The character classes above are implemented with swashes. The second group
1012 * (just below) contains the ones implemented without. These are also sorted
1013 * in rough order of the frequency of their use, except that \v should be last,
1014 * as it isn't a real Posix character class, and some (small) inefficiencies in
1015 * regular expression handling would be introduced by putting it in the middle
1016 * of those that are. Also, cntrl and ascii come after the others as it may be
1017 * useful to group these which have no members that match above Latin1, (or
1018 * above ASCII in the latter case) */
1020 # define _CC_SPACE 10 /* \s, [:space:] */
1021 # define _CC_BLANK 11 /* [:blank:] */
1022 # define _CC_XDIGIT 12 /* [:xdigit:] */
1023 # define _CC_CNTRL 13 /* [:cntrl:] */
1024 # define _CC_ASCII 14 /* [:ascii:] */
1025 # define _CC_VERTSPACE 15 /* \v */
1027 # define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1029 /* The members of the third group below do not need to be coordinated with data
1030 * structures in regcomp.[ch] and regexec.c. */
1031 # define _CC_IDFIRST 16
1032 # define _CC_CHARNAME_CONT 17
1033 # define _CC_NONLATIN1_FOLD 18
1034 # define _CC_NONLATIN1_SIMPLE_FOLD 19
1035 # define _CC_QUOTEMETA 20
1036 # define _CC_NON_FINAL_FOLD 21
1037 # define _CC_IS_IN_SOME_FOLD 22
1038 # define _CC_MNEMONIC_CNTRL 23
1040 /* This next group is only used on EBCDIC platforms, so theoretically could be
1041 * shared with something entirely different that's only on ASCII platforms */
1042 # define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1043 # define _CC_UTF8_IS_START 29
1044 # define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1045 # define _CC_UTF8_IS_CONTINUATION 31
1047 * If more bits are needed, one could add a second word for non-64bit
1048 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
1049 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1050 * is used only for optimization (as of this writing), and differs in the
1051 * Latin1 range from the ALPHA bit only in two relatively unimportant
1052 * characters: the masculine and feminine ordinal indicators, so removing it
1053 * would just cause /i regexes which match them to run less efficiently.
1054 * Similarly the EBCDIC-only bits are used just for speed, and could be
1055 * replaced by other means */
1057 #if defined(PERL_CORE) || defined(PERL_EXT)
1058 /* An enum version of the character class numbers, to help compilers
1061 _CC_ENUM_ALPHA = _CC_ALPHA,
1062 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1063 _CC_ENUM_ASCII = _CC_ASCII,
1064 _CC_ENUM_BLANK = _CC_BLANK,
1065 _CC_ENUM_CASED = _CC_CASED,
1066 _CC_ENUM_CNTRL = _CC_CNTRL,
1067 _CC_ENUM_DIGIT = _CC_DIGIT,
1068 _CC_ENUM_GRAPH = _CC_GRAPH,
1069 _CC_ENUM_LOWER = _CC_LOWER,
1070 _CC_ENUM_PRINT = _CC_PRINT,
1071 _CC_ENUM_PUNCT = _CC_PUNCT,
1072 _CC_ENUM_SPACE = _CC_SPACE,
1073 _CC_ENUM_UPPER = _CC_UPPER,
1074 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
1075 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
1076 _CC_ENUM_XDIGIT = _CC_XDIGIT
1077 } _char_class_number;
1080 #define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
1081 #define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
1083 #if defined(PERL_IN_UTF8_C) \
1084 || defined(PERL_IN_REGCOMP_C) \
1085 || defined(PERL_IN_REGEXEC_C)
1086 # if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
1087 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1088 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
1089 #error Need to adjust order of swash_property_names[]
1092 /* This is declared static in each of the few files that this is #defined for
1093 * to keep them from being publicly accessible. Hence there is a small amount
1094 * of wasted space */
1096 static const char* const swash_property_names[] = {
1112 EXTCONST U32 PL_charclass[] = {
1113 # include "l1_char_class_tab.h"
1116 # else /* ! DOINIT */
1117 EXTCONST U32 PL_charclass[];
1121 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
1122 # define _CC_mask(classnum) (1U << (classnum))
1124 /* For internal core Perl use only: the base macro for defining macros like
1126 # define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
1127 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
1129 /* The mask for the _A versions of the macros; it just adds in the bit for
1131 # define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1133 /* For internal core Perl use only: the base macro for defining macros like
1134 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1135 * the ASCII bit are present */
1136 # define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1137 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1138 == _CC_mask_A(classnum)))
1140 # define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1141 # define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
1142 # define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1143 # define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
1144 # define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
1145 # define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1146 # define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1147 # define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1148 # define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1149 # define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1150 # define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1151 # define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
1152 # define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1154 # define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
1155 # define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1156 # define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1157 # define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1159 /* continuation character for legal NAME in \N{NAME} */
1160 # define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1162 # define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1163 # define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1164 # define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1165 # define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1166 # define isPSXSPC_L1(c) isSPACE_L1(c)
