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 /* This is a helper macro to avoid preprocessor issues, replaced by nothing
281 * unless under DEBUGGING, where it expands to an assert of its argument,
282 * followed by a comma (hence the comma operator). If we just used a straight
283 * assert(), we would get a comma with nothing before it when not DEBUGGING */
285 # define __ASSERT_(statement) assert(statement),
287 # define __ASSERT_(statement)
291 =head1 SV-Body Allocation
293 =for apidoc Ama|SV*|newSVpvs|const char* s
294 Like C<newSVpvn>, but takes a literal C<NUL>-terminated string instead of a
297 =for apidoc Ama|SV*|newSVpvs_flags|const char* s|U32 flags
298 Like C<newSVpvn_flags>, but takes a literal C<NUL>-terminated string instead of
299 a string/length pair.
301 =for apidoc Ama|SV*|newSVpvs_share|const char* s
302 Like C<newSVpvn_share>, but takes a literal C<NUL>-terminated string instead of
303 a string/length pair and omits the hash parameter.
305 =for apidoc Am|void|sv_catpvs_flags|SV* sv|const char* s|I32 flags
306 Like C<sv_catpvn_flags>, but takes a literal C<NUL>-terminated string instead
307 of a string/length pair.
309 =for apidoc Am|void|sv_catpvs_nomg|SV* sv|const char* s
310 Like C<sv_catpvn_nomg>, but takes a literal string instead of a
313 =for apidoc Am|void|sv_catpvs|SV* sv|const char* s
314 Like C<sv_catpvn>, but takes a literal string instead of a string/length pair.
316 =for apidoc Am|void|sv_catpvs_mg|SV* sv|const char* s
317 Like C<sv_catpvn_mg>, but takes a literal string instead of a
320 =for apidoc Am|void|sv_setpvs|SV* sv|const char* s
321 Like C<sv_setpvn>, but takes a literal string instead of a string/length pair.
323 =for apidoc Am|void|sv_setpvs_mg|SV* sv|const char* s
324 Like C<sv_setpvn_mg>, but takes a literal string instead of a
327 =for apidoc Am|SV *|sv_setref_pvs|const char* s
328 Like C<sv_setref_pvn>, but takes a literal string instead of a
331 =head1 Memory Management
333 =for apidoc Ama|char*|savepvs|const char* s
334 Like C<savepvn>, but takes a literal C<NUL>-terminated string instead of a
337 =for apidoc Ama|char*|savesharedpvs|const char* s
338 A version of C<savepvs()> which allocates the duplicate string in memory
339 which is shared between threads.
343 =for apidoc Am|HV*|gv_stashpvs|const char* name|I32 create
344 Like C<gv_stashpvn>, but takes a literal string instead of a string/length pair.
346 =head1 Hash Manipulation Functions
348 =for apidoc Am|SV**|hv_fetchs|HV* tb|const char* key|I32 lval
349 Like C<hv_fetch>, but takes a literal string instead of a string/length pair.
351 =for apidoc Am|SV**|hv_stores|HV* tb|const char* key|NULLOK SV* val
352 Like C<hv_store>, but takes a literal string instead of a string/length pair
353 and omits the hash parameter.
355 =head1 Lexer interface
357 =for apidoc Amx|void|lex_stuff_pvs|const char *pv|U32 flags
359 Like L</lex_stuff_pvn>, but takes a literal string instead of a
365 /* concatenating with "" ensures that only literal strings are accepted as
367 #define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
369 /* note that STR_WITH_LEN() can't be used as argument to macros or functions
370 * that under some configurations might be macros, which means that it requires
371 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
374 /* STR_WITH_LEN() shortcuts */
375 #define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
376 #define newSVpvs_flags(str,flags) \
377 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
378 #define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
379 #define sv_catpvs_flags(sv, str, flags) \
380 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
381 #define sv_catpvs_nomg(sv, str) \
382 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
383 #define sv_catpvs(sv, str) \
384 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
385 #define sv_catpvs_mg(sv, str) \
386 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
387 #define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
388 #define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
389 #define sv_setref_pvs(rv, classname, str) \
390 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
391 #define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
392 #define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
393 #define gv_stashpvs(str, create) \
394 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
395 #define gv_fetchpvs(namebeg, add, sv_type) \
396 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
397 #define gv_fetchpvn(namebeg, len, add, sv_type) \
398 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
399 #define sv_catxmlpvs(dsv, str, utf8) \
400 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
401 #define hv_fetchs(hv,key,lval) \
402 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
403 (lval) ? (HV_FETCH_JUST_SV | HV_FETCH_LVALUE) \
404 : HV_FETCH_JUST_SV, NULL, 0))
406 #define hv_stores(hv,key,val) \
407 ((SV **)Perl_hv_common(aTHX_ (hv), NULL, STR_WITH_LEN(key), 0, \
408 (HV_FETCH_ISSTORE|HV_FETCH_JUST_SV), (val), 0))
410 #define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
412 #define get_cvs(str, flags) \
413 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
416 =head1 Miscellaneous Functions
418 =for apidoc Am|bool|strNE|char* s1|char* s2
419 Test two strings to see if they are different. Returns true or
422 =for apidoc Am|bool|strEQ|char* s1|char* s2
423 Test two strings to see if they are equal. Returns true or false.
425 =for apidoc Am|bool|strLT|char* s1|char* s2
426 Test two strings to see if the first, C<s1>, is less than the second,
427 C<s2>. Returns true or false.
429 =for apidoc Am|bool|strLE|char* s1|char* s2
430 Test two strings to see if the first, C<s1>, is less than or equal to the
431 second, C<s2>. Returns true or false.
433 =for apidoc Am|bool|strGT|char* s1|char* s2
434 Test two strings to see if the first, C<s1>, is greater than the second,
435 C<s2>. Returns true or false.
437 =for apidoc Am|bool|strGE|char* s1|char* s2
438 Test two strings to see if the first, C<s1>, is greater than or equal to
439 the second, C<s2>. Returns true or false.
441 =for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
442 Test two strings to see if they are different. The C<len> parameter
443 indicates the number of bytes to compare. Returns true or false. (A
444 wrapper for C<strncmp>).
446 =for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
447 Test two strings to see if they are equal. The C<len> parameter indicates
448 the number of bytes to compare. Returns true or false. (A wrapper for
451 =for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
452 Test two buffers (which may contain embedded C<NUL> characters, to see if they
453 are equal. The C<len> parameter indicates the number of bytes to compare.
454 Returns zero if equal, or non-zero if non-equal.
456 =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
457 Test two buffers (which may contain embedded C<NUL> characters, to see if they
458 are not equal. The C<len> parameter indicates the number of bytes to compare.
459 Returns zero if non-equal, or non-zero if equal.
464 #define strNE(s1,s2) (strcmp(s1,s2))
465 #define strEQ(s1,s2) (!strcmp(s1,s2))
466 #define strLT(s1,s2) (strcmp(s1,s2) < 0)
467 #define strLE(s1,s2) (strcmp(s1,s2) <= 0)
468 #define strGT(s1,s2) (strcmp(s1,s2) > 0)
469 #define strGE(s1,s2) (strcmp(s1,s2) >= 0)
470 #define strnNE(s1,s2,l) (strncmp(s1,s2,l))
471 #define strnEQ(s1,s2,l) (!strncmp(s1,s2,l))
474 # define memNE(s1,s2,l) (memcmp(s1,s2,l))
475 # define memEQ(s1,s2,l) (!memcmp(s1,s2,l))
477 # define memNE(s1,s2,l) (bcmp(s1,s2,l))
478 # define memEQ(s1,s2,l) (!bcmp(s1,s2,l))
481 #define memEQs(s1, l, s2) \
482 (sizeof(s2)-1 == l && memEQ(s1, ("" s2 ""), (sizeof(s2)-1)))
483 #define memNEs(s1, l, s2) !memEQs(s1, l, s2)
488 * Unfortunately, the introduction of locales means that we
489 * can't trust isupper(), etc. to tell the truth. And when
490 * it comes to /\w+/ with tainting enabled, we *must* be able
491 * to trust our character classes.
493 * Therefore, the default tests in the text of Perl will be
494 * independent of locale. Any code that wants to depend on
495 * the current locale will use the tests that begin with "lc".
498 #ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
506 =head1 Character classification
507 This section is about functions (really macros) that classify characters
508 into types, such as punctuation versus alphabetic, etc. Most of these are
509 analogous to regular expression character classes. (See
510 L<perlrecharclass/POSIX Character Classes>.) There are several variants for
511 each class. (Not all macros have all variants; each item below lists the
512 ones valid for it.) None are affected by C<use bytes>, and only the ones
513 with C<LC> in the name are affected by the current locale.
