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_safe> is like C<isFOO_uvchr>, but is used for UTF-8
569 encoded strings. Each call classifies one character, even if the string
570 contains many. This variant takes two parameters. The first, C<p>, is a
571 pointer to the first byte of the character to be classified. (Recall that it
572 may take more than one byte to represent a character in UTF-8 strings.) The
573 second parameter, C<e>, points to anywhere in the string beyond the first
574 character, up to one byte past the end of the entire string. The suffix
575 C<_safe> in the function's name indicates that it will not attempt to read
576 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
577 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
578 character is malformed in some way, the program may croak, or the function may
579 return FALSE, at the discretion of the implementation, and subject to change in
582 Variant C<isFOO_utf8> is like C<isFOO_utf8_safe>, but takes just a single
583 parameter, C<p>, which has the same meaning as the corresponding parameter does
584 in C<isFOO_utf8_safe>. The function therefore can't check if it is reading
585 beyond the end of the string.
587 Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
588 result is based on the current locale, which is what C<LC> in the name stands
589 for. If Perl can determine that the current locale is a UTF-8 locale, it uses
590 the published Unicode rules; otherwise, it uses the C library function that
591 gives the named classification. For example, C<isDIGIT_LC()> when not in a
592 UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
593 returned if the input won't fit into an octet. On some platforms where the C
594 library function is known to be defective, Perl changes its result to follow
595 the POSIX standard's rules.
597 Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
598 returns the same as C<isFOO_LC> for input code points less than 256, and
599 returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
601 Variant C<isFOO_LC_utf8_safe> is like C<isFOO_LC_uvchr>, but is used for UTF-8
602 encoded strings. Each call classifies one character, even if the string
603 contains many. This variant takes two parameters. The first, C<p>, is a
604 pointer to the first byte of the character to be classified. (Recall that it
605 may take more than one byte to represent a character in UTF-8 strings.) The
606 second parameter, C<e>, points to anywhere in the string beyond the first
607 character, up to one byte past the end of the entire string. The suffix
608 C<_safe> in the function's name indicates that it will not attempt to read
609 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
610 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
611 character is malformed in some way, the program may croak, or the function may
612 return FALSE, at the discretion of the implementation, and subject to change in
615 Variant C<isFOO_LC_utf8> is like C<isFOO_LC_utf8_safe>, but takes just a single
616 parameter, C<p>, which has the same meaning as the corresponding parameter does
617 in C<isFOO_LC_utf8_safe>. The function therefore can't check if it is reading
618 beyond the end of the string.
620 =for apidoc Am|bool|isALPHA|char ch
621 Returns a boolean indicating whether the specified character is an
622 alphabetic character, analogous to C<m/[[:alpha:]]/>.
623 See the L<top of this section|/Character classification> for an explanation of
625 C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8_safe>,
626 C<isALPHA_LC>, C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8_safe>.
628 =for apidoc Am|bool|isALPHANUMERIC|char ch
629 Returns a boolean indicating whether the specified character is a either an
630 alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
631 See the L<top of this section|/Character classification> for an explanation of
633 C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
634 C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>,
635 and C<isALPHANUMERIC_LC_utf8_safe>.
637 =for apidoc Am|bool|isASCII|char ch
638 Returns a boolean indicating whether the specified character is one of the 128
639 characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
640 On non-ASCII platforms, it returns TRUE iff this
641 character corresponds to an ASCII character. Variants C<isASCII_A()> and
642 C<isASCII_L1()> are identical to C<isASCII()>.
643 See the L<top of this section|/Character classification> for an explanation of
645 C<isASCII_uvchr>, C<isASCII_utf8_safe>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
646 C<isASCII_LC_utf8_safe>. Note, however, that some platforms do not have the C
647 library routine C<isascii()>. In these cases, the variants whose names contain
648 C<LC> are the same as the corresponding ones without.
650 Also note, that because all ASCII characters are UTF-8 invariant (meaning they
651 have the exact same representation (always a single byte) whether encoded in
652 UTF-8 or not), C<isASCII> will give the correct results when called with any
653 byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8_safe>
654 will work properly on any string encoded or not in UTF-8.
656 =for apidoc Am|bool|isBLANK|char ch
657 Returns a boolean indicating whether the specified character is a
658 character considered to be a blank, analogous to C<m/[[:blank:]]/>.
659 See the L<top of this section|/Character classification> for an explanation of
661 C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8_safe>,
662 C<isBLANK_LC>, C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8_safe>. Note,
663 however, that some platforms do not have the C library routine
664 C<isblank()>. In these cases, the variants whose names contain C<LC> are
665 the same as the corresponding ones without.
667 =for apidoc Am|bool|isCNTRL|char ch
668 Returns a boolean indicating whether the specified character is a
669 control character, analogous to C<m/[[:cntrl:]]/>.
670 See the L<top of this section|/Character classification> for an explanation of
672 C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8_safe>,
673 C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8_safe> On EBCDIC
674 platforms, you almost always want to use the C<isCNTRL_L1> variant.
676 =for apidoc Am|bool|isDIGIT|char ch
677 Returns a boolean indicating whether the specified character is a
678 digit, analogous to C<m/[[:digit:]]/>.
679 Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
680 See the L<top of this section|/Character classification> for an explanation of
682 C<isDIGIT_uvchr>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
683 C<isDIGIT_LC_utf8_safe>.
685 =for apidoc Am|bool|isGRAPH|char ch
686 Returns a boolean indicating whether the specified character is a
687 graphic character, analogous to C<m/[[:graph:]]/>.
688 See the L<top of this section|/Character classification> for an explanation of
689 variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8_safe>,
690 C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8_safe>.
692 =for apidoc Am|bool|isLOWER|char ch
693 Returns a boolean indicating whether the specified character is a
694 lowercase character, analogous to C<m/[[:lower:]]/>.
695 See the L<top of this section|/Character classification> for an explanation of
697 C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8_safe>,
698 C<isLOWER_LC>, C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8_safe>.
700 =for apidoc Am|bool|isOCTAL|char ch
701 Returns a boolean indicating whether the specified character is an
703 The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
706 =for apidoc Am|bool|isPUNCT|char ch
707 Returns a boolean indicating whether the specified character is a
708 punctuation character, analogous to C<m/[[:punct:]]/>.
709 Note that the definition of what is punctuation isn't as
710 straightforward as one might desire. See L<perlrecharclass/POSIX Character
711 Classes> for details.
712 See the L<top of this section|/Character classification> for an explanation of
713 variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8_safe>,
714 C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8_safe>.
716 =for apidoc Am|bool|isSPACE|char ch
717 Returns a boolean indicating whether the specified character is a
718 whitespace character. This is analogous
719 to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
720 this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
721 locale forms of this macro (the ones with C<LC> in their names) matched
722 precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
723 in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
724 (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
725 See the L<top of this section|/Character classification> for an explanation of
727 C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8_safe>,
728 C<isSPACE_LC>, C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8_safe>.
730 =for apidoc Am|bool|isPSXSPC|char ch
731 (short for Posix Space)
732 Starting in 5.18, this is identical in all its forms to the
733 corresponding C<isSPACE()> macros.
734 The locale forms of this macro are identical to their corresponding
735 C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
736 non-locale forms differ from their C<isSPACE()> forms only in that the
737 C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
738 Otherwise they are identical. Thus this macro is analogous to what
739 C<m/[[:space:]]/> matches in a regular expression.
740 See the L<top of this section|/Character classification> for an explanation of
741 variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8_safe>,
742 C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8_safe>.
744 =for apidoc Am|bool|isUPPER|char ch
745 Returns a boolean indicating whether the specified character is an
746 uppercase character, analogous to C<m/[[:upper:]]/>.
