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 PERL_HANDY_H_ /* Guard against nested #inclusion */
18 # define Null(type) ((type)NULL)
23 =for apidoc AmU||Nullch
24 Null character pointer. (No longer available when C<PERL_CORE> is
27 =for apidoc AmU||Nullsv
28 Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
33 # define Nullch Null(char*)
34 # define Nullfp Null(PerlIO*)
35 # define Nullsv Null(SV*)
47 /* The MUTABLE_*() macros cast pointers to the types shown, in such a way
48 * (compiler permitting) that casting away const-ness will give a warning;
52 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
53 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
56 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
57 # define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
59 # define MUTABLE_PTR(p) ((void *) (p))
62 #define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
63 #define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
64 #define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
65 #define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
66 #define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
67 #define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
69 #if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
76 /* bool is built-in for g++-2.6.3 and later, which might be used
77 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
78 be sure _G_config.h will be included before this file. _G_config.h
79 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
80 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
81 g++ can be identified by __GNUG__.
82 Andy Dougherty February 2000
84 #ifdef __GNUG__ /* GNU g++ has bool built-in */
85 # ifndef PERL_BOOL_AS_CHAR
100 /* cast-to-bool. A simple (bool) cast may not do the right thing: if bool is
101 * defined as char for example, then the cast from int is
102 * implementation-defined (bool)!!(cbool) in a ternary triggers a bug in xlc on
104 #define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
106 /* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
107 * XXX Should really be a Configure probe, with HAS__FUNCTION__
108 * and FUNCTION__ as results.
109 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
110 #if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
111 # define FUNCTION__ __func__
112 #elif (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
113 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
114 # define FUNCTION__ ""
116 # define FUNCTION__ __FUNCTION__ /* Common extension. */
119 /* XXX A note on the perl source internal type system. The
120 original intent was that I32 be *exactly* 32 bits.
122 Currently, we only guarantee that I32 is *at least* 32 bits.
123 Specifically, if int is 64 bits, then so is I32. (This is the case
124 for the Cray.) This has the advantage of meshing nicely with
125 standard library calls (where we pass an I32 and the library is
126 expecting an int), but the disadvantage that an I32 is not 32 bits.
127 Andy Dougherty August 1996
129 There is no guarantee that there is *any* integral type with
130 exactly 32 bits. It is perfectly legal for a system to have
131 sizeof(short) == sizeof(int) == sizeof(long) == 8.
133 Similarly, there is no guarantee that I16 and U16 have exactly 16
136 For dealing with issues that may arise from various 32/64-bit
137 systems, we will ask Configure to check out
139 SHORTSIZE == sizeof(short)
140 INTSIZE == sizeof(int)
141 LONGSIZE == sizeof(long)
142 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
143 PTRSIZE == sizeof(void *)
144 DOUBLESIZE == sizeof(double)
145 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
149 #ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
150 # include <inttypes.h>
151 # ifdef INT32_MIN_BROKEN
153 # define INT32_MIN (-2147483647-1)
155 # ifdef INT64_MIN_BROKEN
157 # define INT64_MIN (-9223372036854775807LL-1)
173 #if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
175 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
176 Please search CHAR_MAX in perl.h for further details. */
177 #define U8_MAX UINT8_MAX
178 #define U8_MIN UINT8_MIN
180 #define I16_MAX INT16_MAX
181 #define I16_MIN INT16_MIN
182 #define U16_MAX UINT16_MAX
183 #define U16_MIN UINT16_MIN
185 #define I32_MAX INT32_MAX
186 #define I32_MIN INT32_MIN
187 #ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
188 # define U32_MAX UINT32_MAX
190 # define U32_MAX 4294967295U
192 #define U32_MIN UINT32_MIN
196 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
197 Please search CHAR_MAX in perl.h for further details. */
198 #define U8_MAX PERL_UCHAR_MAX
199 #define U8_MIN PERL_UCHAR_MIN
201 #define I16_MAX PERL_SHORT_MAX
202 #define I16_MIN PERL_SHORT_MIN
203 #define U16_MAX PERL_USHORT_MAX
204 #define U16_MIN PERL_USHORT_MIN
207 # define I32_MAX PERL_INT_MAX
208 # define I32_MIN PERL_INT_MIN
209 # define U32_MAX PERL_UINT_MAX
210 # define U32_MIN PERL_UINT_MIN
212 # define I32_MAX PERL_LONG_MAX
213 # define I32_MIN PERL_LONG_MIN
214 # define U32_MAX PERL_ULONG_MAX
215 # define U32_MIN PERL_ULONG_MIN
220 /* These C99 typedefs are useful sometimes for, say, loop variables whose
221 * maximum values are small, but for which speed trumps size. If we have a C99
222 * compiler, use that. Otherwise, a plain 'int' should be good enough.