1167 # define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1168 # define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1169 # define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1170 # define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1171 # define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
1174 # define isASCII(c) _generic_isCC(c, _CC_ASCII)
1177 /* Participates in a single-character fold with a character above 255 */
1178 # 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)))
1180 /* Like the above, but also can be part of a multi-char fold */
1181 # 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)))
1183 # define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
1184 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1185 _generic_isCC(c, _CC_NON_FINAL_FOLD)
1186 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1187 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1188 # define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1189 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1190 #else /* else we don't have perl.h H_PERL */
1192 /* If we don't have perl.h, we are compiling a utility program. Below we
1193 * hard-code various macro definitions that wouldn't otherwise be available
1194 * to it. Most are coded based on first principles. These are written to
1195 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
1196 # define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1197 # define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1198 # define isSPACE_A(c) (isBLANK_A(c) \
1203 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1204 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1205 * performance (if optimization isn't on), and allow the same code to be
1206 * used for both platform types */
1207 # define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1209 || ((c) >= 'j' && (c) <= 'r') \
1211 # define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1213 || ((c) >= 'J' && (c) <= 'R') \
1215 # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1216 # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
1217 # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
1218 # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1219 # define isXDIGIT_A(c) (isDIGIT_A(c) \
1220 || ((c) >= 'a' && (c) <= 'f') \
1221 || ((c) <= 'F' && (c) >= 'A'))
1222 # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1223 || (c) == '#' || (c) == '$' || (c) == '%' \
1224 || (c) == '&' || (c) == '\'' || (c) == '(' \
1225 || (c) == ')' || (c) == '*' || (c) == '+' \
1226 || (c) == ',' || (c) == '.' || (c) == '/' \
1227 || (c) == ':' || (c) == ';' || (c) == '<' \
1228 || (c) == '=' || (c) == '>' || (c) == '?' \
1229 || (c) == '@' || (c) == '[' || (c) == '\\' \
1230 || (c) == ']' || (c) == '^' || (c) == '_' \
1231 || (c) == '`' || (c) == '{' || (c) == '|' \
1232 || (c) == '}' || (c) == '~')
1233 # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1234 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1237 /* The below is accurate for the 3 EBCDIC code pages traditionally
1238 * supported by perl. The only difference between them in the controls
1239 * is the position of \n, and that is represented symbolically below */
1240 # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1241 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1242 || (c) == '\t' || (c) == '\v' \
1243 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1244 || (c) == 7 /* U+7F DEL */ \
1245 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1246 /* DLE, DC[1-3] */ \
1247 || (c) == 0x18 /* U+18 CAN */ \
1248 || (c) == 0x19 /* U+19 EOM */ \
1249 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1250 || (c) == 0x26 /* U+17 ETB */ \
1251 || (c) == 0x27 /* U+1B ESC */ \
1252 || (c) == 0x2D /* U+05 ENQ */ \
1253 || (c) == 0x2E /* U+06 ACK */ \
1254 || (c) == 0x32 /* U+16 SYN */ \
1255 || (c) == 0x37 /* U+04 EOT */ \
1256 || (c) == 0x3C /* U+14 DC4 */ \
1257 || (c) == 0x3D /* U+15 NAK */ \
1258 || (c) == 0x3F)/* U+1A SUB */
1259 # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1260 # else /* isASCII is already defined for ASCII platforms, so can use that to
1262 # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
1265 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
1266 * during debugging, and it seems best to keep them. We may be called
1267 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1268 * do anything anyway, so make it not a problem */
1269 # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1270 # define NATIVE_TO_LATIN1(ch) (ch)
1272 # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1273 # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1274 # define isBLANK_L1(c) (isBLANK_A(c) \
1275 || (FITS_IN_8_BITS(c) \
1276 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1277 # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1278 # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1279 # define isLOWER_L1(c) (isLOWER_A(c) \
1280 || (FITS_IN_8_BITS(c) \
1281 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1282 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1283 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1284 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1285 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1286 # define isPRINT_L1(c) (isPRINT_A(c) \
1287 || (FITS_IN_8_BITS(c) \
1288 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1289 # define isPUNCT_L1(c) (isPUNCT_A(c) \
1290 || (FITS_IN_8_BITS(c) \
1291 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1292 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1293 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1294 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1295 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1296 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1297 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1298 # define isSPACE_L1(c) (isSPACE_A(c) \
1299 || (FITS_IN_8_BITS(c) \