747 See the L<top of this section|/Character classification> for an explanation of
748 variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8_safe>,
749 C<isUPPER_LC>, C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8_safe>.
751 =for apidoc Am|bool|isPRINT|char ch
752 Returns a boolean indicating whether the specified character is a
753 printable character, analogous to C<m/[[:print:]]/>.
754 See the L<top of this section|/Character classification> for an explanation of
756 C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8_safe>,
757 C<isPRINT_LC>, C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8_safe>.
759 =for apidoc Am|bool|isWORDCHAR|char ch
760 Returns a boolean indicating whether the specified character is a character
761 that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
762 in a regular expression. A word character is an alphabetic character, a
763 decimal digit, a connecting punctuation character (such as an underscore), or
764 a "mark" character that attaches to one of those (like some sort of accent).
765 C<isALNUM()> is a synonym provided for backward compatibility, even though a
766 word character includes more than the standard C language meaning of
768 See the L<top of this section|/Character classification> for an explanation of
769 variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and
770 C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and
771 C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
772 include the platform's native underscore.
774 =for apidoc Am|bool|isXDIGIT|char ch
775 Returns a boolean indicating whether the specified character is a hexadecimal
776 digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
777 and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
778 See the L<top of this section|/Character classification> for an explanation of
780 C<isXDIGIT_uvchr>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>,
781 and C<isXDIGIT_LC_utf8_safe>.
783 =for apidoc Am|bool|isIDFIRST|char ch
784 Returns a boolean indicating whether the specified character can be the first
785 character of an identifier. This is very close to, but not quite the same as
786 the official Unicode property C<XID_Start>. The difference is that this
787 returns true only if the input character also matches L</isWORDCHAR>.
788 See the L<top of this section|/Character classification> for an explanation of
790 C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8_safe>,
791 C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8_safe>.
793 =for apidoc Am|bool|isIDCONT|char ch
794 Returns a boolean indicating whether the specified character can be the
795 second or succeeding character of an identifier. This is very close to, but
796 not quite the same as the official Unicode property C<XID_Continue>. The
797 difference is that this returns true only if the input character also matches
798 L</isWORDCHAR>. See the L<top of this section|/Character classification> for
800 explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
801 C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
802 C<isIDCONT_LC_utf8_safe>.
804 =head1 Miscellaneous Functions
806 =for apidoc Am|U8|READ_XDIGIT|char str*
807 Returns the value of an ASCII-range hex digit and advances the string pointer.
808 Behaviour is only well defined when isXDIGIT(*str) is true.
810 =head1 Character case changing
811 Perl uses "full" Unicode case mappings. This means that converting a single
812 character to another case may result in a sequence of more than one character.
813 For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
814 character sequence C<SS>. This presents some complications The lowercase of
815 all characters in the range 0..255 is a single character, and thus
816 C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
817 return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
818 an API that does allow every possible legal result to be returned.) Likewise
819 no other function that is crippled by not being able to give the correct
820 results for the full range of possible inputs has been implemented here.
822 =for apidoc Am|U8|toUPPER|U8 ch
823 Converts the specified character to uppercase. If the input is anything but an
824 ASCII lowercase character, that input character itself is returned. Variant
825 C<toUPPER_A> is equivalent.
827 =for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
828 Converts the code point C<cp> to its uppercase version, and
829 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
830 point is interpreted as native if less than 256; otherwise as Unicode. Note
831 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
832 bytes since the uppercase version may be longer than the original character.
834 The first code point of the uppercased version is returned
835 (but note, as explained at L<the top of this section|/Character case
836 changing>, that there may be more.)
838 =for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
839 Converts the UTF-8 encoded character at C<p> to its uppercase version, and
840 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
841 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
842 bytes since the uppercase version may be longer than the original character.
844 The first code point of the uppercased version is returned
845 (but note, as explained at L<the top of this section|/Character case
846 changing>, that there may be more).
848 The input character at C<p> is assumed to be well-formed.
850 =for apidoc Am|U8|toFOLD|U8 ch
851 Converts the specified character to foldcase. If the input is anything but an
852 ASCII uppercase character, that input character itself is returned. Variant
853 C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
854 Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
856 =for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
857 Converts the code point C<cp> to its foldcase version, and
858 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
859 point is interpreted as native if less than 256; otherwise as Unicode. Note
860 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
861 bytes since the foldcase version may be longer than the original character.
863 The first code point of the foldcased version is returned
864 (but note, as explained at L<the top of this section|/Character case
865 changing>, that there may be more).
867 =for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
868 Converts the UTF-8 encoded character at C<p> to its foldcase version, and
869 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
870 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
871 bytes since the foldcase version may be longer than the original character.
873 The first code point of the foldcased version is returned
874 (but note, as explained at L<the top of this section|/Character case
875 changing>, that there may be more).
877 The input character at C<p> is assumed to be well-formed.
879 =for apidoc Am|U8|toLOWER|U8 ch
880 Converts the specified character to lowercase. If the input is anything but an
881 ASCII uppercase character, that input character itself is returned. Variant
882 C<toLOWER_A> is equivalent.
884 =for apidoc Am|U8|toLOWER_L1|U8 ch
885 Converts the specified Latin1 character to lowercase. The results are
886 undefined if the input doesn't fit in a byte.
888 =for apidoc Am|U8|toLOWER_LC|U8 ch
889 Converts the specified character to lowercase using the current locale's rules,
890 if possible; otherwise returns the input character itself.
892 =for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
893 Converts the code point C<cp> to its lowercase version, and
894 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
895 point is interpreted as native if less than 256; otherwise as Unicode. Note
896 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
897 bytes since the lowercase version may be longer than the original character.
899 The first code point of the lowercased version is returned
900 (but note, as explained at L<the top of this section|/Character case
901 changing>, that there may be more).
903 =for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
904 Converts the UTF-8 encoded character at C<p> to its lowercase version, and
905 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
906 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
907 bytes since the lowercase version may be longer than the original character.
909 The first code point of the lowercased version is returned
910 (but note, as explained at L<the top of this section|/Character case
911 changing>, that there may be more).
913 The input character at C<p> is assumed to be well-formed.
915 =for apidoc Am|U8|toTITLE|U8 ch
916 Converts the specified character to titlecase. If the input is anything but an
917 ASCII lowercase character, that input character itself is returned. Variant
918 C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
919 range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
920 not a concept used in locale handling, so there is no functionality for that.)
922 =for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
923 Converts the code point C<cp> to its titlecase version, and
924 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
925 point is interpreted as native if less than 256; otherwise as Unicode. Note
926 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
927 bytes since the titlecase version may be longer than the original character.
929 The first code point of the titlecased version is returned
930 (but note, as explained at L<the top of this section|/Character case
931 changing>, that there may be more).
933 =for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
934 Converts the UTF-8 encoded character at C<p> to its titlecase version, and
935 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
936 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
937 bytes since the titlecase version may be longer than the original character.
939 The first code point of the titlecased version is returned
940 (but note, as explained at L<the top of this section|/Character case
941 changing>, that there may be more).