224 * Restrict these to core for now until we are more certain this is a good
226 #if defined(PERL_CORE) || defined(PERL_EXT)
228 typedef int_fast8_t PERL_INT_FAST8_T;
229 typedef uint_fast8_t PERL_UINT_FAST8_T;
230 typedef int_fast16_t PERL_INT_FAST16_T;
231 typedef uint_fast16_t PERL_UINT_FAST16_T;
233 typedef int PERL_INT_FAST8_T;
234 typedef unsigned int PERL_UINT_FAST8_T;
235 typedef int PERL_INT_FAST16_T;
236 typedef unsigned int PERL_UINT_FAST16_T;
240 /* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
241 * anyone is grepping for it */
242 #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
243 #define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
244 #define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
246 /* Unused by core; should be deprecated */
247 #define Ctl(ch) ((ch) & 037)
249 #if defined(PERL_CORE) || defined(PERL_EXT)
251 # define MIN(a,b) ((a) < (b) ? (a) : (b))
254 # define MAX(a,b) ((a) > (b) ? (a) : (b))
258 /* Returns a boolean as to whether the input unsigned number is a power of 2
259 * (2**0, 2**1, etc). In other words if it has just a single bit set.
260 * If not, subtracting 1 would leave the uppermost bit set, so the & would
262 #if defined(PERL_CORE) || defined(PERL_EXT)
263 # define isPOWER_OF_2(n) (n && (n & (n-1)) == 0)
266 /* This is a helper macro to avoid preprocessor issues, replaced by nothing
267 * unless under DEBUGGING, where it expands to an assert of its argument,
268 * followed by a comma (hence the comma operator). If we just used a straight
269 * assert(), we would get a comma with nothing before it when not DEBUGGING.
271 * We also use empty definition under Coverity since the __ASSERT__
272 * checks often check for things that Really Cannot Happen, and Coverity
273 * detects that and gets all excited. */
275 #if defined(DEBUGGING) && !defined(__COVERITY__)
276 # define __ASSERT_(statement) assert(statement),
278 # define __ASSERT_(statement)
282 =head1 SV Manipulation Functions
284 =for apidoc Ama|SV*|newSVpvs|"literal string" s
285 Like C<newSVpvn>, but takes a literal string instead of a
288 =for apidoc Ama|SV*|newSVpvs_flags|"literal string" s|U32 flags
289 Like C<newSVpvn_flags>, but takes a literal string instead of
290 a string/length pair.
292 =for apidoc Ama|SV*|newSVpvs_share|"literal string" s
293 Like C<newSVpvn_share>, but takes a literal string instead of
294 a string/length pair and omits the hash parameter.
296 =for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string" s|I32 flags
297 Like C<sv_catpvn_flags>, but takes a literal string instead
298 of a string/length pair.
300 =for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string" s
301 Like C<sv_catpvn_nomg>, but takes a literal string instead of
302 a string/length pair.
304 =for apidoc Am|void|sv_catpvs|SV* sv|"literal string" s
305 Like C<sv_catpvn>, but takes a literal string instead of a
308 =for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string" s
309 Like C<sv_catpvn_mg>, but takes a literal string instead of a
312 =for apidoc Am|void|sv_setpvs|SV* sv|"literal string" s
313 Like C<sv_setpvn>, but takes a literal string instead of a
316 =for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string" s
317 Like C<sv_setpvn_mg>, but takes a literal string instead of a
320 =for apidoc Am|SV *|sv_setref_pvs|"literal string" s
321 Like C<sv_setref_pvn>, but takes a literal string instead of
322 a string/length pair.