515 The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
516 C<U8>) as input and returns a boolean as to whether or not the character
517 represented by that octet is (or on non-ASCII platforms, corresponds to) an
518 ASCII character in the named class based on platform, Unicode, and Perl rules.
519 If the input is a number that doesn't fit in an octet, FALSE is returned.
521 Variant C<isFOO_A> (e.g., C<isALPHA_A()>) is identical to the base function
522 with no suffix C<"_A">.
524 Variant C<isFOO_L1> imposes the Latin-1 (or EBCDIC equivlalent) character set
525 onto the platform. That is, the code points that are ASCII are unaffected,
526 since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
527 as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
528 true when called with the code point 0xDF, which is a word character in both
529 ASCII and EBCDIC (though it represents different characters in each).
531 Variant C<isFOO_uni> is like the C<isFOO_L1> variant, but accepts any UV code
532 point as input. If the code point is larger than 255, Unicode rules are used
533 to determine if it is in the character class. For example,
534 C<isWORDCHAR_uni(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
535 WITH MACRON in Unicode, and is a word character.
537 Variant C<isFOO_utf8> is like C<isFOO_uni>, but the input is a pointer to a
538 (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
539 classification of just the first (possibly multi-byte) character in the string
542 Variant C<isFOO_LC> is like the C<isFOO_A> and C<isFOO_L1> variants, but the
543 result is based on the current locale, which is what C<LC> in the name stands
544 for. If Perl can determine that the current locale is a UTF-8 locale, it uses
545 the published Unicode rules; otherwise, it uses the C library function that
546 gives the named classification. For example, C<isDIGIT_LC()> when not in a
547 UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
548 returned if the input won't fit into an octet. On some platforms where the C
549 library function is known to be defective, Perl changes its result to follow
550 the POSIX standard's rules.
552 Variant C<isFOO_LC_uvchr> is like C<isFOO_LC>, but is defined on any UV. It
553 returns the same as C<isFOO_LC> for input code points less than 256, and
554 returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
556 Variant C<isFOO_LC_utf8> is like C<isFOO_LC_uvchr>, but the input is a pointer
557 to a (known to be well-formed) UTF-8 encoded string (C<U8*> or C<char*>). The
558 classification of just the first (possibly multi-byte) character in the string
561 =for apidoc Am|bool|isALPHA|char ch
562 Returns a boolean indicating whether the specified character is an
563 alphabetic character, analogous to C<m/[[:alpha:]]/>.
564 See the L<top of this section|/Character classification> for an explanation of
566 C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uni>, C<isALPHA_utf8>, C<isALPHA_LC>,
567 C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8>.
569 =for apidoc Am|bool|isALPHANUMERIC|char ch
570 Returns a boolean indicating whether the specified character is a either an
571 alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
572 See the L<top of this section|/Character classification> for an explanation of
574 C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uni>,
575 C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, and
576 C<isALPHANUMERIC_LC_utf8>.
578 =for apidoc Am|bool|isASCII|char ch
579 Returns a boolean indicating whether the specified character is one of the 128
580 characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
581 On non-ASCII platforms, it returns TRUE iff this
582 character corresponds to an ASCII character. Variants C<isASCII_A()> and
583 C<isASCII_L1()> are identical to C<isASCII()>.
584 See the L<top of this section|/Character classification> for an explanation of
586 C<isASCII_uni>, C<isASCII_utf8>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
587 C<isASCII_LC_utf8>. Note, however, that some platforms do not have the C
588 library routine C<isascii()>. In these cases, the variants whose names contain
589 C<LC> are the same as the corresponding ones without.
591 Also note, that because all ASCII characters are UTF-8 invariant (meaning they
592 have the exact same representation (always a single byte) whether encoded in
593 UTF-8 or not), C<isASCII> will give the correct results when called with any
594 byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> will
595 work properly on any string encoded or not in UTF-8.
597 =for apidoc Am|bool|isBLANK|char ch
598 Returns a boolean indicating whether the specified character is a
599 character considered to be a blank, analogous to C<m/[[:blank:]]/>.
600 See the L<top of this section|/Character classification> for an explanation of
602 C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uni>, C<isBLANK_utf8>, C<isBLANK_LC>,
603 C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8>. Note, however, that some
604 platforms do not have the C library routine C<isblank()>. In these cases, the
605 variants whose names contain C<LC> are the same as the corresponding ones
608 =for apidoc Am|bool|isCNTRL|char ch
609 Returns a boolean indicating whether the specified character is a
610 control character, analogous to C<m/[[:cntrl:]]/>.
611 See the L<top of this section|/Character classification> for an explanation of
613 C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uni>, C<isCNTRL_utf8>, C<isCNTRL_LC>,
614 C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8>
615 On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
617 =for apidoc Am|bool|isDIGIT|char ch
618 Returns a boolean indicating whether the specified character is a
619 digit, analogous to C<m/[[:digit:]]/>.
620 Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
621 See the L<top of this section|/Character classification> for an explanation of
623 C<isDIGIT_uni>, C<isDIGIT_utf8>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
626 =for apidoc Am|bool|isGRAPH|char ch
627 Returns a boolean indicating whether the specified character is a
628 graphic character, analogous to C<m/[[:graph:]]/>.
629 See the L<top of this section|/Character classification> for an explanation of
631 C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uni>, C<isGRAPH_utf8>, C<isGRAPH_LC>,
632 C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8>.
634 =for apidoc Am|bool|isLOWER|char ch
635 Returns a boolean indicating whether the specified character is a
636 lowercase character, analogous to C<m/[[:lower:]]/>.
637 See the L<top of this section|/Character classification> for an explanation of
639 C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uni>, C<isLOWER_utf8>, C<isLOWER_LC>,
640 C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8>.
642 =for apidoc Am|bool|isOCTAL|char ch
643 Returns a boolean indicating whether the specified character is an
645 The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
648 =for apidoc Am|bool|isPUNCT|char ch
649 Returns a boolean indicating whether the specified character is a
650 punctuation character, analogous to C<m/[[:punct:]]/>.
651 Note that the definition of what is punctuation isn't as
652 straightforward as one might desire. See L<perlrecharclass/POSIX Character
653 Classes> for details.
654 See the L<top of this section|/Character classification> for an explanation of
656 C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uni>, C<isPUNCT_utf8>, C<isPUNCT_LC>,
657 C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8>.
659 =for apidoc Am|bool|isSPACE|char ch
660 Returns a boolean indicating whether the specified character is a
661 whitespace character. This is analogous
662 to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
663 this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
664 locale forms of this macro (the ones with C<LC> in their names) matched
665 precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
666 in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
667 (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
668 See the L<top of this section|/Character classification> for an explanation of
670 C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uni>, C<isSPACE_utf8>, C<isSPACE_LC>,
671 C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8>.
673 =for apidoc Am|bool|isPSXSPC|char ch
674 (short for Posix Space)
675 Starting in 5.18, this is identical in all its forms to the
676 corresponding C<isSPACE()> macros.
677 The locale forms of this macro are identical to their corresponding
678 C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
679 non-locale forms differ from their C<isSPACE()> forms only in that the
680 C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
681 Otherwise they are identical. Thus this macro is analogous to what
682 C<m/[[:space:]]/> matches in a regular expression.
683 See the L<top of this section|/Character classification> for an explanation of
684 variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uni>, C<isPSXSPC_utf8>,
685 C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8>.
687 =for apidoc Am|bool|isUPPER|char ch
688 Returns a boolean indicating whether the specified character is an
689 uppercase character, analogous to C<m/[[:upper:]]/>.
690 See the L<top of this section|/Character classification> for an explanation of
692 C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uni>, C<isUPPER_utf8>, C<isUPPER_LC>,
693 C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8>.
695 =for apidoc Am|bool|isPRINT|char ch
696 Returns a boolean indicating whether the specified character is a
697 printable character, analogous to C<m/[[:print:]]/>.
698 See the L<top of this section|/Character classification> for an explanation of
700 C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uni>, C<isPRINT_utf8>, C<isPRINT_LC>,
701 C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8>.
703 =for apidoc Am|bool|isWORDCHAR|char ch
704 Returns a boolean indicating whether the specified character is a character
705 that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
706 in a regular expression. A word character is an alphabetic character, a
707 decimal digit, a connecting punctuation character (such as an underscore), or
708 a "mark" character that attaches to one of those (like some sort of accent).