943 The input character at C<p> is assumed to be well-formed.
947 XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
948 really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
950 Note that these macros are repeated in Devel::PPPort, so should also be
951 patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
955 /* Specify the widest unsigned type on the platform. Use U64TYPE because U64
956 * is known only in the perl core, and this macro can be called from outside
959 # define WIDEST_UTYPE U64TYPE
961 # define WIDEST_UTYPE U32
964 /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
965 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
966 * bits of information are lost even on a 64-bit machine, but to get the
967 * compiler to optimize it out if possible. This is because Configure makes
968 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
969 * sizeof() guarantees that this evaluates to a constant true at compile time.
971 * For Coverity, be always true, because otherwise Coverity thinks
972 * it finds several expressions that are always true, independent
973 * of operands. Well, they are, but that is kind of the point.
976 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
978 #define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
979 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
981 #define FITS_IN_8_BITS(c) (1)
986 /* The native libc isascii() et.al. functions return the wrong results
987 * on at least z/OS unless this is defined. */
988 # error _ALL_SOURCE should probably be defined
991 /* There is a simple definition of ASCII for ASCII platforms. But the
992 * EBCDIC one isn't so simple, so is defined using table look-up like the
993 * other macros below.
995 * The cast here is used instead of '(c) >= 0', because some compilers emit
996 * a warning that that test is always true when the parameter is an
997 * unsigned type. khw supposes that it could be written as
998 * && ((c) == '\0' || (c) > 0)
999 * to avoid the message, but the cast will likely avoid extra branches even
1000 * with stupid compilers.
1002 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1004 # define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
1007 /* Take the eight possible bit patterns of the lower 3 bits and you get the
1008 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1009 * can be ignored. If the rest match '0', we have an octal */
1010 #define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
1012 #ifdef H_PERL /* If have access to perl.h, lookup in its table */
1014 /* Character class numbers. For internal core Perl use only. The ones less
1015 * than 32 are used in PL_charclass[] and the ones up through the one that
1016 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1017 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1018 * their actual definitions are here. If that file has a name not used here,
1021 * The first group of these is ordered in what I (khw) estimate to be the
1022 * frequency of their use. This gives a slight edge to exiting a loop earlier
1023 * (in reginclass() in regexec.c) */
1024 # define _CC_WORDCHAR 0 /* \w and [:word:] */
1025 # define _CC_DIGIT 1 /* \d and [:digit:] */
1026 # define _CC_ALPHA 2 /* [:alpha:] */
1027 # define _CC_LOWER 3 /* [:lower:] */
1028 # define _CC_UPPER 4 /* [:upper:] */
1029 # define _CC_PUNCT 5 /* [:punct:] */
1030 # define _CC_PRINT 6 /* [:print:] */
1031 # define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1032 # define _CC_GRAPH 8 /* [:graph:] */
1033 # define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1035 #define _FIRST_NON_SWASH_CC 10
1036 /* The character classes above are implemented with swashes. The second group
1037 * (just below) contains the ones implemented without. These are also sorted
1038 * in rough order of the frequency of their use, except that \v should be last,
1039 * as it isn't a real Posix character class, and some (small) inefficiencies in
1040 * regular expression handling would be introduced by putting it in the middle
1041 * of those that are. Also, cntrl and ascii come after the others as it may be
1042 * useful to group these which have no members that match above Latin1, (or
1043 * above ASCII in the latter case) */
1045 # define _CC_SPACE 10 /* \s, [:space:] */
1046 # define _CC_BLANK 11 /* [:blank:] */
1047 # define _CC_XDIGIT 12 /* [:xdigit:] */
1048 # define _CC_CNTRL 13 /* [:cntrl:] */
1049 # define _CC_ASCII 14 /* [:ascii:] */
1050 # define _CC_VERTSPACE 15 /* \v */
1052 # define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1054 /* The members of the third group below do not need to be coordinated with data
1055 * structures in regcomp.[ch] and regexec.c. */
1056 # define _CC_IDFIRST 16
1057 # define _CC_CHARNAME_CONT 17
1058 # define _CC_NONLATIN1_FOLD 18
1059 # define _CC_NONLATIN1_SIMPLE_FOLD 19
1060 # define _CC_QUOTEMETA 20
1061 # define _CC_NON_FINAL_FOLD 21
1062 # define _CC_IS_IN_SOME_FOLD 22
1063 # define _CC_MNEMONIC_CNTRL 23
1065 /* This next group is only used on EBCDIC platforms, so theoretically could be
1066 * shared with something entirely different that's only on ASCII platforms */
1067 # define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1068 # define _CC_UTF8_IS_START 29
1069 # define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1070 # define _CC_UTF8_IS_CONTINUATION 31
1072 * If more bits are needed, one could add a second word for non-64bit
1073 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
1074 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1075 * is used only for optimization (as of this writing), and differs in the
1076 * Latin1 range from the ALPHA bit only in two relatively unimportant
1077 * characters: the masculine and feminine ordinal indicators, so removing it
1078 * would just cause /i regexes which match them to run less efficiently.
1079 * Similarly the EBCDIC-only bits are used just for speed, and could be
1080 * replaced by other means */
1082 #if defined(PERL_CORE) || defined(PERL_EXT)
1083 /* An enum version of the character class numbers, to help compilers
1086 _CC_ENUM_ALPHA = _CC_ALPHA,
1087 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1088 _CC_ENUM_ASCII = _CC_ASCII,
1089 _CC_ENUM_BLANK = _CC_BLANK,
1090 _CC_ENUM_CASED = _CC_CASED,
1091 _CC_ENUM_CNTRL = _CC_CNTRL,
1092 _CC_ENUM_DIGIT = _CC_DIGIT,
1093 _CC_ENUM_GRAPH = _CC_GRAPH,
1094 _CC_ENUM_LOWER = _CC_LOWER,
1095 _CC_ENUM_PRINT = _CC_PRINT,
1096 _CC_ENUM_PUNCT = _CC_PUNCT,
1097 _CC_ENUM_SPACE = _CC_SPACE,
1098 _CC_ENUM_UPPER = _CC_UPPER,
1099 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
1100 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
1101 _CC_ENUM_XDIGIT = _CC_XDIGIT
1102 } _char_class_number;
1105 #define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
1106 #define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
1108 #if defined(PERL_IN_UTF8_C) \
1109 || defined(PERL_IN_REGCOMP_C) \
1110 || defined(PERL_IN_REGEXEC_C)
1111 # if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
1112 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1113 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
1114 #error Need to adjust order of swash_property_names[]
1117 /* This is declared static in each of the few files that this is #defined for
1118 * to keep them from being publicly accessible. Hence there is a small amount
1119 * of wasted space */
1121 static const char* const swash_property_names[] = {
1137 EXTCONST U32 PL_charclass[] = {
1138 # include "l1_char_class_tab.h"
1141 # else /* ! DOINIT */
1142 EXTCONST U32 PL_charclass[];
1146 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
1147 # define _CC_mask(classnum) (1U << (classnum))
1149 /* For internal core Perl use only: the base macro for defining macros like
1151 # define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
1152 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
1154 /* The mask for the _A versions of the macros; it just adds in the bit for
1156 # define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1158 /* For internal core Perl use only: the base macro for defining macros like
1159 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1160 * the ASCII bit are present */
1161 # define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1162 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1163 == _CC_mask_A(classnum)))
1165 # define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1166 # define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
1167 # define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1168 # define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
1169 # define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
1170 # define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1171 # define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1172 # define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1173 # define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1174 # define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1175 # define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1176 # define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
1177 # define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1179 # define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
1180 # define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1181 # define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1182 # define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1184 /* continuation character for legal NAME in \N{NAME} */