1300 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1301 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1302 # define isUPPER_L1(c) (isUPPER_A(c) \
1303 || (FITS_IN_8_BITS(c) \
1304 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1305 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1306 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1307 # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1308 # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1309 # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1314 /* The following are not fully accurate in the above-ASCII range. I (khw)
1315 * don't think it's necessary to be so for the purposes where this gets
1317 # define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1318 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1320 /* And these aren't accurate at all. They are useful only for above
1321 * Latin1, which utilities and bootstrapping don't deal with */
1322 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
1323 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1324 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1326 /* Many of the macros later in this file are defined in terms of these. By
1327 * implementing them with a function, which converts the class number into
1328 * a call to the desired macro, all of the later ones work. However, that
1329 * function won't be actually defined when building a utility program (no
1330 * perl.h), and so a compiler error will be generated if one is attempted
1331 * to be used. And the above-Latin1 code points require Unicode tables to
1332 * be present, something unlikely to be the case when bootstrapping */
1333 # define _generic_isCC(c, classnum) \
1334 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1335 # define _generic_isCC_A(c, classnum) \
1336 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
1337 #endif /* End of no perl.h H_PERL */
1339 #define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1340 #define isALPHA(c) isALPHA_A(c)
1341 #define isASCII_A(c) isASCII(c)
1342 #define isASCII_L1(c) isASCII(c)
1343 #define isBLANK(c) isBLANK_A(c)
1344 #define isCNTRL(c) isCNTRL_A(c)
1345 #define isDIGIT(c) isDIGIT_A(c)
1346 #define isGRAPH(c) isGRAPH_A(c)
1347 #define isIDFIRST(c) isIDFIRST_A(c)
1348 #define isLOWER(c) isLOWER_A(c)
1349 #define isPRINT(c) isPRINT_A(c)
1350 #define isPSXSPC_A(c) isSPACE_A(c)
1351 #define isPSXSPC(c) isPSXSPC_A(c)
1352 #define isPSXSPC_L1(c) isSPACE_L1(c)
1353 #define isPUNCT(c) isPUNCT_A(c)
1354 #define isSPACE(c) isSPACE_A(c)
1355 #define isUPPER(c) isUPPER_A(c)
1356 #define isWORDCHAR(c) isWORDCHAR_A(c)
1357 #define isXDIGIT(c) isXDIGIT_A(c)
1359 /* ASCII casing. These could also be written as
1360 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1361 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1362 which uses table lookup and mask instead of subtraction. (This would
1363 work because the _MOD does not apply in the ASCII range) */
1364 #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1365 #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
1367 /* In the ASCII range, these are equivalent to what they're here defined to be.
1368 * But by creating these definitions, other code doesn't have to be aware of
1370 #define toFOLD(c) toLOWER(c)
1371 #define toTITLE(c) toUPPER(c)
1373 #define toLOWER_A(c) toLOWER(c)
1374 #define toUPPER_A(c) toUPPER(c)
1375 #define toFOLD_A(c) toFOLD(c)
1376 #define toTITLE_A(c) toTITLE(c)
1378 /* Use table lookup for speed; returns the input itself if is out-of-range */
1379 #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
1381 : PL_latin1_lc[ (U8) (c) ])
1382 #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1384 /* Modified uc. Is correct uc except for three non-ascii chars which are
1385 * all mapped to one of them, and these need special handling; returns the
1386 * input itself if is out-of-range */
1387 #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
1389 : PL_mod_latin1_uc[ (U8) (c) ])
1390 #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
1392 /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1394 /* For internal core Perl use only: the base macro for defining macros like
1395 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
1396 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1397 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1398 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1399 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1400 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1401 * code points under locale rules, use the macros based on _generic_LC_uvchr
1402 * instead of this. */
1403 #define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1404 (! FITS_IN_8_BITS(c) \
1406 : IN_UTF8_CTYPE_LOCALE \
1407 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
1410 /* For internal core Perl use only: a helper macro for defining macros like
1411 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1412 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1413 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1414 * _generic_LC_base for more info */
1415 #define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1416 _generic_LC_base(c,utf8_locale_classnum, \
1417 non_utf8_func( (_LC_CAST) (c)))
1419 /* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1420 * 'c' is the platform's native underscore character */
1421 #define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1422 _generic_LC_base(c, utf8_locale_classnum, \
1423 (non_utf8_func( (_LC_CAST) (c)) \
1424 || (char)(c) == '_'))
1426 /* These next three are also for internal core Perl use only: case-change
1428 #define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1430 : (IN_UTF8_CTYPE_LOCALE) \
1431 ? PL_latin1_lc[ (U8) (c) ] \
1432 : (cast)function((cast)(c)))
1434 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1435 * returns a single value, so can't adequately return the upper case of LATIN
1436 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1437 * values "SS"); instead it asserts against that under DEBUGGING, and
1438 * otherwise returns its input */
1439 #define _generic_toUPPER_LC(c, function, cast) \