709 C<isALNUM()> is a synonym provided for backward compatibility, even though a
710 word character includes more than the standard C language meaning of
712 See the L<top of this section|/Character classification> for an explanation of
714 C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uni>, and C<isWORDCHAR_utf8>.
715 C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and C<isWORDCHAR_LC_utf8> are also as
716 described there, but additionally include the platform's native underscore.
718 =for apidoc Am|bool|isXDIGIT|char ch
719 Returns a boolean indicating whether the specified character is a hexadecimal
720 digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
721 and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
722 See the L<top of this section|/Character classification> for an explanation of
724 C<isXDIGIT_uni>, C<isXDIGIT_utf8>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, and
727 =for apidoc Am|bool|isIDFIRST|char ch
728 Returns a boolean indicating whether the specified character can be the first
729 character of an identifier. This is very close to, but not quite the same as
730 the official Unicode property C<XID_Start>. The difference is that this
731 returns true only if the input character also matches L</isWORDCHAR>.
732 See the L<top of this section|/Character classification> for an explanation of
734 C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uni>, C<isIDFIRST_utf8>,
735 C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8>.
737 =for apidoc Am|bool|isIDCONT|char ch
738 Returns a boolean indicating whether the specified character can be the
739 second or succeeding character of an identifier. This is very close to, but
740 not quite the same as the official Unicode property C<XID_Continue>. The
741 difference is that this returns true only if the input character also matches
742 L</isWORDCHAR>. See the L<top of this section|/Character classification> for
744 explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uni>,
745 C<isIDCONT_utf8>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
748 =head1 Miscellaneous Functions
750 =for apidoc Am|U8|READ_XDIGIT|char str*
751 Returns the value of an ASCII-range hex digit and advances the string pointer.
752 Behaviour is only well defined when isXDIGIT(*str) is true.
754 =head1 Character case changing
756 =for apidoc Am|U8|toUPPER|U8 ch
757 Converts the specified character to uppercase. If the input is anything but an
758 ASCII lowercase character, that input character itself is returned. Variant
759 C<toUPPER_A> is equivalent.
761 =for apidoc Am|UV|toUPPER_uni|UV cp|U8* s|STRLEN* lenp
762 Converts the Unicode code point C<cp> to its uppercase version, and
763 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
764 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
765 bytes since the uppercase version may be longer than the original character.
767 The first code point of the uppercased version is returned
768 (but note, as explained just above, that there may be more.)
770 =for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
771 Converts the UTF-8 encoded character at C<p> to its uppercase version, and
772 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
773 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
774 bytes since the uppercase version may be longer than the original character.
776 The first code point of the uppercased version is returned
777 (but note, as explained just above, that there may be more.)
779 The input character at C<p> is assumed to be well-formed.
781 =for apidoc Am|U8|toFOLD|U8 ch
782 Converts the specified character to foldcase. If the input is anything but an
783 ASCII uppercase character, that input character itself is returned. Variant
784 C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
785 Latin1 range, as the full generality of L</toFOLD_uni> is needed there.)
787 =for apidoc Am|UV|toFOLD_uni|UV cp|U8* s|STRLEN* lenp
788 Converts the Unicode code point C<cp> to its foldcase version, and
789 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
790 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
791 bytes since the foldcase version may be longer than the original character.
793 The first code point of the foldcased version is returned
794 (but note, as explained just above, that there may be more.)
796 =for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
797 Converts the UTF-8 encoded character at C<p> to its foldcase version, and
798 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
799 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
800 bytes since the foldcase version may be longer than the original character.
802 The first code point of the foldcased version is returned
803 (but note, as explained just above, that there may be more.)
805 The input character at C<p> is assumed to be well-formed.
807 =for apidoc Am|U8|toLOWER|U8 ch
808 Converts the specified character to lowercase. If the input is anything but an
809 ASCII uppercase character, that input character itself is returned. Variant
810 C<toLOWER_A> is equivalent.
812 =for apidoc Am|U8|toLOWER_L1|U8 ch
813 Converts the specified Latin1 character to lowercase. The results are
814 undefined if the input doesn't fit in a byte.
816 =for apidoc Am|U8|toLOWER_LC|U8 ch
817 Converts the specified character to lowercase using the current locale's rules,
818 if possible; otherwise returns the input character itself.
820 =for apidoc Am|UV|toLOWER_uni|UV cp|U8* s|STRLEN* lenp
821 Converts the Unicode code point C<cp> to its lowercase version, and
822 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
823 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
824 bytes since the lowercase version may be longer than the original character.
826 The first code point of the lowercased version is returned
827 (but note, as explained just above, that there may be more.)
829 =for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
830 Converts the UTF-8 encoded character at C<p> to its lowercase version, and
831 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
832 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
833 bytes since the lowercase version may be longer than the original character.
835 The first code point of the lowercased version is returned
836 (but note, as explained just above, that there may be more.)
838 The input character at C<p> is assumed to be well-formed.
840 =for apidoc Am|U8|toTITLE|U8 ch
841 Converts the specified character to titlecase. If the input is anything but an
842 ASCII lowercase character, that input character itself is returned. Variant
843 C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
844 range, as the full generality of L</toTITLE_uni> is needed there. Titlecase is
845 not a concept used in locale handling, so there is no functionality for that.)
847 =for apidoc Am|UV|toTITLE_uni|UV cp|U8* s|STRLEN* lenp
848 Converts the Unicode code point C<cp> to its titlecase version, and
849 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
850 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
851 bytes since the titlecase version may be longer than the original character.
853 The first code point of the titlecased version is returned
854 (but note, as explained just above, that there may be more.)
856 =for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
857 Converts the UTF-8 encoded character at C<p> to its titlecase version, and
858 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
859 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
860 bytes since the titlecase version may be longer than the original character.
862 The first code point of the titlecased version is returned
863 (but note, as explained just above, that there may be more.)
865 The input character at C<p> is assumed to be well-formed.
869 XXX Still undocumented isVERTWS_uni and _utf8; it's unclear what their names
870 really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change.
872 Note that these macros are repeated in Devel::PPPort, so should also be
873 patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
877 /* Specify the widest unsigned type on the platform. Use U64TYPE because U64
878 * is known only in the perl core, and this macro can be called from outside
881 # define WIDEST_UTYPE U64TYPE
883 # define WIDEST_UTYPE U32
886 /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
887 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
888 * bits of information are lost even on a 64-bit machine, but to get the
889 * compiler to optimize it out if possible. This is because Configure makes
890 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
891 * sizeof() guarantees that this evaluates to a constant true at compile time.