1185 # define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1187 # define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1188 # define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1189 # define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1190 # define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1191 # define isPSXSPC_L1(c) isSPACE_L1(c)
1192 # define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1193 # define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1194 # define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1195 # define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1196 # define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
1199 # define isASCII(c) _generic_isCC(c, _CC_ASCII)
1202 /* Participates in a single-character fold with a character above 255 */
1203 # 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)))
1205 /* Like the above, but also can be part of a multi-char fold */
1206 # 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)))
1208 # define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
1209 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1210 _generic_isCC(c, _CC_NON_FINAL_FOLD)
1211 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1212 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1213 # define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1214 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1215 #else /* else we don't have perl.h H_PERL */
1217 /* If we don't have perl.h, we are compiling a utility program. Below we
1218 * hard-code various macro definitions that wouldn't otherwise be available
1219 * to it. Most are coded based on first principles. These are written to
1220 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
1221 # define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1222 # define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1223 # define isSPACE_A(c) (isBLANK_A(c) \
1228 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1229 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1230 * performance (if optimization isn't on), and allow the same code to be
1231 * used for both platform types */
1232 # define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1234 || ((c) >= 'j' && (c) <= 'r') \
1236 # define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1238 || ((c) >= 'J' && (c) <= 'R') \
1240 # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1241 # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
1242 # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
1243 # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1244 # define isXDIGIT_A(c) (isDIGIT_A(c) \
1245 || ((c) >= 'a' && (c) <= 'f') \
1246 || ((c) <= 'F' && (c) >= 'A'))
1247 # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1248 || (c) == '#' || (c) == '$' || (c) == '%' \
1249 || (c) == '&' || (c) == '\'' || (c) == '(' \
1250 || (c) == ')' || (c) == '*' || (c) == '+' \
1251 || (c) == ',' || (c) == '.' || (c) == '/' \
1252 || (c) == ':' || (c) == ';' || (c) == '<' \
1253 || (c) == '=' || (c) == '>' || (c) == '?' \
1254 || (c) == '@' || (c) == '[' || (c) == '\\' \
1255 || (c) == ']' || (c) == '^' || (c) == '_' \
1256 || (c) == '`' || (c) == '{' || (c) == '|' \
1257 || (c) == '}' || (c) == '~')
1258 # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1259 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1262 /* The below is accurate for the 3 EBCDIC code pages traditionally
1263 * supported by perl. The only difference between them in the controls
1264 * is the position of \n, and that is represented symbolically below */
1265 # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1266 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1267 || (c) == '\t' || (c) == '\v' \
1268 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1269 || (c) == 7 /* U+7F DEL */ \
1270 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1271 /* DLE, DC[1-3] */ \
1272 || (c) == 0x18 /* U+18 CAN */ \
1273 || (c) == 0x19 /* U+19 EOM */ \
1274 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1275 || (c) == 0x26 /* U+17 ETB */ \
1276 || (c) == 0x27 /* U+1B ESC */ \
1277 || (c) == 0x2D /* U+05 ENQ */ \
1278 || (c) == 0x2E /* U+06 ACK */ \
1279 || (c) == 0x32 /* U+16 SYN */ \
1280 || (c) == 0x37 /* U+04 EOT */ \
1281 || (c) == 0x3C /* U+14 DC4 */ \
1282 || (c) == 0x3D /* U+15 NAK */ \
1283 || (c) == 0x3F)/* U+1A SUB */
1284 # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1285 # else /* isASCII is already defined for ASCII platforms, so can use that to
1287 # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
1290 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
1291 * during debugging, and it seems best to keep them. We may be called
1292 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1293 * do anything anyway, so make it not a problem */
1294 # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1295 # define NATIVE_TO_LATIN1(ch) (ch)
1297 # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1298 # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1299 # define isBLANK_L1(c) (isBLANK_A(c) \
1300 || (FITS_IN_8_BITS(c) \
1301 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1302 # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1303 # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1304 # define isLOWER_L1(c) (isLOWER_A(c) \
1305 || (FITS_IN_8_BITS(c) \
1306 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1307 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1308 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1309 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1310 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1311 # define isPRINT_L1(c) (isPRINT_A(c) \
1312 || (FITS_IN_8_BITS(c) \
1313 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1314 # define isPUNCT_L1(c) (isPUNCT_A(c) \
1315 || (FITS_IN_8_BITS(c) \
1316 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1317 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1318 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1319 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1320 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1321 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1322 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1323 # define isSPACE_L1(c) (isSPACE_A(c) \
1324 || (FITS_IN_8_BITS(c) \
1325 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1326 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1327 # define isUPPER_L1(c) (isUPPER_A(c) \
1328 || (FITS_IN_8_BITS(c) \
1329 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1330 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1331 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1332 # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1333 # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1334 # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1339 /* The following are not fully accurate in the above-ASCII range. I (khw)
1340 * don't think it's necessary to be so for the purposes where this gets
1342 # define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1343 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1345 /* And these aren't accurate at all. They are useful only for above
1346 * Latin1, which utilities and bootstrapping don't deal with */
1347 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
1348 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1349 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1351 /* Many of the macros later in this file are defined in terms of these. By
1352 * implementing them with a function, which converts the class number into
1353 * a call to the desired macro, all of the later ones work. However, that
1354 * function won't be actually defined when building a utility program (no
1355 * perl.h), and so a compiler error will be generated if one is attempted
1356 * to be used. And the above-Latin1 code points require Unicode tables to
1357 * be present, something unlikely to be the case when bootstrapping */
1358 # define _generic_isCC(c, classnum) \
1359 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1360 # define _generic_isCC_A(c, classnum) \
1361 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
1362 #endif /* End of no perl.h H_PERL */
1364 #define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1365 #define isALPHA(c) isALPHA_A(c)
1366 #define isASCII_A(c) isASCII(c)
1367 #define isASCII_L1(c) isASCII(c)
1368 #define isBLANK(c) isBLANK_A(c)
1369 #define isCNTRL(c) isCNTRL_A(c)
1370 #define isDIGIT(c) isDIGIT_A(c)
1371 #define isGRAPH(c) isGRAPH_A(c)
1372 #define isIDFIRST(c) isIDFIRST_A(c)
1373 #define isLOWER(c) isLOWER_A(c)
1374 #define isPRINT(c) isPRINT_A(c)
1375 #define isPSXSPC_A(c) isSPACE_A(c)
1376 #define isPSXSPC(c) isPSXSPC_A(c)
1377 #define isPSXSPC_L1(c) isSPACE_L1(c)
1378 #define isPUNCT(c) isPUNCT_A(c)
1379 #define isSPACE(c) isSPACE_A(c)
1380 #define isUPPER(c) isUPPER_A(c)
1381 #define isWORDCHAR(c) isWORDCHAR_A(c)
1382 #define isXDIGIT(c) isXDIGIT_A(c)
1384 /* ASCII casing. These could also be written as
1385 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1386 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1387 which uses table lookup and mask instead of subtraction. (This would
1388 work because the _MOD does not apply in the ASCII range) */
1389 #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1390 #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
1392 /* In the ASCII range, these are equivalent to what they're here defined to be.