1440 (! FITS_IN_8_BITS(c) \
1442 : ((! IN_UTF8_CTYPE_LOCALE) \
1443 ? (cast)function((cast)(c)) \
1444 : ((((U8)(c)) == MICRO_SIGN) \
1445 ? GREEK_CAPITAL_LETTER_MU \
1446 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1447 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1448 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1449 ? (__ASSERT_(0) (c)) \
1450 : PL_mod_latin1_uc[ (U8) (c) ])))))
1452 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1453 * returns a single value, so can't adequately return the fold case of LATIN
1454 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1455 * values "ss"); instead it asserts against that under DEBUGGING, and
1456 * otherwise returns its input */
1457 #define _generic_toFOLD_LC(c, function, cast) \
1458 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
1459 ? GREEK_SMALL_LETTER_MU \
1460 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1461 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1462 _generic_toLOWER_LC(c, function, cast)))
1464 /* Use the libc versions for these if available. */
1465 #if defined(HAS_ISASCII)
1466 # define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1468 # define isASCII_LC(c) isASCII(c)
1471 #if defined(HAS_ISBLANK)
1472 # define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
1473 #else /* Unlike isASCII, varies if in a UTF-8 locale */
1474 # define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
1480 /* The Windows functions don't bother to follow the POSIX standard, which
1481 * for example says that something can't both be a printable and a control.
1482 * But Windows treats the \t control as a printable, and does such things
1483 * as making superscripts into both digits and punctuation. This tames
1484 * these flaws by assuming that the definitions of both controls and space
1485 * are correct, and then making sure that other definitions don't have
1486 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1487 * Not all possible weirdnesses are checked for, just the ones that were
1488 * detected on actual Microsoft code pages */
1490 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1491 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1493 # define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1494 && isALPHANUMERIC_LC(c))
1495 # define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1497 # define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1498 isALPHANUMERIC_LC(c))
1499 # define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1500 # define isIDFIRST_LC(c) (((c) == '_') \
1501 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1502 # define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1503 # define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1504 # define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1505 # define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1506 # define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1507 # define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1508 && isALPHANUMERIC_LC(c))
1510 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1511 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1512 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1514 #elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
1515 /* For most other platforms */
1517 # define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1518 # define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1519 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1520 # define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1521 # define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1522 # define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1523 # define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1524 # define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1525 # define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1526 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1527 # define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1528 # define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1529 # define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1532 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1533 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1534 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1536 #else /* The final fallback position */
1538 # define isALPHA_LC(c) (isascii(c) && isalpha(c))
1539 # define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1540 # define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1541 # define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1542 # define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1543 # define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1544 # define isLOWER_LC(c) (isascii(c) && islower(c))
1545 # define isPRINT_LC(c) (isascii(c) && isprint(c))
1546 # define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1547 # define isSPACE_LC(c) (isascii(c) && isspace(c))
1548 # define isUPPER_LC(c) (isascii(c) && isupper(c))
1549 # define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1550 # define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1552 # define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1553 # define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1554 # define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
1558 #define isIDCONT(c) isWORDCHAR(c)
1559 #define isIDCONT_A(c) isWORDCHAR_A(c)
1560 #define isIDCONT_L1(c) isWORDCHAR_L1(c)
1561 #define isIDCONT_LC(c) isWORDCHAR_LC(c)
1562 #define isPSXSPC_LC(c) isSPACE_LC(c)
1564 /* For internal core Perl use only: the base macros for defining macros like
1565 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1566 * number defined earlier in this file. _generic_uvchr() is used for POSIX
1567 * classes where there is a macro or function 'above_latin1' that takes the
1568 * single argument 'c' and returns the desired value. These exist for those
1569 * classes which have simple definitions, avoiding the overhead of a hash
1570 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
1571 * for classes where that overhead is faster than a direct lookup.