893 #define FITS_IN_8_BITS(c) ((sizeof(c) == 1) || !(((WIDEST_UTYPE)(c)) & ~0xFF))
897 /* The native libc isascii() et.al. functions return the wrong results
898 * on at least z/OS unless this is defined. */
899 # error _ALL_SOURCE should probably be defined
902 /* There is a simple definition of ASCII for ASCII platforms. But the
903 * EBCDIC one isn't so simple, so is defined using table look-up like the
904 * other macros below */
905 # define isASCII(c) ((WIDEST_UTYPE)(c) < 128)
908 /* The lower 3 bits in both the ASCII and EBCDIC representations of '0' are 0,
909 * and the 8 possible permutations of those bits exactly comprise the 8 octal
911 #define isOCTAL_A(c) cBOOL(FITS_IN_8_BITS(c) && (0xF8 & (c)) == '0')
913 #ifdef H_PERL /* If have access to perl.h, lookup in its table */
915 /* Character class numbers. For internal core Perl use only. The ones less
916 * than 32 are used in PL_charclass[] and the ones up through the one that
917 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
918 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
919 * their actual definitions are here. If that file has a name not used here,
922 * The first group of these is ordered in what I (khw) estimate to be the
923 * frequency of their use. This gives a slight edge to exiting a loop earlier
924 * (in reginclass() in regexec.c) */
925 # define _CC_WORDCHAR 0 /* \w and [:word:] */
926 # define _CC_DIGIT 1 /* \d and [:digit:] */
927 # define _CC_ALPHA 2 /* [:alpha:] */
928 # define _CC_LOWER 3 /* [:lower:] */
929 # define _CC_UPPER 4 /* [:upper:] */
930 # define _CC_PUNCT 5 /* [:punct:] */
931 # define _CC_PRINT 6 /* [:print:] */
932 # define _CC_ALPHANUMERIC 7 /* [:alnum:] */
933 # define _CC_GRAPH 8 /* [:graph:] */
934 # define _CC_CASED 9 /* [:lower:] and [:upper:] under /i */
936 #define _FIRST_NON_SWASH_CC 10
937 /* The character classes above are implemented with swashes. The second group
938 * (just below) contains the ones implemented without. These are also sorted
939 * in rough order of the frequency of their use, except that \v should be last,
940 * as it isn't a real Posix character class, and some (small) inefficiencies in
941 * regular expression handling would be introduced by putting it in the middle
942 * of those that are. Also, cntrl and ascii come after the others as it may be
943 * useful to group these which have no members that match above Latin1, (or
944 * above ASCII in the latter case) */
946 # define _CC_SPACE 10 /* \s, [:space:] */
947 # define _CC_BLANK 11 /* [:blank:] */
948 # define _CC_XDIGIT 12 /* [:xdigit:] */
949 # define _CC_CNTRL 13 /* [:cntrl:] */
950 # define _CC_ASCII 14 /* [:ascii:] */
951 # define _CC_VERTSPACE 15 /* \v */
953 # define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
955 /* The members of the third group below do not need to be coordinated with data
956 * structures in regcomp.[ch] and regexec.c. */
957 # define _CC_IDFIRST 16
958 # define _CC_CHARNAME_CONT 17
959 # define _CC_NONLATIN1_FOLD 18
960 # define _CC_NONLATIN1_SIMPLE_FOLD 19
961 # define _CC_QUOTEMETA 20
962 # define _CC_NON_FINAL_FOLD 21
963 # define _CC_IS_IN_SOME_FOLD 22
964 # define _CC_MNEMONIC_CNTRL 23
966 /* This next group is only used on EBCDIC platforms, so theoretically could be
967 * shared with something entirely different that's only on ASCII platforms */
968 # define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
969 # define _CC_UTF8_IS_START 29
970 # define _CC_UTF8_IS_DOWNGRADEABLE_START 30
971 # define _CC_UTF8_IS_CONTINUATION 31
973 * If more bits are needed, one could add a second word for non-64bit
974 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
975 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
976 * is used only for optimization (as of this writing), and differs in the
977 * Latin1 range from the ALPHA bit only in two relatively unimportant
978 * characters: the masculine and feminine ordinal indicators, so removing it
979 * would just cause /i regexes which match them to run less efficiently.
980 * Similarly the EBCDIC-only bits are used just for speed, and could be
981 * replaced by other means */
983 #if defined(PERL_CORE) || defined(PERL_EXT)
984 /* An enum version of the character class numbers, to help compilers
987 _CC_ENUM_ALPHA = _CC_ALPHA,
988 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
989 _CC_ENUM_ASCII = _CC_ASCII,
990 _CC_ENUM_BLANK = _CC_BLANK,
991 _CC_ENUM_CASED = _CC_CASED,
992 _CC_ENUM_CNTRL = _CC_CNTRL,
993 _CC_ENUM_DIGIT = _CC_DIGIT,
994 _CC_ENUM_GRAPH = _CC_GRAPH,
995 _CC_ENUM_LOWER = _CC_LOWER,
996 _CC_ENUM_PRINT = _CC_PRINT,
997 _CC_ENUM_PUNCT = _CC_PUNCT,
998 _CC_ENUM_SPACE = _CC_SPACE,
999 _CC_ENUM_UPPER = _CC_UPPER,
1000 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
1001 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
1002 _CC_ENUM_XDIGIT = _CC_XDIGIT
1003 } _char_class_number;
1006 #define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
1007 #define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
1009 #if defined(PERL_IN_UTF8_C) \
1010 || defined(PERL_IN_REGCOMP_C) \
1011 || defined(PERL_IN_REGEXEC_C)
1012 # if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
1013 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1014 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
1015 #error Need to adjust order of swash_property_names[]
1018 /* This is declared static in each of the few files that this is #defined for
1019 * to keep them from being publicly accessible. Hence there is a small amount
1020 * of wasted space */
1022 static const char* const swash_property_names[] = {
1038 EXTCONST U32 PL_charclass[] = {
1039 # include "l1_char_class_tab.h"
1042 # else /* ! DOINIT */
1043 EXTCONST U32 PL_charclass[];
1047 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
1048 # define _CC_mask(classnum) (1U << (classnum))
1050 /* For internal core Perl use only: the base macro for defining macros like
1052 # define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
1053 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
1055 /* The mask for the _A versions of the macros; it just adds in the bit for
1057 # define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1059 /* For internal core Perl use only: the base macro for defining macros like
1060 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1061 * the ASCII bit are present */
1062 # define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1063 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1064 == _CC_mask_A(classnum)))
1066 # define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1067 # define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
1068 # define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1069 # define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
1070 # define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
1071 # define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1072 # define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1073 # define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1074 # define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1075 # define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1076 # define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1077 # define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
1078 # define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1080 # define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
1081 # define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1082 # define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1083 # define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1085 /* continuation character for legal NAME in \N{NAME} */
1086 # define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1088 # define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1089 # define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1090 # define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1091 # define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1092 # define isPSXSPC_L1(c) isSPACE_L1(c)
1093 # define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1094 # define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1095 # define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1096 # define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1097 # define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
1100 # define isASCII(c) _generic_isCC(c, _CC_ASCII)
1103 /* Participates in a single-character fold with a character above 255 */
1104 # 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)))
1106 /* Like the above, but also can be part of a multi-char fold */
1107 # 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)))
1109 # define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
1110 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1111 _generic_isCC(c, _CC_NON_FINAL_FOLD)
1112 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1113 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1114 # define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1115 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1116 #else /* else we don't have perl.h H_PERL */
1118 /* If we don't have perl.h, we are compiling a utility program. Below we
1119 * hard-code various macro definitions that wouldn't otherwise be available
1120 * to it. Most are coded based on first principals. First some ones common
1121 * to both ASCII and EBCDIC */
1122 # define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1123 # define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1124 # define isSPACE_A(c) (isBLANK_A(c) \
1129 # ifdef EBCDIC /* There are gaps between 'i' and 'j'; 'r' and 's'. Same
1130 for uppercase. This is ordered to exclude most things
1132 # define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1134 || ((c) >= 'j' && (c) <= 'r') \
1136 # define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1138 || ((c) >= 'J' && (c) <= 'R') \
1140 # else /* ASCII platform. */
1141 # define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z')
1142 # define isUPPER_A(c) ((c) <= 'Z' && (c) >= 'A')
1145 /* Some more ASCII, non-ASCII common definitions */
1146 # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1147 # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
1148 # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
1149 # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1150 # define isXDIGIT_A(c) (isDIGIT_A(c) \
1151 || ((c) >= 'a' && (c) <= 'f') \
1152 || ((c) <= 'F' && (c) >= 'A'))
1155 # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1156 || (c) == '#' || (c) == '$' || (c) == '%' \
1157 || (c) == '&' || (c) == '\'' || (c) == '(' \
1158 || (c) == ')' || (c) == '*' || (c) == '+' \
1159 || (c) == ',' || (c) == '.' || (c) == '/' \
1160 || (c) == ':' || (c) == ';' || (c) == '<' \
1161 || (c) == '=' || (c) == '>' || (c) == '?' \
1162 || (c) == '@' || (c) == '[' || (c) == '\\' \
1163 || (c) == ']' || (c) == '^' || (c) == '_' \
1164 || (c) == '`' || (c) == '{' || (c) == '|' \
1165 || (c) == '}' || (c) == '~')
1166 # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1167 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1169 # ifdef QUESTION_MARK_CTRL
1170 # define _isQMC(c) ((c) == QUESTION_MARK_CTRL)
1172 # define _isQMC(c) 0
1175 /* I (khw) can't think of a way to define all the ASCII controls
1176 * without resorting to a libc (locale-sensitive) call. But we know
1177 * that all controls but the question-mark one are in the range 0-0x3f.