324 =head1 Memory Management
326 =for apidoc Ama|char*|savepvs|"literal string" s
327 Like C<savepvn>, but takes a literal string instead of a
330 =for apidoc Ama|char*|savesharedpvs|"literal string" s
331 A version of C<savepvs()> which allocates the duplicate string in memory
332 which is shared between threads.
336 =for apidoc Am|HV*|gv_stashpvs|"literal string" name|I32 create
337 Like C<gv_stashpvn>, but takes a literal string instead of a
340 =head1 Hash Manipulation Functions
342 =for apidoc Am|SV**|hv_fetchs|HV* tb|"literal string" key|I32 lval
343 Like C<hv_fetch>, but takes a literal string instead of a
346 =for apidoc Am|SV**|hv_stores|HV* tb|"literal string" key|SV* val
347 Like C<hv_store>, but takes a literal string instead of a
349 and omits the hash parameter.
351 =head1 Lexer interface
353 =for apidoc Amx|void|lex_stuff_pvs|"literal string" pv|U32 flags
355 Like L</lex_stuff_pvn>, but takes a literal string instead of
356 a string/length pair.
361 /* concatenating with "" ensures that only literal strings are accepted as
363 #define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
365 /* note that STR_WITH_LEN() can't be used as argument to macros or functions
366 * that under some configurations might be macros, which means that it requires
367 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
370 /* STR_WITH_LEN() shortcuts */
371 #define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
372 #define newSVpvs_flags(str,flags) \
373 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
374 #define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
375 #define sv_catpvs_flags(sv, str, flags) \
376 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
377 #define sv_catpvs_nomg(sv, str) \
378 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
379 #define sv_catpvs(sv, str) \
380 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
381 #define sv_catpvs_mg(sv, str) \
382 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
383 #define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
384 #define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
385 #define sv_setref_pvs(rv, classname, str) \
386 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
387 #define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
388 #define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
389 #define gv_stashpvs(str, create) \
390 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
391 #define gv_fetchpvs(namebeg, add, sv_type) \
392 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
393 #define gv_fetchpvn(namebeg, len, add, sv_type) \
394 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
395 #define sv_catxmlpvs(dsv, str, utf8) \
396 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
399 #define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
401 #define get_cvs(str, flags) \
402 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
405 =head1 Miscellaneous Functions
407 =for apidoc Am|bool|strNE|char* s1|char* s2
408 Test two C<NUL>-terminated strings to see if they are different. Returns true
411 =for apidoc Am|bool|strEQ|char* s1|char* s2
412 Test two C<NUL>-terminated strings to see if they are equal. Returns true or
415 =for apidoc Am|bool|strLT|char* s1|char* s2
416 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
417 second, C<s2>. Returns true or false.
419 =for apidoc Am|bool|strLE|char* s1|char* s2
420 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
421 equal to the second, C<s2>. Returns true or false.
423 =for apidoc Am|bool|strGT|char* s1|char* s2
424 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
425 the second, C<s2>. Returns true or false.
427 =for apidoc Am|bool|strGE|char* s1|char* s2
428 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
429 or equal to the second, C<s2>. Returns true or false.
431 =for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
432 Test two C<NUL>-terminated strings to see if they are different. The C<len>
433 parameter indicates the number of bytes to compare. Returns true or false. (A
434 wrapper for C<strncmp>).
436 =for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
437 Test two C<NUL>-terminated strings to see if they are equal. The C<len>
438 parameter indicates the number of bytes to compare. Returns true or false. (A
439 wrapper for C<strncmp>).
441 =for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
442 Test two buffers (which may contain embedded C<NUL> characters, to see if they
443 are equal. The C<len> parameter indicates the number of bytes to compare.
444 Returns zero if equal, or non-zero if non-equal.
446 =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
447 Test two buffers (which may contain embedded C<NUL> characters, to see if they
448 are not equal. The C<len> parameter indicates the number of bytes to compare.
449 Returns zero if non-equal, or non-zero if equal.
453 New macros should use the following conventions for their names (which are
454 based on the underlying C library functions):
456 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
458 Each has two main parameters, string-like operands that are compared
459 against each other, as specified by the macro name. Some macros may
460 additionally have one or potentially even two length parameters. If a length
461 parameter applies to both string parameters, it will be positioned third;
462 otherwise any length parameter immediately follows the string parameter it
465 If the prefix to the name is 'str', the string parameter is a pointer to a C
466 language string. Such a string does not contain embedded NUL bytes; its
467 length may be unknown, but can be calculated by C<strlen()>, since it is
468 terminated by a NUL, which isn't included in its length.