1572 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
1573 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1574 * have duplicate checks here, so could create versions of the macros that
1575 * don't, but experiments show that gcc optimizes them out anyway. */
1577 /* Note that all ignore 'use bytes' */
1578 #define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1579 ? _generic_isCC(c, classnum) \
1581 #define _generic_swash_uvchr(classnum, c) ((c) < 256 \
1582 ? _generic_isCC(c, classnum) \
1583 : _is_uni_FOO(classnum, c))
1584 #define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1585 #define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1586 #define isASCII_uvchr(c) isASCII(c)
1587 #define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1588 #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1589 #define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1590 #define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1591 #define isIDCONT_uvchr(c) _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1592 #define isIDFIRST_uvchr(c) _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
1593 #define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1594 #define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1596 #define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1597 #define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1598 #define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1600 #define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1601 #define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1602 #define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1603 #define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1605 #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1606 #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1607 #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1608 #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1610 /* For backwards compatibility, even though '_uni' should mean official Unicode
1611 * code points, in Perl it means native for those below 256 */
1612 #define isALPHA_uni(c) isALPHA_uvchr(c)
1613 #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1614 #define isASCII_uni(c) isASCII_uvchr(c)
1615 #define isBLANK_uni(c) isBLANK_uvchr(c)
1616 #define isCNTRL_uni(c) isCNTRL_uvchr(c)
1617 #define isDIGIT_uni(c) isDIGIT_uvchr(c)
1618 #define isGRAPH_uni(c) isGRAPH_uvchr(c)
1619 #define isIDCONT_uni(c) isIDCONT_uvchr(c)
1620 #define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1621 #define isLOWER_uni(c) isLOWER_uvchr(c)
1622 #define isPRINT_uni(c) isPRINT_uvchr(c)
1623 #define isPUNCT_uni(c) isPUNCT_uvchr(c)
1624 #define isSPACE_uni(c) isSPACE_uvchr(c)
1625 #define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1626 #define isUPPER_uni(c) isUPPER_uvchr(c)
1627 #define isVERTWS_uni(c) isVERTWS_uvchr(c)
1628 #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1629 #define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1630 #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1631 #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1632 #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1633 #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
1635 /* For internal core Perl use only: the base macros for defining macros like
1636 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1637 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
1638 * simple class definitions; the other for more complex. These are like
1639 * _generic_uvchr, so see it for more info. */
1640 #define _generic_LC_uvchr(latin1, above_latin1, c) \
1641 (c < 256 ? latin1(c) : above_latin1(c))
1642 #define _generic_LC_swash_uvchr(latin1, classnum, c) \
1643 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1645 #define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1646 #define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1647 _CC_ALPHANUMERIC, c)
1648 #define isASCII_LC_uvchr(c) isASCII_LC(c)
1649 #define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1650 is_HORIZWS_cp_high, c)
1651 #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
1652 #define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1653 #define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
1654 #define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
1655 _is_uni_perl_idcont, c)
1656 #define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
1657 _is_uni_perl_idstart, c)
1658 #define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1659 #define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
1660 #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
1661 #define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
1662 #define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
1663 is_XPERLSPACE_cp_high, c)
1664 #define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
1665 #define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
1667 #define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1668 is_XDIGIT_cp_high, c)
1670 #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
1672 /* For internal core Perl use only: the base macros for defining macros like
1673 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1674 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1675 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1676 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1677 * same representation whether utf8 or not. Note that it assumes that the utf8
1678 * has been validated, and ignores 'use bytes' */
1679 #define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
1680 ? _generic_isCC(*(p), classnum) \
1681 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1683 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
1687 /* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1688 * 'above_latin1' can be a macro */
1689 #define _generic_func_utf8(classnum, above_latin1, p) \
1690 _generic_utf8(classnum, p, above_latin1(p))
1691 /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
1692 * 'above_latin1' parameter */
1693 #define _generic_swash_utf8(classnum, p) \
1694 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
1696 /* Like the above, but should be used only when it is known that there are no
1697 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1698 * class is TRUE for. Hence it can skip the tests for this range.
1699 * 'above_latin1' should include its arguments */
1700 #define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
1701 (UTF8_IS_INVARIANT(*(p)) \
1702 ? _generic_isCC(*(p), classnum) \
1703 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
1707 /* NOTE that some of these macros have very similar ones in regcharclass.h.