1178 * This makes sure that all the controls that have names are included,
1179 * and all controls that are also considered ASCII in the locale. This
1180 * may include more or fewer than what it actually should, but the
1181 * wrong ones are less-important controls, so likely won't impact
1182 * things (keep in mind that this is compiled only if perl.h isn't
1183 * available). The question mark control is included if available */
1184 # define isCNTRL_A(c) (((c) < 0x40 && isascii(c)) \
1185 || (c) == '\0' || (c) == '\a' || (c) == '\b' \
1186 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1187 || (c) == '\t' || (c) == '\v' || _isQMC(c))
1189 # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1190 # else /* ASCII platform; things are simpler, and isASCII has already
1192 # define isGRAPH_A(c) (((c) > ' ' && (c) < 127))
1193 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1194 # define isPUNCT_A(c) (isGRAPH_A(c) && (! isALPHANUMERIC_A(c)))
1195 # define isCNTRL_A(c) (isASCII(c) && (! isPRINT_A(c)))
1198 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
1199 * during debugging, and it seems best to keep them. We may be called
1200 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1201 * do anything anyway, so make it not a problem */
1202 # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1203 # define NATIVE_TO_LATIN1(ch) (ch)
1205 # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1206 # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1207 # define isBLANK_L1(c) (isBLANK_A(c) \
1208 || (FITS_IN_8_BITS(c) \
1209 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1210 # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1211 # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1212 # define isLOWER_L1(c) (isLOWER_A(c) \
1213 || (FITS_IN_8_BITS(c) \
1214 && ((NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1215 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1216 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1217 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1218 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1219 # define isPRINT_L1(c) (isPRINT_A(c) \
1220 || (FITS_IN_8_BITS(c) \
1221 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1222 # define isPUNCT_L1(c) (isPUNCT_A(c) \
1223 || (FITS_IN_8_BITS(c) \
1224 && (NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1225 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1226 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1227 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1228 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1229 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1230 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1231 # define isSPACE_L1(c) (isSPACE_A(c) \
1232 || (FITS_IN_8_BITS(c) \
1233 && (NATIVE_TO_LATIN1((U8) c) == 0x85 \
1234 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1235 # define isUPPER_L1(c) (isUPPER_A(c) \
1236 || (FITS_IN_8_BITS(c) \
1237 && (NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1238 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1239 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1240 # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1241 # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1242 # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1247 /* The following are not fully accurate in the above-ASCII range. I (khw)
1248 * don't think it's necessary to be so for the purposes where this gets
1250 # define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1251 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1253 /* And these aren't accurate at all. They are useful only for above
1254 * Latin1, which utilities and bootstrapping don't deal with */
1255 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
1256 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1257 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1259 /* Many of the macros later in this file are defined in terms of these. By
1260 * implementing them with a function, which converts the class number into
1261 * a call to the desired macro, all of the later ones work. However, that
1262 * function won't be actually defined when building a utility program (no
1263 * perl.h), and so a compiler error will be generated if one is attempted
1264 * to be used. And the above-Latin1 code points require Unicode tables to
1265 * be present, something unlikely to be the case when bootstrapping */
1266 # define _generic_isCC(c, classnum) \
1267 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1268 # define _generic_isCC_A(c, classnum) \
1269 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
1270 #endif /* End of no perl.h H_PERL */
1272 #define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1273 #define isALPHA(c) isALPHA_A(c)
1274 #define isASCII_A(c) isASCII(c)
1275 #define isASCII_L1(c) isASCII(c)
1276 #define isBLANK(c) isBLANK_A(c)
1277 #define isCNTRL(c) isCNTRL_A(c)
1278 #define isDIGIT(c) isDIGIT_A(c)
1279 #define isGRAPH(c) isGRAPH_A(c)
1280 #define isIDFIRST(c) isIDFIRST_A(c)
1281 #define isLOWER(c) isLOWER_A(c)
1282 #define isPRINT(c) isPRINT_A(c)
1283 #define isPSXSPC_A(c) isSPACE_A(c)
1284 #define isPSXSPC(c) isPSXSPC_A(c)
1285 #define isPSXSPC_L1(c) isSPACE_L1(c)
1286 #define isPUNCT(c) isPUNCT_A(c)
1287 #define isSPACE(c) isSPACE_A(c)
1288 #define isUPPER(c) isUPPER_A(c)
1289 #define isWORDCHAR(c) isWORDCHAR_A(c)
1290 #define isXDIGIT(c) isXDIGIT_A(c)
1292 /* ASCII casing. These could also be written as
1293 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1294 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1295 which uses table lookup and mask instead of subtraction. (This would
1296 work because the _MOD does not apply in the ASCII range) */
1297 #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1298 #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
1300 /* In the ASCII range, these are equivalent to what they're here defined to be.
1301 * But by creating these definitions, other code doesn't have to be aware of
1303 #define toFOLD(c) toLOWER(c)
1304 #define toTITLE(c) toUPPER(c)
1306 #define toLOWER_A(c) toLOWER(c)
1307 #define toUPPER_A(c) toUPPER(c)
1308 #define toFOLD_A(c) toFOLD(c)
1309 #define toTITLE_A(c) toTITLE(c)
1311 /* Use table lookup for speed; returns the input itself if is out-of-range */
1312 #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
1314 : PL_latin1_lc[ (U8) (c) ])
1315 #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1317 /* Modified uc. Is correct uc except for three non-ascii chars which are
1318 * all mapped to one of them, and these need special handling; returns the
1319 * input itself if is out-of-range */
1320 #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
1322 : PL_mod_latin1_uc[ (U8) (c) ])
1323 #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
1325 /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1327 /* For internal core Perl use only: the base macro for defining macros like
1328 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
1329 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1330 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1331 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1332 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1333 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1334 * code points under locale rules, use the macros based on _generic_LC_uvchr
1335 * instead of this. */
1336 #define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1337 (! FITS_IN_8_BITS(c) \
1339 : IN_UTF8_CTYPE_LOCALE \
1340 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
1343 /* For internal core Perl use only: a helper macro for defining macros like
1344 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1345 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1346 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1347 * _generic_LC_base for more info */
1348 #define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1349 _generic_LC_base(c,utf8_locale_classnum, \
1350 non_utf8_func( (_LC_CAST) (c)))
1352 /* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1353 * 'c' is the platform's native underscore character */
1354 #define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1355 _generic_LC_base(c, utf8_locale_classnum, \
1356 (non_utf8_func( (_LC_CAST) (c)) \
1357 || (char)(c) == '_'))
1359 /* These next three are also for internal core Perl use only: case-change
1361 #define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1363 : (IN_UTF8_CTYPE_LOCALE) \
1364 ? PL_latin1_lc[ (U8) (c) ] \
1365 : (cast)function((cast)(c)))
1367 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1368 * returns a single value, so can't adequately return the upper case of LATIN
1369 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1370 * values "SS"); instead it asserts against that under DEBUGGING, and
1371 * otherwise returns its input */
1372 #define _generic_toUPPER_LC(c, function, cast) \
1373 (! FITS_IN_8_BITS(c) \
1375 : ((! IN_UTF8_CTYPE_LOCALE) \
1376 ? (cast)function((cast)(c)) \
1377 : ((((U8)(c)) == MICRO_SIGN) \
1378 ? GREEK_CAPITAL_LETTER_MU \
1379 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1380 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1381 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1382 ? (__ASSERT_(0) (c)) \
1383 : PL_mod_latin1_uc[ (U8) (c) ])))))
1385 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1386 * returns a single value, so can't adequately return the fold case of LATIN
1387 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1388 * values "ss"); instead it asserts against that under DEBUGGING, and
1389 * otherwise returns its input */
1390 #define _generic_toFOLD_LC(c, function, cast) \
1391 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
1392 ? GREEK_SMALL_LETTER_MU \
1393 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1394 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1395 _generic_toLOWER_LC(c, function, cast)))
1397 /* Use the libc versions for these if available. */
1398 #if defined(HAS_ISASCII)
1399 # define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1401 # define isASCII_LC(c) isASCII(c)
1404 #if defined(HAS_ISBLANK)
1405 # define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
1406 #else /* Unlike isASCII, varies if in a UTF-8 locale */
1407 # define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
1413 /* The Windows functions don't bother to follow the POSIX standard, which
1414 * for example says that something can't both be a printable and a control.