470 The optional 'n' following 'str' means that that there is a third parameter,
471 giving the maximum number of bytes to look at in each string. Even if both
472 strings are longer than the length parameter, those extra bytes will be
475 The 's' suffix means that the 2nd byte string parameter is a literal C
476 double-quoted string. Its length will automatically be calculated by the
477 macro, so no length parameter will ever be needed for it.
479 If the prefix is 'mem', the string parameters don't have to be C strings;
480 they may contain embedded NUL bytes, do not necessarily have a terminating
481 NUL, and their lengths can be known only through other means, which in
482 practice are additional parameter(s) passed to the function. All 'mem'
483 functions have at least one length parameter. Barring any 'l' or 's' suffix,
484 there is a single length parameter, in position 3, which applies to both
485 string parameters. The 's' suffix means, as described above, that the 2nd
486 string is a literal double-quoted C string (hence its length is calculated by
487 the macro, and the length parameter to the function applies just to the first
488 string parameter, and hence is positioned just after it). An 'l' suffix
489 means that the 2nd string parameter has its own length parameter, and the
490 signature will look like memFOOl(s1, l1, s2, l2).
492 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
493 substring of the 1st string. 'P' if present indicates that the substring
494 must be a "proper" one in tha mathematical sense that the first one must be
495 strictly larger than the 2nd.
500 #define strNE(s1,s2) (strcmp(s1,s2) != 0)
501 #define strEQ(s1,s2) (strcmp(s1,s2) == 0)
502 #define strLT(s1,s2) (strcmp(s1,s2) < 0)
503 #define strLE(s1,s2) (strcmp(s1,s2) <= 0)
504 #define strGT(s1,s2) (strcmp(s1,s2) > 0)
505 #define strGE(s1,s2) (strcmp(s1,s2) >= 0)
507 #define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
508 #define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
510 #define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
511 #define memNE(s1,s2,l) (! memEQ(s1,s2,l))
513 /* memEQ and memNE where second comparand is a string constant */
514 #define memEQs(s1, l, s2) \
515 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
516 #define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
518 /* Keep these private until we decide it was a good idea */
519 #if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
521 #define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
523 #define memBEGINs(s1, l, s2) \
524 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
525 && memEQ(s1, "" s2 "", sizeof(s2)-1))
526 #define memBEGINPs(s1, l, s2) \
527 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
528 && memEQ(s1, "" s2 "", sizeof(s2)-1))
529 #define memENDs(s1, l, s2) \
530 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
531 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
532 #define memENDPs(s1, l, s2) \
533 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
534 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
535 #endif /* End of making macros private */
537 #define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
538 #define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
539 #define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
540 #define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
545 * Unfortunately, the introduction of locales means that we
546 * can't trust isupper(), etc. to tell the truth. And when
547 * it comes to /\w+/ with tainting enabled, we *must* be able
548 * to trust our character classes.
550 * Therefore, the default tests in the text of Perl will be
551 * independent of locale. Any code that wants to depend on
552 * the current locale will use the tests that begin with "lc".
555 #ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
563 =head1 Character classification
564 This section is about functions (really macros) that classify characters
565 into types, such as punctuation versus alphabetic, etc. Most of these are
566 analogous to regular expression character classes. (See
567 L<perlrecharclass/POSIX Character Classes>.) There are several variants for
568 each class. (Not all macros have all variants; each item below lists the
569 ones valid for it.) None are affected by C<use bytes>, and only the ones
570 with C<LC> in the name are affected by the current locale.
572 The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
573 C<U8>) as input and returns a boolean as to whether or not the character
574 represented by that octet is (or on non-ASCII platforms, corresponds to) an
575 ASCII character in the named class based on platform, Unicode, and Perl rules.
576 If the input is a number that doesn't fit in an octet, FALSE is returned.
578 Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
579 with no suffix C<"_A">. This variant is used to emphasize by its name that
580 only ASCII-range characters can return TRUE.