1708 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1709 * there, differing in name only by an underscore from the one here
1710 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1711 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1712 * points; the regcharclass.h ones are implemented as a series of
1713 * "if-else-if-else ..." */
1715 #define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1716 #define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1717 #define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
1718 utf8, the non-utf8 macro works
1720 #define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
1723 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1724 * more efficient macro instead of the more general one */
1725 # define isCNTRL_utf8(p) isCNTRL_L1(*(p))
1727 # define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1730 #define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1731 _is_utf8_FOO(_CC_DIGIT, p))
1732 #define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
1733 #define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
1734 _is_utf8_perl_idcont, p)
1736 /* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1737 * IDFIRST is an alnum. See
1738 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
1739 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1740 * This used to be not the XID version, but we decided to go with the more
1741 * modern Unicode definition */
1742 #define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
1743 _is_utf8_perl_idstart, p)
1745 #define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1746 #define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1747 #define isPSXSPC_utf8(p) isSPACE_utf8(p)
1748 #define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1749 #define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1750 #define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1751 #define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1752 #define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1753 #define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
1756 #define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
1757 #define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1758 #define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1759 #define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
1761 /* For internal core Perl use only: the base macros for defining macros like
1762 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1763 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1764 * 'macro' parameter */
1765 #define _generic_LC_utf8(macro, p, utf8) \
1766 (UTF8_IS_INVARIANT(*(p)) \
1768 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1769 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
1772 #define _generic_LC_swash_utf8(macro, classnum, p) \
1773 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1774 #define _generic_LC_func_utf8(macro, above_latin1, p) \
1775 _generic_LC_utf8(macro, p, above_latin1(p))
1777 #define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
1778 _CC_ALPHANUMERIC, p)
1779 #define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1780 #define isASCII_LC_utf8(p) isASCII_LC(*p)
1781 #define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1783 #define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1784 #define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1785 #define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1786 #define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1787 _is_utf8_perl_idcont, p)
1788 #define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1789 _is_utf8_perl_idstart, p)
1790 #define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1791 #define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1792 #define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1793 #define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1794 #define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1795 is_XPERLSPACE_high, p)
1796 #define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
1797 #define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1799 #define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1802 /* Macros for backwards compatibility and for completeness when the ASCII and
1803 * Latin1 values are identical */
1804 #define isALPHAU(c) isALPHA_L1(c)
1805 #define isDIGIT_L1(c) isDIGIT_A(c)
1806 #define isOCTAL(c) isOCTAL_A(c)
1807 #define isOCTAL_L1(c) isOCTAL_A(c)
1808 #define isXDIGIT_L1(c) isXDIGIT_A(c)
1809 #define isALNUM(c) isWORDCHAR(c)
1810 #define isALNUMU(c) isWORDCHAR_L1(c)
1811 #define isALNUM_LC(c) isWORDCHAR_LC(c)
1812 #define isALNUM_uni(c) isWORDCHAR_uni(c)
1813 #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
1814 #define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1815 #define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1816 #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1817 #define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1818 #define isALNUMC(c) isALPHANUMERIC(c)
1819 #define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1820 #define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
1821 #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
1822 #define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
1823 #define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
1825 /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1826 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1827 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1828 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1829 * the outlier from the block that contains the other controls, just like
1830 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1831 * block. If it weren't special cased, it would yield a non-control.
1832 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1835 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
1837 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1839 ? (UNLIKELY((c) == '?') \
1840 ? QUESTION_MARK_CTRL \
1841 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1842 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1844 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
1847 /* Line numbers are unsigned, 32 bits. */
1849 #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
1851 /* Helpful alias for version prescan */
1852 #define is_LAX_VERSION(a,b) \
1853 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1855 #define is_STRICT_VERSION(a,b) \
1856 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1858 #define BADVERSION(a,b,c) \
1864 /* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1865 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1866 * which is advanced. The input is validated only by an assert() in DEBUGGING
1867 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1868 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1869 #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1872 #define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1876 /* Converts a character known to represent an octal digit (0-7) to its numeric
1877 * value. The input is validated only by an assert() in DEBUGGING builds. In
1878 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1879 #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1881 /* Efficiently returns a boolean as to if two native characters are equivalent
1882 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1883 * the ALPHA in the name is to remind you of that. This is asserted() in
1884 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1885 * works (on valid input) for both non- and UTF-8-encoded bytes.
1887 * When one of the inputs is a compile-time constant and gets folded by the
1888 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1889 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1890 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
1891 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1892 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1893 * just a single 0, in the bit position where the upper- and lowercase differ.
1895 #define isALPHA_FOLD_EQ(c1, c2) \
1896 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1897 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1898 #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1901 =head1 Memory Management
1903 =for apidoc Am|void|Newx|void* ptr|int nitems|type
1904 The XSUB-writer's interface to the C C<malloc> function.
1906 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
1908 In 5.9.3, Newx() and friends replace the older New() API, and drops
1909 the first parameter, I<x>, a debug aid which allowed callers to identify
1910 themselves. This aid has been superseded by a new build option,
1911 PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
1912 there for use in XS modules supporting older perls.