1415 * But Windows treats the \t control as a printable, and does such things
1416 * as making superscripts into both digits and punctuation. This tames
1417 * these flaws by assuming that the definitions of both controls and space
1418 * are correct, and then making sure that other definitions don't have
1419 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1420 * Not all possible weirdnesses are checked for, just the ones that were
1421 * detected on actual Microsoft code pages */
1423 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1424 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1426 # define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1427 && isALPHANUMERIC_LC(c))
1428 # define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1430 # define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1431 isALPHANUMERIC_LC(c))
1432 # define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1433 # define isIDFIRST_LC(c) (((c) == '_') \
1434 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1435 # define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1436 # define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1437 # define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1438 # define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1439 # define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1440 # define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1441 && isALPHANUMERIC_LC(c))
1443 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1444 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1445 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1447 #elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
1448 /* For most other platforms */
1450 # define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1451 # define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1452 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1453 # define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1454 # define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1455 # define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1456 # define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1457 # define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1458 # define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1459 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1460 # define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1461 # define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1462 # define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1465 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1466 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1467 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1469 #else /* The final fallback position */
1471 # define isALPHA_LC(c) (isascii(c) && isalpha(c))
1472 # define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1473 # define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1474 # define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1475 # define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1476 # define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1477 # define isLOWER_LC(c) (isascii(c) && islower(c))
1478 # define isPRINT_LC(c) (isascii(c) && isprint(c))
1479 # define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1480 # define isSPACE_LC(c) (isascii(c) && isspace(c))
1481 # define isUPPER_LC(c) (isascii(c) && isupper(c))
1482 # define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1483 # define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1485 # define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1486 # define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1487 # define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
1491 #define isIDCONT(c) isWORDCHAR(c)
1492 #define isIDCONT_A(c) isWORDCHAR_A(c)
1493 #define isIDCONT_L1(c) isWORDCHAR_L1(c)
1494 #define isIDCONT_LC(c) isWORDCHAR_LC(c)
1495 #define isPSXSPC_LC(c) isSPACE_LC(c)
1497 /* For internal core Perl use only: the base macros for defining macros like
1498 * isALPHA_uni. 'c' is the code point to check. 'classnum' is the POSIX class
1499 * number defined earlier in this file. _generic_uni() is used for POSIX
1500 * classes where there is a macro or function 'above_latin1' that takes the
1501 * single argument 'c' and returns the desired value. These exist for those
1502 * classes which have simple definitions, avoiding the overhead of a hash
1503 * lookup or inversion list binary search. _generic_swash_uni() can be used
1504 * for classes where that overhead is faster than a direct lookup.
1505 * _generic_uni() won't compile if 'c' isn't unsigned, as it won't match the
1506 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1507 * have duplicate checks here, so could create versions of the macros that
1508 * don't, but experiments show that gcc optimizes them out anyway. */
1510 /* Note that all ignore 'use bytes' */
1511 #define _generic_uni(classnum, above_latin1, c) ((c) < 256 \
1512 ? _generic_isCC(c, classnum) \
1514 #define _generic_swash_uni(classnum, c) ((c) < 256 \
1515 ? _generic_isCC(c, classnum) \
1516 : _is_uni_FOO(classnum, c))
1517 #define isALPHA_uni(c) _generic_swash_uni(_CC_ALPHA, c)
1518 #define isALPHANUMERIC_uni(c) _generic_swash_uni(_CC_ALPHANUMERIC, c)
1519 #define isASCII_uni(c) isASCII(c)
1520 #define isBLANK_uni(c) _generic_uni(_CC_BLANK, is_HORIZWS_cp_high, c)
1521 #define isCNTRL_uni(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1522 #define isDIGIT_uni(c) _generic_swash_uni(_CC_DIGIT, c)
1523 #define isGRAPH_uni(c) _generic_swash_uni(_CC_GRAPH, c)
1524 #define isIDCONT_uni(c) _generic_uni(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1525 #define isIDFIRST_uni(c) _generic_uni(_CC_IDFIRST, _is_uni_perl_idstart, c)
1526 #define isLOWER_uni(c) _generic_swash_uni(_CC_LOWER, c)
1527 #define isPRINT_uni(c) _generic_swash_uni(_CC_PRINT, c)
1529 #define isPUNCT_uni(c) _generic_swash_uni(_CC_PUNCT, c)
1530 #define isSPACE_uni(c) _generic_uni(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1531 #define isPSXSPC_uni(c) isSPACE_uni(c)
1533 #define isUPPER_uni(c) _generic_swash_uni(_CC_UPPER, c)
1534 #define isVERTWS_uni(c) _generic_uni(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1535 #define isWORDCHAR_uni(c) _generic_swash_uni(_CC_WORDCHAR, c)
1536 #define isXDIGIT_uni(c) _generic_uni(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1538 #define toFOLD_uni(c,s,l) to_uni_fold(c,s,l)
1539 #define toLOWER_uni(c,s,l) to_uni_lower(c,s,l)
1540 #define toTITLE_uni(c,s,l) to_uni_title(c,s,l)
1541 #define toUPPER_uni(c,s,l) to_uni_upper(c,s,l)
1543 /* For internal core Perl use only: the base macros for defining macros like
1544 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1545 * point, not just 0-255. Like _generic_uni, there are two versions, one for
1546 * simple class definitions; the other for more complex. These are like
1547 * _generic_uni, so see it for more info. */
1548 #define _generic_LC_uvchr(latin1, above_latin1, c) \
1549 (c < 256 ? latin1(c) : above_latin1(c))
1550 #define _generic_LC_swash_uvchr(latin1, classnum, c) \
1551 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1553 #define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1554 #define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1555 _CC_ALPHANUMERIC, c)
1556 #define isASCII_LC_uvchr(c) isASCII_LC(c)
1557 #define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1558 is_HORIZWS_cp_high, c)
1559 #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
1560 #define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1561 #define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
1562 #define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
1563 _is_uni_perl_idcont, c)
1564 #define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
1565 _is_uni_perl_idstart, c)
1566 #define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1567 #define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
1568 #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
1569 #define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
1570 #define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
1571 is_XPERLSPACE_cp_high, c)
1572 #define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
1573 #define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
1575 #define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1576 is_XDIGIT_cp_high, c)
1578 #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
1580 /* For internal core Perl use only: the base macros for defining macros like
1581 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1582 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1583 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1584 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1585 * same representation whether utf8 or not. Note that it assumes that the utf8
1586 * has been validated, and ignores 'use bytes' */
1587 #define _generic_utf8(classnum, p, utf8) (UTF8_IS_INVARIANT(*(p)) \
1588 ? _generic_isCC(*(p), classnum) \
1589 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1591 EIGHT_BIT_UTF8_TO_NATIVE(*(p), \
1595 /* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1596 * 'above_latin1' can be a macro */
1597 #define _generic_func_utf8(classnum, above_latin1, p) \
1598 _generic_utf8(classnum, p, above_latin1(p))
1599 /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
1600 * 'above_latin1' parameter */
1601 #define _generic_swash_utf8(classnum, p) \
1602 _generic_utf8(classnum, p, _is_utf8_FOO(classnum, p))
1604 /* Like the above, but should be used only when it is known that there are no
1605 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1606 * class is TRUE for. Hence it can skip the tests for this range.
1607 * 'above_latin1' should include its arguments */
1608 #define _generic_utf8_no_upper_latin1(classnum, p, above_latin1) \
1609 (UTF8_IS_INVARIANT(*(p)) \
1610 ? _generic_isCC(*(p), classnum) \
1611 : (UTF8_IS_ABOVE_LATIN1(*(p))) \
1615 /* NOTE that some of these macros have very similar ones in regcharclass.h.