1393 * But by creating these definitions, other code doesn't have to be aware of
1395 #define toFOLD(c) toLOWER(c)
1396 #define toTITLE(c) toUPPER(c)
1398 #define toLOWER_A(c) toLOWER(c)
1399 #define toUPPER_A(c) toUPPER(c)
1400 #define toFOLD_A(c) toFOLD(c)
1401 #define toTITLE_A(c) toTITLE(c)
1403 /* Use table lookup for speed; returns the input itself if is out-of-range */
1404 #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
1406 : PL_latin1_lc[ (U8) (c) ])
1407 #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1409 /* Modified uc. Is correct uc except for three non-ascii chars which are
1410 * all mapped to one of them, and these need special handling; returns the
1411 * input itself if is out-of-range */
1412 #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
1414 : PL_mod_latin1_uc[ (U8) (c) ])
1415 #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
1417 /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1419 /* For internal core Perl use only: the base macro for defining macros like
1420 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
1421 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1422 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1423 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1424 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1425 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1426 * code points under locale rules, use the macros based on _generic_LC_uvchr
1427 * instead of this. */
1428 #define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1429 (! FITS_IN_8_BITS(c) \
1431 : IN_UTF8_CTYPE_LOCALE \
1432 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
1435 /* For internal core Perl use only: a helper macro for defining macros like
1436 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1437 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1438 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1439 * _generic_LC_base for more info */
1440 #define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1441 _generic_LC_base(c,utf8_locale_classnum, \
1442 non_utf8_func( (_LC_CAST) (c)))
1444 /* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1445 * 'c' is the platform's native underscore character */
1446 #define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1447 _generic_LC_base(c, utf8_locale_classnum, \
1448 (non_utf8_func( (_LC_CAST) (c)) \
1449 || (char)(c) == '_'))
1451 /* These next three are also for internal core Perl use only: case-change
1453 #define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1455 : (IN_UTF8_CTYPE_LOCALE) \
1456 ? PL_latin1_lc[ (U8) (c) ] \
1457 : (cast)function((cast)(c)))
1459 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1460 * returns a single value, so can't adequately return the upper case of LATIN
1461 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1462 * values "SS"); instead it asserts against that under DEBUGGING, and
1463 * otherwise returns its input */
1464 #define _generic_toUPPER_LC(c, function, cast) \
1465 (! FITS_IN_8_BITS(c) \
1467 : ((! IN_UTF8_CTYPE_LOCALE) \
1468 ? (cast)function((cast)(c)) \
1469 : ((((U8)(c)) == MICRO_SIGN) \
1470 ? GREEK_CAPITAL_LETTER_MU \
1471 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1472 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1473 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1474 ? (__ASSERT_(0) (c)) \
1475 : PL_mod_latin1_uc[ (U8) (c) ])))))
1477 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1478 * returns a single value, so can't adequately return the fold case of LATIN
1479 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1480 * values "ss"); instead it asserts against that under DEBUGGING, and
1481 * otherwise returns its input */
1482 #define _generic_toFOLD_LC(c, function, cast) \
1483 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
1484 ? GREEK_SMALL_LETTER_MU \
1485 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1486 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1487 _generic_toLOWER_LC(c, function, cast)))
1489 /* Use the libc versions for these if available. */
1490 #if defined(HAS_ISASCII)
1491 # define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1493 # define isASCII_LC(c) isASCII(c)
1496 #if defined(HAS_ISBLANK)
1497 # define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
1498 #else /* Unlike isASCII, varies if in a UTF-8 locale */
1499 # define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
1505 /* The Windows functions don't bother to follow the POSIX standard, which
1506 * for example says that something can't both be a printable and a control.
1507 * But Windows treats the \t control as a printable, and does such things
1508 * as making superscripts into both digits and punctuation. This tames
1509 * these flaws by assuming that the definitions of both controls and space
1510 * are correct, and then making sure that other definitions don't have
1511 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1512 * Not all possible weirdnesses are checked for, just the ones that were
1513 * detected on actual Microsoft code pages */
1515 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1516 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1518 # define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1519 && isALPHANUMERIC_LC(c))
1520 # define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1522 # define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1523 isALPHANUMERIC_LC(c))
1524 # define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1525 # define isIDFIRST_LC(c) (((c) == '_') \
1526 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1527 # define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1528 # define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1529 # define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1530 # define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1531 # define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1532 # define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1533 && isALPHANUMERIC_LC(c))
1535 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1536 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1537 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1539 #elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
1540 /* For most other platforms */
1542 # define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1543 # define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1544 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1545 # define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1546 # define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1547 # define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1548 # define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1549 # define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1550 # define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1551 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1552 # define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1553 # define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1554 # define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1557 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1558 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1559 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1561 #else /* The final fallback position */
1563 # define isALPHA_LC(c) (isascii(c) && isalpha(c))
1564 # define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1565 # define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1566 # define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1567 # define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1568 # define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1569 # define isLOWER_LC(c) (isascii(c) && islower(c))
1570 # define isPRINT_LC(c) (isascii(c) && isprint(c))
1571 # define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1572 # define isSPACE_LC(c) (isascii(c) && isspace(c))
1573 # define isUPPER_LC(c) (isascii(c) && isupper(c))
1574 # define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1575 # define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1577 # define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1578 # define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1579 # define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
1583 #define isIDCONT(c) isWORDCHAR(c)
1584 #define isIDCONT_A(c) isWORDCHAR_A(c)
1585 #define isIDCONT_L1(c) isWORDCHAR_L1(c)
1586 #define isIDCONT_LC(c) isWORDCHAR_LC(c)
1587 #define isPSXSPC_LC(c) isSPACE_LC(c)
1589 /* For internal core Perl use only: the base macros for defining macros like
1590 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1591 * number defined earlier in this file. _generic_uvchr() is used for POSIX
1592 * classes where there is a macro or function 'above_latin1' that takes the
1593 * single argument 'c' and returns the desired value. These exist for those
1594 * classes which have simple definitions, avoiding the overhead of a hash
1595 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
1596 * for classes where that overhead is faster than a direct lookup.