1914 =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
1915 The XSUB-writer's interface to the C C<malloc> function, with
1916 cast. See also C<L</Newx>>.
1918 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
1920 =for apidoc Am|void|Newxz|void* ptr|int nitems|type
1921 The XSUB-writer's interface to the C C<malloc> function. The allocated
1922 memory is zeroed with C<memzero>. See also C<L</Newx>>.
1924 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
1926 =for apidoc Am|void|Renew|void* ptr|int nitems|type
1927 The XSUB-writer's interface to the C C<realloc> function.
1929 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
1931 =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1932 The XSUB-writer's interface to the C C<realloc> function, with
1935 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
1937 =for apidoc Am|void|Safefree|void* ptr
1938 The XSUB-writer's interface to the C C<free> function.
1940 This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
1942 =for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1943 The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
1944 source, C<dest> is the destination, C<nitems> is the number of items, and
1945 C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
1947 =for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
1948 Like C<Move> but returns C<dest>. Useful
1949 for encouraging compilers to tail-call
1952 =for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1953 The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
1954 source, C<dest> is the destination, C<nitems> is the number of items, and
1955 C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
1957 =for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1959 Like C<Copy> but returns C<dest>. Useful
1960 for encouraging compilers to tail-call
1963 =for apidoc Am|void|Zero|void* dest|int nitems|type
1965 The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1966 destination, C<nitems> is the number of items, and C<type> is the type.
1968 =for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1970 Like C<Zero> but returns dest. Useful
1971 for encouraging compilers to tail-call
1974 =for apidoc Am|void|StructCopy|type *src|type *dest|type
1975 This is an architecture-independent macro to copy one structure to another.
1977 =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1979 Fill up memory with a byte pattern (a byte repeated over and over
1980 again) that hopefully catches attempts to access uninitialized memory.
1982 =for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1984 PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1986 =for apidoc Am|void|PoisonFree|void* dest|int nitems|type
1988 PoisonWith(0xEF) for catching access to freed memory.
1990 =for apidoc Am|void|Poison|void* dest|int nitems|type
1992 PoisonWith(0xEF) for catching access to freed memory.
1996 /* Maintained for backwards-compatibility only. Use newSV() instead. */
1998 #define NEWSV(x,len) newSV(len)
2001 #define MEM_SIZE_MAX ((MEM_SIZE)~0)
2004 #ifdef PERL_MALLOC_WRAP
2006 /* This expression will be constant-folded at compile time. It checks
2007 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2008 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2009 * As well as avoiding the need for a run-time check in some cases, it's
2010 * designed to avoid compiler warnings like:
2011 * comparison is always false due to limited range of data type
2012 * It's mathematically equivalent to
2013 * max(n) * sizeof(t) > MEM_SIZE_MAX
2016 # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
2017 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
2019 /* This is written in a slightly odd way to avoid various spurious
2020 * compiler warnings. We *want* to write the expression as
2021 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2022 * (for some compile-time constant C), but even when the LHS
2023 * constant-folds to false at compile-time, g++ insists on emitting
2024 * warnings about the RHS (e.g. "comparison is always false"), so instead
2027 * (cond ? n : X) > C
2029 * where X is a constant with X > C always false. Choosing a value for X
2030 * is tricky. If 0, some compilers will complain about 0 > C always being
2031 * false; if 1, Coverity complains when n happens to be the constant value
2032 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2033 * for X and hope that nothing else whines.
2036 # define _MEM_WRAP_WILL_WRAP(n,t) \
2037 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2038 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
2040 # define MEM_WRAP_CHECK(n,t) \
2041 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2042 && (croak_memory_wrap(),0))
2044 # define MEM_WRAP_CHECK_1(n,t,a) \
2045 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2046 && (Perl_croak_nocontext("%s",(a)),0))
2048 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
2050 #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)))
2053 #define MEM_WRAP_CHECK(n,t)
2054 #define MEM_WRAP_CHECK_1(n,t,a)
2055 #define MEM_WRAP_CHECK_2(n,t,a,b)
2056 #define MEM_WRAP_CHECK_(n,t)
2058 #define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
2064 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
2065 * go through functions, which are handy for debugging breakpoints, but
2066 * which more importantly get the immediate calling environment (file and
2067 * line number, and C function name if available) passed in. This info can
2068 * then be used for logging the calls, for which one gets a sample
2069 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
2072 * - not all memory allocs get logged, only those
2073 * that go through Newx() and derivatives (while all
2074 * Safefrees do get logged)
2075 * - __FILE__ and __LINE__ do not work everywhere
2076 * - __func__ or __FUNCTION__ even less so
2077 * - I think more goes on after the perlio frees but
2078 * the thing is that STDERR gets closed (as do all
2079 * the file descriptors)
2080 * - no deeper calling stack than the caller of the Newx()
2081 * or the kind, but do I look like a C reflection/introspection
2083 * - the function prototypes for the logging functions
2084 * probably should maybe be somewhere else than handy.h
2085 * - one could consider inlining (macrofying) the logging
2086 * for speed, but I am too lazy
2087 * - one could imagine recording the allocations in a hash,
2088 * (keyed by the allocation address?), and maintain that
2089 * through reallocs and frees, but how to do that without
2090 * any News() happening...?