1616 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1617 * there, differing in name only by an underscore from the one here
1618 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1619 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1620 * points; the regcharclass.h ones are implemented as a series of
1621 * "if-else-if-else ..." */
1623 #define isALPHA_utf8(p) _generic_swash_utf8(_CC_ALPHA, p)
1624 #define isALPHANUMERIC_utf8(p) _generic_swash_utf8(_CC_ALPHANUMERIC, p)
1625 #define isASCII_utf8(p) isASCII(*p) /* Because ASCII is invariant under
1626 utf8, the non-utf8 macro works
1628 #define isBLANK_utf8(p) _generic_func_utf8(_CC_BLANK, is_HORIZWS_high, p)
1631 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1632 * more efficient macro instead of the more general one */
1633 # define isCNTRL_utf8(p) isCNTRL_L1(*(p))
1635 # define isCNTRL_utf8(p) _generic_utf8(_CC_CNTRL, p, 0)
1638 #define isDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_DIGIT, p, \
1639 _is_utf8_FOO(_CC_DIGIT, p))
1640 #define isGRAPH_utf8(p) _generic_swash_utf8(_CC_GRAPH, p)
1641 #define isIDCONT_utf8(p) _generic_func_utf8(_CC_WORDCHAR, \
1642 _is_utf8_perl_idcont, p)
1644 /* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1645 * IDFIRST is an alnum. See
1646 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
1647 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1648 * This used to be not the XID version, but we decided to go with the more
1649 * modern Unicode definition */
1650 #define isIDFIRST_utf8(p) _generic_func_utf8(_CC_IDFIRST, \
1651 _is_utf8_perl_idstart, p)
1653 #define isLOWER_utf8(p) _generic_swash_utf8(_CC_LOWER, p)
1654 #define isPRINT_utf8(p) _generic_swash_utf8(_CC_PRINT, p)
1655 #define isPSXSPC_utf8(p) isSPACE_utf8(p)
1656 #define isPUNCT_utf8(p) _generic_swash_utf8(_CC_PUNCT, p)
1657 #define isSPACE_utf8(p) _generic_func_utf8(_CC_SPACE, is_XPERLSPACE_high, p)
1658 #define isUPPER_utf8(p) _generic_swash_utf8(_CC_UPPER, p)
1659 #define isVERTWS_utf8(p) _generic_func_utf8(_CC_VERTSPACE, is_VERTWS_high, p)
1660 #define isWORDCHAR_utf8(p) _generic_swash_utf8(_CC_WORDCHAR, p)
1661 #define isXDIGIT_utf8(p) _generic_utf8_no_upper_latin1(_CC_XDIGIT, p, \
1664 #define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
1665 #define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1666 #define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1667 #define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
1669 /* For internal core Perl use only: the base macros for defining macros like
1670 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1671 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1672 * 'macro' parameter */
1673 #define _generic_LC_utf8(macro, p, utf8) \
1674 (UTF8_IS_INVARIANT(*(p)) \
1676 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1677 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1)))\
1680 #define _generic_LC_swash_utf8(macro, classnum, p) \
1681 _generic_LC_utf8(macro, p, _is_utf8_FOO(classnum, p))
1682 #define _generic_LC_func_utf8(macro, above_latin1, p) \
1683 _generic_LC_utf8(macro, p, above_latin1(p))
1685 #define isALPHANUMERIC_LC_utf8(p) _generic_LC_swash_utf8(isALPHANUMERIC_LC, \
1686 _CC_ALPHANUMERIC, p)
1687 #define isALPHA_LC_utf8(p) _generic_LC_swash_utf8(isALPHA_LC, _CC_ALPHA, p)
1688 #define isASCII_LC_utf8(p) isASCII_LC(*p)
1689 #define isBLANK_LC_utf8(p) _generic_LC_func_utf8(isBLANK_LC, \
1691 #define isCNTRL_LC_utf8(p) _generic_LC_utf8(isCNTRL_LC, p, 0)
1692 #define isDIGIT_LC_utf8(p) _generic_LC_swash_utf8(isDIGIT_LC, _CC_DIGIT, p)
1693 #define isGRAPH_LC_utf8(p) _generic_LC_swash_utf8(isGRAPH_LC, _CC_GRAPH, p)
1694 #define isIDCONT_LC_utf8(p) _generic_LC_func_utf8(isIDCONT_LC, \
1695 _is_utf8_perl_idcont, p)
1696 #define isIDFIRST_LC_utf8(p) _generic_LC_func_utf8(isIDFIRST_LC, \
1697 _is_utf8_perl_idstart, p)
1698 #define isLOWER_LC_utf8(p) _generic_LC_swash_utf8(isLOWER_LC, _CC_LOWER, p)
1699 #define isPRINT_LC_utf8(p) _generic_LC_swash_utf8(isPRINT_LC, _CC_PRINT, p)
1700 #define isPSXSPC_LC_utf8(p) isSPACE_LC_utf8(p)
1701 #define isPUNCT_LC_utf8(p) _generic_LC_swash_utf8(isPUNCT_LC, _CC_PUNCT, p)
1702 #define isSPACE_LC_utf8(p) _generic_LC_func_utf8(isSPACE_LC, \
1703 is_XPERLSPACE_high, p)
1704 #define isUPPER_LC_utf8(p) _generic_LC_swash_utf8(isUPPER_LC, _CC_UPPER, p)
1705 #define isWORDCHAR_LC_utf8(p) _generic_LC_swash_utf8(isWORDCHAR_LC, \
1707 #define isXDIGIT_LC_utf8(p) _generic_LC_func_utf8(isXDIGIT_LC, \
1710 /* Macros for backwards compatibility and for completeness when the ASCII and
1711 * Latin1 values are identical */
1712 #define isALPHAU(c) isALPHA_L1(c)
1713 #define isDIGIT_L1(c) isDIGIT_A(c)
1714 #define isOCTAL(c) isOCTAL_A(c)
1715 #define isOCTAL_L1(c) isOCTAL_A(c)
1716 #define isXDIGIT_L1(c) isXDIGIT_A(c)
1717 #define isALNUM(c) isWORDCHAR(c)
1718 #define isALNUMU(c) isWORDCHAR_L1(c)
1719 #define isALNUM_LC(c) isWORDCHAR_LC(c)
1720 #define isALNUM_uni(c) isWORDCHAR_uni(c)
1721 #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
1722 #define isALNUM_utf8(p) isWORDCHAR_utf8(p)
1723 #define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
1724 #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
1725 #define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
1726 #define isALNUMC(c) isALPHANUMERIC(c)
1727 #define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
1728 #define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
1729 #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
1730 #define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
1731 #define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
1733 /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
1734 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
1735 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
1736 * '?' is special-cased on EBCDIC to APC, which is the control there that is
1737 * the outlier from the block that contains the other controls, just like
1738 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
1739 * block. If it weren't special cased, it would yield a non-control.
1740 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
1743 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
1745 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
1747 ? (UNLIKELY((c) == '?') \
1748 ? QUESTION_MARK_CTRL \
1749 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
1750 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
1752 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
1755 /* Line numbers are unsigned, 32 bits. */
1757 #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
1759 /* Helpful alias for version prescan */
1760 #define is_LAX_VERSION(a,b) \
1761 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
1763 #define is_STRICT_VERSION(a,b) \
1764 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
1766 #define BADVERSION(a,b,c) \
1772 /* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
1773 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
1774 * which is advanced. The input is validated only by an assert() in DEBUGGING
1775 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
1776 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
1777 #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
1780 #define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
1784 /* Converts a character known to represent an octal digit (0-7) to its numeric
1785 * value. The input is validated only by an assert() in DEBUGGING builds. In
1786 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
1787 #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
1789 /* Efficiently returns a boolean as to if two native characters are equivalent
1790 * case-insenstively. At least one of the characters must be one of [A-Za-z];
1791 * the ALPHA in the name is to remind you of that. This is asserted() in
1792 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
1793 * works (on valid input) for both non- and UTF-8-encoded bytes.
1795 * When one of the inputs is a compile-time constant and gets folded by the
1796 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
1797 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
1798 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
1799 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
1800 * exclusive 'or' to find that bit and then inverts it to form a mask, with
1801 * just a single 0, in the bit position where the upper- and lowercase differ.
1803 #define isALPHA_FOLD_EQ(c1, c2) \
1804 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
1805 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
1806 #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
1809 =head1 Memory Management
1811 =for apidoc Am|void|Newx|void* ptr|int nitems|type
1812 The XSUB-writer's interface to the C C<malloc> function.
1814 Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1816 In 5.9.3, Newx() and friends replace the older New() API, and drops
1817 the first parameter, I<x>, a debug aid which allowed callers to identify
1818 themselves. This aid has been superseded by a new build option,
1819 PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
1820 there for use in XS modules supporting older perls.
1822 =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
1823 The XSUB-writer's interface to the C C<malloc> function, with
1824 cast. See also C<L</Newx>>.
1826 Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1828 =for apidoc Am|void|Newxz|void* ptr|int nitems|type
1829 The XSUB-writer's interface to the C C<malloc> function. The allocated
1830 memory is zeroed with C<memzero>. See also C<L</Newx>>.
1832 Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1834 =for apidoc Am|void|Renew|void* ptr|int nitems|type
1835 The XSUB-writer's interface to the C C<realloc> function.
1837 Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1839 =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
1840 The XSUB-writer's interface to the C C<realloc> function, with
1843 Memory obtained by this should B<ONLY> be freed with L<"Safefree">.
1845 =for apidoc Am|void|Safefree|void* ptr
1846 The XSUB-writer's interface to the C C<free> function.
1848 This should B<ONLY> be used on memory obtained using L<"Newx"> and friends.
1850 =for apidoc Am|void|Move|void* src|void* dest|int nitems|type
1851 The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
1852 source, C<dest> is the destination, C<nitems> is the number of items, and
1853 C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
1855 =for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
1856 Like C<Move> but returns C<dest>. Useful
1857 for encouraging compilers to tail-call
1860 =for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
1861 The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
1862 source, C<dest> is the destination, C<nitems> is the number of items, and
1863 C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
1865 =for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
1867 Like C<Copy> but returns C<dest>. Useful
1868 for encouraging compilers to tail-call
1871 =for apidoc Am|void|Zero|void* dest|int nitems|type
1873 The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
1874 destination, C<nitems> is the number of items, and C<type> is the type.