582 Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
583 onto the platform. That is, the code points that are ASCII are unaffected,
584 since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
585 as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
586 true when called with the code point 0xDF, which is a word character in both
587 ASCII and EBCDIC (though it represents different characters in each).
589 Variant C<isI<FOO>_uvchr> is like the C<isI<FOO>_L1> variant, but accepts any UV code
590 point as input. If the code point is larger than 255, Unicode rules are used
591 to determine if it is in the character class. For example,
592 C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
593 WITH MACRON in Unicode, and is a word character.
595 Variant C<isI<FOO>_utf8_safe> is like C<isI<FOO>_uvchr>, but is used for UTF-8
596 encoded strings. Each call classifies one character, even if the string
597 contains many. This variant takes two parameters. The first, C<p>, is a
598 pointer to the first byte of the character to be classified. (Recall that it
599 may take more than one byte to represent a character in UTF-8 strings.) The
600 second parameter, C<e>, points to anywhere in the string beyond the first
601 character, up to one byte past the end of the entire string. The suffix
602 C<_safe> in the function's name indicates that it will not attempt to read
603 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
604 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
605 character is malformed in some way, the program may croak, or the function may
606 return FALSE, at the discretion of the implementation, and subject to change in
609 Variant C<isI<FOO>_utf8> is like C<isI<FOO>_utf8_safe>, but takes just a single
610 parameter, C<p>, which has the same meaning as the corresponding parameter does
611 in C<isI<FOO>_utf8_safe>. The function therefore can't check if it is reading
612 beyond the end of the string. Starting in Perl v5.30, it will take a second
613 parameter, becoming a synonym for C<isI<FOO>_utf8_safe>. At that time every
614 program that uses it will have to be changed to successfully compile. In the
615 meantime, the first runtime call to C<isI<FOO>_utf8> from each call point in the
616 program will raise a deprecation warning, enabled by default. You can convert
617 your program now to use C<isI<FOO>_utf8_safe>, and avoid the warnings, and get an
618 extra measure of protection, or you can wait until v5.30, when you'll be forced
619 to add the C<e> parameter.
621 Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants, but the
622 result is based on the current locale, which is what C<LC> in the name stands
623 for. If Perl can determine that the current locale is a UTF-8 locale, it uses
624 the published Unicode rules; otherwise, it uses the C library function that
625 gives the named classification. For example, C<isDIGIT_LC()> when not in a
626 UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
627 returned if the input won't fit into an octet. On some platforms where the C
628 library function is known to be defective, Perl changes its result to follow
629 the POSIX standard's rules.
631 Variant C<isI<FOO>_LC_uvchr> is like C<isI<FOO>_LC>, but is defined on any UV. It
632 returns the same as C<isI<FOO>_LC> for input code points less than 256, and
633 returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
635 Variant C<isI<FOO>_LC_utf8_safe> is like C<isI<FOO>_LC_uvchr>, but is used for UTF-8
636 encoded strings. Each call classifies one character, even if the string
637 contains many. This variant takes two parameters. The first, C<p>, is a
638 pointer to the first byte of the character to be classified. (Recall that it
639 may take more than one byte to represent a character in UTF-8 strings.) The
640 second parameter, C<e>, points to anywhere in the string beyond the first
641 character, up to one byte past the end of the entire string. The suffix
642 C<_safe> in the function's name indicates that it will not attempt to read
643 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
644 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
645 character is malformed in some way, the program may croak, or the function may
646 return FALSE, at the discretion of the implementation, and subject to change in
649 Variant C<isI<FOO>_LC_utf8> is like C<isI<FOO>_LC_utf8_safe>, but takes just a single
650 parameter, C<p>, which has the same meaning as the corresponding parameter does
651 in C<isI<FOO>_LC_utf8_safe>. The function therefore can't check if it is reading
652 beyond the end of the string. Starting in Perl v5.30, it will take a second
653 parameter, becoming a synonym for C<isI<FOO>_LC_utf8_safe>. At that time every
654 program that uses it will have to be changed to successfully compile. In the
655 meantime, the first runtime call to C<isI<FOO>_LC_utf8> from each call point in
656 the program will raise a deprecation warning, enabled by default. You can
657 convert your program now to use C<isI<FOO>_LC_utf8_safe>, and avoid the warnings,
658 and get an extra measure of protection, or you can wait until v5.30, when
659 you'll be forced to add the C<e> parameter.