1597 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
1598 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1599 * have duplicate checks here, so could create versions of the macros that
1600 * don't, but experiments show that gcc optimizes them out anyway. */
1602 /* Note that all ignore 'use bytes' */
1603 #define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1604 ? _generic_isCC(c, classnum) \
1606 #define _generic_swash_uvchr(classnum, c) ((c) < 256 \
1607 ? _generic_isCC(c, classnum) \
1608 : _is_uni_FOO(classnum, c))
1609 #define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1610 #define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1611 #define isASCII_uvchr(c) isASCII(c)
1612 #define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1613 #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1614 #define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1615 #define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1616 #define isIDCONT_uvchr(c) _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1617 #define isIDFIRST_uvchr(c) _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
1618 #define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1619 #define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1621 #define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1622 #define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1623 #define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1625 #define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1626 #define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1627 #define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1628 #define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1630 #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1631 #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1632 #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1633 #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1635 /* For backwards compatibility, even though '_uni' should mean official Unicode
1636 * code points, in Perl it means native for those below 256 */
1637 #define isALPHA_uni(c) isALPHA_uvchr(c)
1638 #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1639 #define isASCII_uni(c) isASCII_uvchr(c)
1640 #define isBLANK_uni(c) isBLANK_uvchr(c)
1641 #define isCNTRL_uni(c) isCNTRL_uvchr(c)
1642 #define isDIGIT_uni(c) isDIGIT_uvchr(c)
1643 #define isGRAPH_uni(c) isGRAPH_uvchr(c)
1644 #define isIDCONT_uni(c) isIDCONT_uvchr(c)
1645 #define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1646 #define isLOWER_uni(c) isLOWER_uvchr(c)
1647 #define isPRINT_uni(c) isPRINT_uvchr(c)
1648 #define isPUNCT_uni(c) isPUNCT_uvchr(c)
1649 #define isSPACE_uni(c) isSPACE_uvchr(c)
1650 #define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1651 #define isUPPER_uni(c) isUPPER_uvchr(c)
1652 #define isVERTWS_uni(c) isVERTWS_uvchr(c)
1653 #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1654 #define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1655 #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1656 #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1657 #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1658 #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
1660 /* For internal core Perl use only: the base macros for defining macros like
1661 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1662 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
1663 * simple class definitions; the other for more complex. These are like
1664 * _generic_uvchr, so see it for more info. */
1665 #define _generic_LC_uvchr(latin1, above_latin1, c) \
1666 (c < 256 ? latin1(c) : above_latin1(c))
1667 #define _generic_LC_swash_uvchr(latin1, classnum, c) \
1668 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1670 #define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1671 #define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1672 _CC_ALPHANUMERIC, c)
1673 #define isASCII_LC_uvchr(c) isASCII_LC(c)
1674 #define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1675 is_HORIZWS_cp_high, c)
1676 #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
1677 #define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1678 #define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
1679 #define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
1680 _is_uni_perl_idcont, c)
1681 #define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
1682 _is_uni_perl_idstart, c)
1683 #define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1684 #define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
1685 #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
1686 #define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
1687 #define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
1688 is_XPERLSPACE_cp_high, c)
1689 #define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
1690 #define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
1692 #define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1693 is_XDIGIT_cp_high, c)
1695 #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
1697 /* For internal core Perl use only: the base macros for defining macros like
1698 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1699 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1700 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1701 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1702 * same representation whether utf8 or not. Note that it assumes that the utf8
1703 * has been validated, and ignores 'use bytes' */
1704 #define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
1705 ? _generic_isCC(*(p), classnum) \
1706 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1708 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
1713 /* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
1714 * they don't otherwise go out of their way to look for malformed UTF-8. If
1715 * they can return accurate results without knowing if the input is otherwise
1716 * malformed, they do so. For example isASCII is accurate in spite of any
1717 * non-length malformations because it looks only at a single byte. Likewise
1718 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
1719 * variant ones return FALSE. But, if the input has to be well-formed in order
1720 * for the results to be accurate, the macros will test and if malformed will
1721 * call a routine to die
1723 * Except for toke.c, the macros do assume that e > p, asserting that on
1724 * DEBUGGING builds. Much code that calls these depends on this being true,
1725 * for other reasons. toke.c is treated specially as using the regular
1726 * assertion breaks it in many ways. All strings that these operate on there
1727 * are supposed to have an extra NUL character at the end, so that *e = \0. A
1728 * bunch of code in toke.c assumes that this is true, so the assertion allows
1730 #ifdef PERL_IN_TOKE_C
1731 # define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
1733 # define _utf8_safe_assert(p,e) ((e) > (p))
1736 #define _generic_utf8_safe(classnum, p, e, above_latin1) \
1737 (__ASSERT_(_utf8_safe_assert(p, e)) \
1738 (UTF8_IS_INVARIANT(*(p))) \
1739 ? _generic_isCC(*(p), classnum) \
1740 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1741 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1742 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
1744 : (_force_out_malformed_utf8_message( \
1745 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1747 /* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1748 * 'above_latin1' can be a macro */
1749 #define _generic_func_utf8(classnum, above_latin1, p) \
1750 _generic_utf8(classnum, p, above_latin1(p))
1751 #define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
1752 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
1753 #define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1754 _generic_utf8_safe(classnum, p, e, \
1755 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1756 ? (_force_out_malformed_utf8_message( \
1757 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1759 /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
1760 * 'above_latin1' parameter */
1761 #define _generic_swash_utf8(classnum, p) \
1762 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
1763 #define _generic_swash_utf8_safe(classnum, p, e) \
1764 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e))
1766 /* Like the above, but should be used only when it is known that there are no
1767 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1768 * class is TRUE for. Hence it can skip the tests for this range.
1769 * 'above_latin1' should include its arguments */
1770 #define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
1771 (UTF8_IS_INVARIANT(*(p)) \
1772 ? _generic_isCC(*(p), classnum) \
1773 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
1777 #define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
1778 (__ASSERT_(_utf8_safe_assert(p, e)) \
1779 (UTF8_IS_INVARIANT(*(p))) \
1780 ? _generic_isCC(*(p), classnum) \
1781 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1782 ? 0 /* Note that doesn't check validity for latin1 */ \
1785 /* NOTE that some of these macros have very similar ones in regcharclass.h.
1786 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1787 * there, differing in name only by an underscore from the one here
1788 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1789 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1790 * points; the regcharclass.h ones are implemented as a series of
1791 * "if-else-if-else ..." */
1793 #define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1794 #define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1795 #define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
1796 utf8, the non-utf8 macro works
1798 #define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
1800 #define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e)
1801 #define isALPHANUMERIC_utf8_safe(p, e) \
1802 _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e)
1803 #define isASCII_utf8_safe(p, e) \
1804 /* Because ASCII is invariant under utf8, the non-utf8 macro \
1806 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
1807 #define isBLANK_utf8_safe(p, e) \
1808 _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
1811 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1812 * more efficient macro instead of the more general one */
1813 # define isCNTRL_utf8(p) isCNTRL_L1(*(p))
1814 # define isCNTRL_utf8_safe(p, e) \
1815 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p))
1817 # define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1818 # define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
1821 #define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1822 _is_utf8_FOO(_CC_DIGIT, p))
1823 #define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
1824 #define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
1825 _is_utf8_perl_idcont, p)
1826 #define isDIGIT_utf8_safe(p, e) \
1827 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
1828 _is_utf8_FOO_with_len(_CC_DIGIT, p, e))
1829 #define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e)
1830 #define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
1831 _is_utf8_perl_idcont_with_len, p, e)
1833 /* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1834 * IDFIRST is an alnum. See
1835 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
1836 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1837 * This used to be not the XID version, but we decided to go with the more
1838 * modern Unicode definition */
1839 #define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
1840 _is_utf8_perl_idstart, p)
1842 #define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1843 #define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1844 #define isPSXSPC_utf8(p) isSPACE_utf8(p)
1845 #define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1846 #define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1847 #define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1848 #define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1849 #define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1850 #define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
1852 #define isIDFIRST_utf8_safe(p, e) \
1853 _generic_func_utf8_safe(_CC_IDFIRST, \
1854 _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e))
1856 #define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e)
1857 #define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e)