1876 =for apidoc Am|void *|ZeroD|void* dest|int nitems|type
1878 Like C<Zero> but returns dest. Useful
1879 for encouraging compilers to tail-call
1882 =for apidoc Am|void|StructCopy|type *src|type *dest|type
1883 This is an architecture-independent macro to copy one structure to another.
1885 =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
1887 Fill up memory with a byte pattern (a byte repeated over and over
1888 again) that hopefully catches attempts to access uninitialized memory.
1890 =for apidoc Am|void|PoisonNew|void* dest|int nitems|type
1892 PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
1894 =for apidoc Am|void|PoisonFree|void* dest|int nitems|type
1896 PoisonWith(0xEF) for catching access to freed memory.
1898 =for apidoc Am|void|Poison|void* dest|int nitems|type
1900 PoisonWith(0xEF) for catching access to freed memory.
1904 /* Maintained for backwards-compatibility only. Use newSV() instead. */
1906 #define NEWSV(x,len) newSV(len)
1909 #define MEM_SIZE_MAX ((MEM_SIZE)~0)
1912 #ifdef PERL_MALLOC_WRAP
1914 /* This expression will be constant-folded at compile time. It checks
1915 * whether or not the type of the count n is so small (e.g. U8 or U16, or
1916 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
1917 * As well as avoiding the need for a run-time check in some cases, it's
1918 * designed to avoid compiler warnings like:
1919 * comparison is always false due to limited range of data type
1922 # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
1923 (sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
1925 /* This is written in a slightly odd way to avoid various spurious
1926 * compiler warnings. We *want* to write the expression as
1927 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
1928 * (for some compile-time constant C), but even when the LHS
1929 * constant-folds to false at compile-time, g++ insists on emitting
1930 * warnings about the RHS (e.g. "comparison is always false"), so instead
1933 * (cond ? n : X) > C
1935 * where X is a constant with X > C always false. Choosing a value for X
1936 * is tricky. If 0, some compilers will complain about 0 > C always being
1937 * false; if 1, Coverity complains when n happens to be the constant value
1938 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
1939 * for X and hope that nothing else whines.
1942 # define _MEM_WRAP_WILL_WRAP(n,t) \
1943 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
1944 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
1946 # define MEM_WRAP_CHECK(n,t) \
1947 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1948 && (croak_memory_wrap(),0))
1950 # define MEM_WRAP_CHECK_1(n,t,a) \
1951 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
1952 && (Perl_croak_nocontext("%s",(a)),0))
1954 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
1956 #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)))
1959 #define MEM_WRAP_CHECK(n,t)
1960 #define MEM_WRAP_CHECK_1(n,t,a)
1961 #define MEM_WRAP_CHECK_2(n,t,a,b)
1962 #define MEM_WRAP_CHECK_(n,t)
1964 #define PERL_STRLEN_ROUNDUP(n) (((n-1+PERL_STRLEN_ROUNDUP_QUANTUM)&~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM-1)))
1970 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
1971 * go through functions, which are handy for debugging breakpoints, but
1972 * which more importantly get the immediate calling environment (file and
1973 * line number, and C function name if available) passed in. This info can
1974 * then be used for logging the calls, for which one gets a sample
1975 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
1978 * - not all memory allocs get logged, only those
1979 * that go through Newx() and derivatives (while all
1980 * Safefrees do get logged)
1981 * - __FILE__ and __LINE__ do not work everywhere
1982 * - __func__ or __FUNCTION__ even less so
1983 * - I think more goes on after the perlio frees but
1984 * the thing is that STDERR gets closed (as do all
1985 * the file descriptors)
1986 * - no deeper calling stack than the caller of the Newx()
1987 * or the kind, but do I look like a C reflection/introspection
1989 * - the function prototypes for the logging functions
1990 * probably should maybe be somewhere else than handy.h
1991 * - one could consider inlining (macrofying) the logging
1992 * for speed, but I am too lazy
1993 * - one could imagine recording the allocations in a hash,
1994 * (keyed by the allocation address?), and maintain that
1995 * through reallocs and frees, but how to do that without
1996 * any News() happening...?
1997 * - lots of -Ddefines to get useful/controllable output
1998 * - lots of ENV reads
2001 PERL_CALLCONV Malloc_t Perl_mem_log_alloc(const UV n, const UV typesize, const char *type_name, Malloc_t newalloc, const char *filename, const int linenumber, const char *funcname);
2003 PERL_CALLCONV Malloc_t Perl_mem_log_realloc(const UV n, const UV typesize, const char *type_name, Malloc_t oldalloc, Malloc_t newalloc, const char *filename, const int linenumber, const char *funcname);
2005 PERL_CALLCONV Malloc_t Perl_mem_log_free(Malloc_t oldalloc, const char *filename, const int linenumber, const char *funcname);
2008 # ifndef PERL_MEM_LOG_NOIMPL
2017 # if defined(PERL_IN_SV_C) /* those are only used in sv.c */
2018 void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2019 void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2026 #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2027 #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
2028 #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
2031 #ifndef MEM_LOG_ALLOC
2032 #define MEM_LOG_ALLOC(n,t,a) (a)
2034 #ifndef MEM_LOG_REALLOC
2035 #define MEM_LOG_REALLOC(n,t,v,a) (a)
2037 #ifndef MEM_LOG_FREE
2038 #define MEM_LOG_FREE(a) (a)
2041 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2042 #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2043 #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
2046 /* pre 5.9.x compatibility */
2047 #define New(x,v,n,t) Newx(v,n,t)
2048 #define Newc(x,v,n,t,c) Newxc(v,n,t,c)
2049 #define Newz(x,v,n,t) Newxz(v,n,t)
2052 #define Renew(v,n,t) \
2053 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2054 #define Renewc(v,n,t,c) \
2055 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2058 #define Safefree(d) \
2059 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
2061 #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
2064 #define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2065 #define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2066 #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) (void)memzero((char*)(d), (n) * sizeof(t)))
2068 #define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2069 #define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2071 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)))
2073 /* Using bzero(), which returns void. */
2074 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) memzero((char*)(d), (n) * sizeof(t)),d)
2077 #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2078 #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2079 #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2080 #define Poison(d,n,t) PoisonFree(d,n,t)
2083 # define PERL_POISON_EXPR(x) x
2085 # define PERL_POISON_EXPR(x)
2088 #ifdef USE_STRUCT_COPY
2089 #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2091 #define StructCopy(s,d,t) Copy(s,d,1,t)
2094 /* C_ARRAY_LENGTH is the number of elements in the C array (so you
2095 * want your zero-based indices to be less than but not equal to).
2097 * C_ARRAY_END is one past the last: half-open/half-closed range,
2098 * not last-inclusive range. */
2099 #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
2100 #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
2104 # define Perl_va_copy(s, d) va_copy(d, s)
2106 # if defined(__va_copy)
2107 # define Perl_va_copy(s, d) __va_copy(d, s)
2109 # define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2114 /* convenience debug macros */
2116 #define pTHX_FORMAT "Perl interpreter: 0x%p"
2117 #define pTHX__FORMAT ", Perl interpreter: 0x%p"
2118 #define pTHX_VALUE_ (void *)my_perl,
2119 #define pTHX_VALUE (void *)my_perl
2120 #define pTHX__VALUE_ ,(void *)my_perl,
2121 #define pTHX__VALUE ,(void *)my_perl
2124 #define pTHX__FORMAT
2127 #define pTHX__VALUE_
2129 #endif /* USE_ITHREADS */
2131 /* Perl_deprecate was not part of the public API, and did not have a deprecate()
2132 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2133 CPAN::Unpack show any users outside the core. */
2135 # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2136 "Use of " s " is deprecated")
2139 /* Internal macros to deal with gids and uids */
2142 # if Uid_t_size > IVSIZE
2143 # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2144 # define SvUID(sv) SvNV(sv)
2146 # if Uid_t_sign <= 0
2147 # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2148 # define SvUID(sv) SvIV(sv)
2150 # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2151 # define SvUID(sv) SvUV(sv)
2153 # endif /* Uid_t_size */
2155 # if Gid_t_size > IVSIZE
2156 # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2157 # define SvGID(sv) SvNV(sv)
2159 # if Gid_t_sign <= 0
2160 # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2161 # define SvGID(sv) SvIV(sv)
2163 # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2164 # define SvGID(sv) SvUV(sv)
2166 # endif /* Gid_t_size */
2170 #endif /* HANDY_H */
2173 * ex: set ts=8 sts=4 sw=4 et:
https://perl5.git.perl.org / perl5.git / blob