2091 * - lots of -Ddefines to get useful/controllable output
2092 * - lots of ENV reads
2096 # ifndef PERL_MEM_LOG_NOIMPL
2105 # if defined(PERL_IN_SV_C) /* those are only used in sv.c */
2106 void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2107 void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2114 #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2115 #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
2116 #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
2119 #ifndef MEM_LOG_ALLOC
2120 #define MEM_LOG_ALLOC(n,t,a) (a)
2122 #ifndef MEM_LOG_REALLOC
2123 #define MEM_LOG_REALLOC(n,t,v,a) (a)
2125 #ifndef MEM_LOG_FREE
2126 #define MEM_LOG_FREE(a) (a)
2129 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2130 #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2131 #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
2134 /* pre 5.9.x compatibility */
2135 #define New(x,v,n,t) Newx(v,n,t)
2136 #define Newc(x,v,n,t,c) Newxc(v,n,t,c)
2137 #define Newz(x,v,n,t) Newxz(v,n,t)
2140 #define Renew(v,n,t) \
2141 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2142 #define Renewc(v,n,t,c) \
2143 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2146 #define Safefree(d) \
2147 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
2149 #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
2152 #define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2153 #define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2154 #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
2156 #define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2157 #define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2159 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
2161 /* Using bzero(), which returns void. */
2162 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
2165 #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2166 #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2167 #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2168 #define Poison(d,n,t) PoisonFree(d,n,t)
2171 # define PERL_POISON_EXPR(x) x
2173 # define PERL_POISON_EXPR(x)
2176 #ifdef USE_STRUCT_COPY
2177 #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2179 #define StructCopy(s,d,t) Copy(s,d,1,t)
2182 /* C_ARRAY_LENGTH is the number of elements in the C array (so you
2183 * want your zero-based indices to be less than but not equal to).
2185 * C_ARRAY_END is one past the last: half-open/half-closed range,
2186 * not last-inclusive range. */
2187 #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
2188 #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
2192 # define Perl_va_copy(s, d) va_copy(d, s)
2194 # if defined(__va_copy)
2195 # define Perl_va_copy(s, d) __va_copy(d, s)
2197 # define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2202 /* convenience debug macros */
2204 #define pTHX_FORMAT "Perl interpreter: 0x%p"
2205 #define pTHX__FORMAT ", Perl interpreter: 0x%p"
2206 #define pTHX_VALUE_ (void *)my_perl,
2207 #define pTHX_VALUE (void *)my_perl
2208 #define pTHX__VALUE_ ,(void *)my_perl,
2209 #define pTHX__VALUE ,(void *)my_perl
2212 #define pTHX__FORMAT
2215 #define pTHX__VALUE_
2217 #endif /* USE_ITHREADS */
2219 /* Perl_deprecate was not part of the public API, and did not have a deprecate()
2220 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2221 CPAN::Unpack show any users outside the core. */
2223 # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2224 "Use of " s " is deprecated")
2227 /* Internal macros to deal with gids and uids */
2230 # if Uid_t_size > IVSIZE
2231 # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2232 # define SvUID(sv) SvNV(sv)
2234 # if Uid_t_sign <= 0
2235 # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2236 # define SvUID(sv) SvIV(sv)
2238 # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2239 # define SvUID(sv) SvUV(sv)
2241 # endif /* Uid_t_size */
2243 # if Gid_t_size > IVSIZE
2244 # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2245 # define SvGID(sv) SvNV(sv)
2247 # if Gid_t_sign <= 0
2248 # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2249 # define SvGID(sv) SvIV(sv)
2251 # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2252 # define SvGID(sv) SvUV(sv)
2254 # endif /* Gid_t_size */
2258 #endif /* HANDY_H */
2261 * ex: set ts=8 sts=4 sw=4 et:
https://perl5.git.perl.org / perl5.git / blob