1858 #define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
1859 #define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e)
1860 #define isSPACE_utf8_safe(p, e) \
1861 _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
1862 #define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e)
1863 #define isVERTWS_utf8_safe(p, e) \
1864 _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
1865 #define isWORDCHAR_utf8_safe(p, e) \
1866 _generic_swash_utf8_safe(_CC_WORDCHAR, p, e)
1867 #define isXDIGIT_utf8_safe(p, e) \
1868 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
1869 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1870 ? (_force_out_malformed_utf8_message( \
1871 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1872 : is_XDIGIT_high(p)))
1874 #define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
1875 #define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1876 #define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1877 #define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
1879 /* For internal core Perl use only: the base macros for defining macros like
1880 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1881 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1882 * 'macro' parameter */
1883 #define _generic_LC_utf8(macro, p, utf8) \
1884 (UTF8_IS_INVARIANT(*(p)) \
1886 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1887 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
1890 #define _generic_LC_swash_utf8(macro, classnum, p) \
1891 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1892 #define _generic_LC_func_utf8(macro, above_latin1, p) \
1893 _generic_LC_utf8(macro, p, above_latin1(p))
1895 #define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
1896 _CC_ALPHANUMERIC, p)
1897 #define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1898 #define isASCII_LC_utf8(p) isASCII_LC(*p)
1899 #define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1901 #define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1902 #define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1903 #define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1904 #define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1905 _is_utf8_perl_idcont, p)
1906 #define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1907 _is_utf8_perl_idstart, p)
1908 #define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1909 #define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1910 #define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1911 #define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1912 #define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1913 is_XPERLSPACE_high, p)
1914 #define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
1915 #define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1917 #define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1919 /* For internal core Perl use only: the base macros for defining macros like
1920 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
1921 * point in 'p' is within the 0-255 range, it uses locale rules from the
1922 * passed-in 'macro' parameter */
1923 #define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
1924 (__ASSERT_(_utf8_safe_assert(p, e)) \
1925 (UTF8_IS_INVARIANT(*(p))) \
1927 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1928 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1929 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
1930 : (_force_out_malformed_utf8_message( \
1931 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1934 #define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \
1935 _generic_LC_utf8_safe(macro, p, e, \
1936 _is_utf8_FOO_with_len(classnum, p, e))
1938 #define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
1939 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
1941 #define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1942 _generic_LC_utf8_safe(classnum, p, e, \
1943 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1944 ? (_force_out_malformed_utf8_message( \
1945 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1948 #define isALPHANUMERIC_LC_utf8_safe(p, e) \
1949 _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \
1950 _CC_ALPHANUMERIC, p, e)
1951 #define isALPHA_LC_utf8_safe(p, e) \
1952 _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
1953 #define isASCII_LC_utf8_safe(p, e) \
1954 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
1955 #define isBLANK_LC_utf8_safe(p, e) \
1956 _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
1957 #define isCNTRL_LC_utf8_safe(p, e) \
1958 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
1959 #define isDIGIT_LC_utf8_safe(p, e) \
1960 _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
1961 #define isGRAPH_LC_utf8_safe(p, e) \
1962 _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
1963 #define isIDCONT_LC_utf8_safe(p, e) \
1964 _generic_LC_func_utf8_safe(isIDCONT_LC, \
1965 _is_utf8_perl_idcont_with_len, p, e)
1966 #define isIDFIRST_LC_utf8_safe(p, e) \
1967 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
1968 _is_utf8_perl_idstart_with_len, p, e)
1969 #define isLOWER_LC_utf8_safe(p, e) \
1970 _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
1971 #define isPRINT_LC_utf8_safe(p, e) \
1972 _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
1973 #define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
1974 #define isPUNCT_LC_utf8_safe(p, e) \
1975 _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
1976 #define isSPACE_LC_utf8_safe(p, e) \
1977 _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
1978 #define isUPPER_LC_utf8_safe(p, e) \
1979 _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
1980 #define isWORDCHAR_LC_utf8_safe(p, e) \
1981 _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
1982 #define isXDIGIT_LC_utf8_safe(p, e) \
1983 _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
1985 /* Macros for backwards compatibility and for completeness when the ASCII and
1986 * Latin1 values are identical */
1987 #define isALPHAU(c) isALPHA_L1(c)
1988 #define isDIGIT_L1(c) isDIGIT_A(c)
1989 #define isOCTAL(c) isOCTAL_A(c)
1990 #define isOCTAL_L1(c) isOCTAL_A(c)
1991 #define isXDIGIT_L1(c) isXDIGIT_A(c)
1992 #define isALNUM(c) isWORDCHAR(c)
1993 #define isALNUMU(c) isWORDCHAR_L1(c)
1994 #define isALNUM_LC(c) isWORDCHAR_LC(c)
1995 #define isALNUM_uni(c) isWORDCHAR_uni(c)
1996 #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
1997 #define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1998 #define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1999 #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2000 #define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2001 #define isALNUMC(c) isALPHANUMERIC(c)
2002 #define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2003 #define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
2004 #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
2005 #define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
2006 #define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
2008 /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2009 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2010 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2011 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2012 * the outlier from the block that contains the other controls, just like
2013 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2014 * block. If it weren't special cased, it would yield a non-control.
2015 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2018 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2020 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2022 ? (UNLIKELY((c) == '?') \
2023 ? QUESTION_MARK_CTRL \
2024 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2025 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2027 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2030 /* Line numbers are unsigned, 32 bits. */
2032 #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
2034 /* Helpful alias for version prescan */
2035 #define is_LAX_VERSION(a,b) \
2036 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2038 #define is_STRICT_VERSION(a,b) \
2039 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2041 #define BADVERSION(a,b,c) \
2047 /* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2048 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2049 * which is advanced. The input is validated only by an assert() in DEBUGGING
2050 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2051 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2052 #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2055 #define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2059 /* Converts a character known to represent an octal digit (0-7) to its numeric
2060 * value. The input is validated only by an assert() in DEBUGGING builds. In
2061 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2062 #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2064 /* Efficiently returns a boolean as to if two native characters are equivalent
2065 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2066 * the ALPHA in the name is to remind you of that. This is asserted() in
2067 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2068 * works (on valid input) for both non- and UTF-8-encoded bytes.
2070 * When one of the inputs is a compile-time constant and gets folded by the
2071 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2072 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2073 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
2074 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2075 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2076 * just a single 0, in the bit position where the upper- and lowercase differ.
2078 #define isALPHA_FOLD_EQ(c1, c2) \
2079 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2080 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2081 #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2084 =head1 Memory Management
2086 =for apidoc Am|void|Newx|void* ptr|int nitems|type
2087 The XSUB-writer's interface to the C C<malloc> function.
2089 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2091 In 5.9.3, Newx() and friends replace the older New() API, and drops
2092 the first parameter, I<x>, a debug aid which allowed callers to identify
2093 themselves. This aid has been superseded by a new build option,
2094 PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
2095 there for use in XS modules supporting older perls.
2097 =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
2098 The XSUB-writer's interface to the C C<malloc> function, with
2099 cast. See also C<L</Newx>>.
2101 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2103 =for apidoc Am|void|Newxz|void* ptr|int nitems|type
2104 The XSUB-writer's interface to the C C<malloc> function. The allocated
2105 memory is zeroed with C<memzero>. See also C<L</Newx>>.
2107 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2109 =for apidoc Am|void|Renew|void* ptr|int nitems|type
2110 The XSUB-writer's interface to the C C<realloc> function.
2112 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2114 =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2115 The XSUB-writer's interface to the C C<realloc> function, with
2118 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2120 =for apidoc Am|void|Safefree|void* ptr
2121 The XSUB-writer's interface to the C C<free> function.
2123 This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
2125 =for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2126 The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
2127 source, C<dest> is the destination, C<nitems> is the number of items, and
2128 C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
2130 =for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
2131 Like C<Move> but returns C<dest>. Useful
2132 for encouraging compilers to tail-call
2135 =for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2136 The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
2137 source, C<dest> is the destination, C<nitems> is the number of items, and
2138 C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
2140 =for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2142 Like C<Copy> but returns C<dest>. Useful
2143 for encouraging compilers to tail-call
2146 =for apidoc Am|void|Zero|void* dest|int nitems|type
2148 The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2149 destination, C<nitems> is the number of items, and C<type> is the type.
2151 =for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2153 Like C<Zero> but returns dest. Useful
2154 for encouraging compilers to tail-call
2157 =for apidoc Am|void|StructCopy|type *src|type *dest|type
2158 This is an architecture-independent macro to copy one structure to another.
2160 =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2162 Fill up memory with a byte pattern (a byte repeated over and over
2163 again) that hopefully catches attempts to access uninitialized memory.
2165 =for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2167 PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2169 =for apidoc Am|void|PoisonFree|void* dest|int nitems|type
2171 PoisonWith(0xEF) for catching access to freed memory.
2173 =for apidoc Am|void|Poison|void* dest|int nitems|type
2175 PoisonWith(0xEF) for catching access to freed memory.
2179 /* Maintained for backwards-compatibility only. Use newSV() instead. */
2181 #define NEWSV(x,len) newSV(len)
2184 #define MEM_SIZE_MAX ((MEM_SIZE)~0)
2187 #ifdef PERL_MALLOC_WRAP
2189 /* This expression will be constant-folded at compile time. It checks
2190 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2191 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2192 * As well as avoiding the need for a run-time check in some cases, it's
2193 * designed to avoid compiler warnings like:
2194 * comparison is always false due to limited range of data type
2195 * It's mathematically equivalent to
2196 * max(n) * sizeof(t) > MEM_SIZE_MAX
2199 # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
2200 (8 * sizeof(n) + sizeof(t) > sizeof(MEM_SIZE))
2202 /* This is written in a slightly odd way to avoid various spurious
2203 * compiler warnings. We *want* to write the expression as
2204 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2205 * (for some compile-time constant C), but even when the LHS
2206 * constant-folds to false at compile-time, g++ insists on emitting
2207 * warnings about the RHS (e.g. "comparison is always false"), so instead
2210 * (cond ? n : X) > C
2212 * where X is a constant with X > C always false. Choosing a value for X
2213 * is tricky. If 0, some compilers will complain about 0 > C always being
2214 * false; if 1, Coverity complains when n happens to be the constant value
2215 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2216 * for X and hope that nothing else whines.
2219 # define _MEM_WRAP_WILL_WRAP(n,t) \
2220 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2221 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
2223 # define MEM_WRAP_CHECK(n,t) \
2224 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2225 && (croak_memory_wrap(),0))
2227 # define MEM_WRAP_CHECK_1(n,t,a) \
2228 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2229 && (Perl_croak_nocontext("%s",(a)),0))
2231 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
2233 #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)))
2236 #define MEM_WRAP_CHECK(n,t)
2237 #define MEM_WRAP_CHECK_1(n,t,a)
2238 #define MEM_WRAP_CHECK_2(n,t,a,b)
2239 #define MEM_WRAP_CHECK_(n,t)
2241 #define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
2247 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
2248 * go through functions, which are handy for debugging breakpoints, but
2249 * which more importantly get the immediate calling environment (file and
2250 * line number, and C function name if available) passed in. This info can
2251 * then be used for logging the calls, for which one gets a sample
2252 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
2255 * - not all memory allocs get logged, only those
2256 * that go through Newx() and derivatives (while all
2257 * Safefrees do get logged)
2258 * - __FILE__ and __LINE__ do not work everywhere
2259 * - __func__ or __FUNCTION__ even less so
2260 * - I think more goes on after the perlio frees but
2261 * the thing is that STDERR gets closed (as do all
2262 * the file descriptors)
2263 * - no deeper calling stack than the caller of the Newx()
2264 * or the kind, but do I look like a C reflection/introspection
2266 * - the function prototypes for the logging functions
2267 * probably should maybe be somewhere else than handy.h
2268 * - one could consider inlining (macrofying) the logging
2269 * for speed, but I am too lazy
2270 * - one could imagine recording the allocations in a hash,
2271 * (keyed by the allocation address?), and maintain that
2272 * through reallocs and frees, but how to do that without
2273 * any News() happening...?
2274 * - lots of -Ddefines to get useful/controllable output
2275 * - lots of ENV reads
2279 # ifndef PERL_MEM_LOG_NOIMPL
2288 # if defined(PERL_IN_SV_C) /* those are only used in sv.c */
2289 void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2290 void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2297 #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2298 #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
2299 #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
2302 #ifndef MEM_LOG_ALLOC
2303 #define MEM_LOG_ALLOC(n,t,a) (a)
2305 #ifndef MEM_LOG_REALLOC
2306 #define MEM_LOG_REALLOC(n,t,v,a) (a)
2308 #ifndef MEM_LOG_FREE
2309 #define MEM_LOG_FREE(a) (a)
2312 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2313 #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2314 #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
2317 /* pre 5.9.x compatibility */
2318 #define New(x,v,n,t) Newx(v,n,t)
2319 #define Newc(x,v,n,t,c) Newxc(v,n,t,c)
2320 #define Newz(x,v,n,t) Newxz(v,n,t)
2323 #define Renew(v,n,t) \
2324 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2325 #define Renewc(v,n,t,c) \
2326 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2329 #define Safefree(d) \
2330 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
2332 #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
2335 #define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2336 #define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2337 #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
2339 #define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2340 #define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2342 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
2344 /* Using bzero(), which returns void. */
2345 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
2348 #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2349 #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2350 #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2351 #define Poison(d,n,t) PoisonFree(d,n,t)
2354 # define PERL_POISON_EXPR(x) x
2356 # define PERL_POISON_EXPR(x)
2359 #ifdef USE_STRUCT_COPY
2360 #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2362 #define StructCopy(s,d,t) Copy(s,d,1,t)
2365 /* C_ARRAY_LENGTH is the number of elements in the C array (so you
2366 * want your zero-based indices to be less than but not equal to).
2368 * C_ARRAY_END is one past the last: half-open/half-closed range,
2369 * not last-inclusive range. */
2370 #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
2371 #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
2375 # define Perl_va_copy(s, d) va_copy(d, s)
2377 # if defined(__va_copy)
2378 # define Perl_va_copy(s, d) __va_copy(d, s)
2380 # define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2385 /* convenience debug macros */
2387 #define pTHX_FORMAT "Perl interpreter: 0x%p"
2388 #define pTHX__FORMAT ", Perl interpreter: 0x%p"
2389 #define pTHX_VALUE_ (void *)my_perl,
2390 #define pTHX_VALUE (void *)my_perl
2391 #define pTHX__VALUE_ ,(void *)my_perl,
2392 #define pTHX__VALUE ,(void *)my_perl
2395 #define pTHX__FORMAT
2398 #define pTHX__VALUE_
2400 #endif /* USE_ITHREADS */
2402 /* Perl_deprecate was not part of the public API, and did not have a deprecate()
2403 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2404 CPAN::Unpack show any users outside the core. */
2406 # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2407 "Use of " s " is deprecated")
2410 /* Internal macros to deal with gids and uids */
2413 # if Uid_t_size > IVSIZE
2414 # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2415 # define SvUID(sv) SvNV(sv)
2417 # if Uid_t_sign <= 0
2418 # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2419 # define SvUID(sv) SvIV(sv)
2421 # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2422 # define SvUID(sv) SvUV(sv)
2424 # endif /* Uid_t_size */
2426 # if Gid_t_size > IVSIZE
2427 # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2428 # define SvGID(sv) SvNV(sv)
2430 # if Gid_t_sign <= 0
2431 # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2432 # define SvGID(sv) SvIV(sv)
2434 # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2435 # define SvGID(sv) SvUV(sv)
2437 # endif /* Gid_t_size */
2441 #endif /* HANDY_H */
2444 * ex: set ts=8 sts=4 sw=4 et:
https://perl5.git.perl.org / perl5.git / blob