661 =for apidoc Am|bool|isALPHA|char ch
662 Returns a boolean indicating whether the specified character is an
663 alphabetic character, analogous to C<m/[[:alpha:]]/>.
664 See the L<top of this section|/Character classification> for an explanation of
666 C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8_safe>,
667 C<isALPHA_LC>, C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8_safe>.
669 =for apidoc Am|bool|isALPHANUMERIC|char ch
670 Returns a boolean indicating whether the specified character is a either an
671 alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
672 See the L<top of this section|/Character classification> for an explanation of
674 C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
675 C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>,
676 and C<isALPHANUMERIC_LC_utf8_safe>.
678 =for apidoc Am|bool|isASCII|char ch
679 Returns a boolean indicating whether the specified character is one of the 128
680 characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
681 On non-ASCII platforms, it returns TRUE iff this
682 character corresponds to an ASCII character. Variants C<isASCII_A()> and
683 C<isASCII_L1()> are identical to C<isASCII()>.
684 See the L<top of this section|/Character classification> for an explanation of
686 C<isASCII_uvchr>, C<isASCII_utf8_safe>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
687 C<isASCII_LC_utf8_safe>. Note, however, that some platforms do not have the C
688 library routine C<isascii()>. In these cases, the variants whose names contain
689 C<LC> are the same as the corresponding ones without.
691 Also note, that because all ASCII characters are UTF-8 invariant (meaning they
692 have the exact same representation (always a single byte) whether encoded in
693 UTF-8 or not), C<isASCII> will give the correct results when called with any
694 byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8_safe>
695 will work properly on any string encoded or not in UTF-8.
697 =for apidoc Am|bool|isBLANK|char ch
698 Returns a boolean indicating whether the specified character is a
699 character considered to be a blank, analogous to C<m/[[:blank:]]/>.
700 See the L<top of this section|/Character classification> for an explanation of
702 C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8_safe>,
703 C<isBLANK_LC>, C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8_safe>. Note,
704 however, that some platforms do not have the C library routine
705 C<isblank()>. In these cases, the variants whose names contain C<LC> are
706 the same as the corresponding ones without.
708 =for apidoc Am|bool|isCNTRL|char ch
709 Returns a boolean indicating whether the specified character is a
710 control character, analogous to C<m/[[:cntrl:]]/>.
711 See the L<top of this section|/Character classification> for an explanation of
713 C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8_safe>,
714 C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8_safe> On EBCDIC
715 platforms, you almost always want to use the C<isCNTRL_L1> variant.
717 =for apidoc Am|bool|isDIGIT|char ch
718 Returns a boolean indicating whether the specified character is a
719 digit, analogous to C<m/[[:digit:]]/>.
720 Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
721 See the L<top of this section|/Character classification> for an explanation of
723 C<isDIGIT_uvchr>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
724 C<isDIGIT_LC_utf8_safe>.
726 =for apidoc Am|bool|isGRAPH|char ch
727 Returns a boolean indicating whether the specified character is a
728 graphic character, analogous to C<m/[[:graph:]]/>.
729 See the L<top of this section|/Character classification> for an explanation of
730 variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8_safe>,
731 C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8_safe>.
733 =for apidoc Am|bool|isLOWER|char ch
734 Returns a boolean indicating whether the specified character is a
735 lowercase character, analogous to C<m/[[:lower:]]/>.
736 See the L<top of this section|/Character classification> for an explanation of
738 C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8_safe>,
739 C<isLOWER_LC>, C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8_safe>.
741 =for apidoc Am|bool|isOCTAL|char ch
742 Returns a boolean indicating whether the specified character is an
744 The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
747 =for apidoc Am|bool|isPUNCT|char ch
748 Returns a boolean indicating whether the specified character is a
749 punctuation character, analogous to C<m/[[:punct:]]/>.
750 Note that the definition of what is punctuation isn't as
751 straightforward as one might desire. See L<perlrecharclass/POSIX Character
752 Classes> for details.
753 See the L<top of this section|/Character classification> for an explanation of
754 variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8_safe>,
755 C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8_safe>.
757 =for apidoc Am|bool|isSPACE|char ch
758 Returns a boolean indicating whether the specified character is a
759 whitespace character. This is analogous
760 to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
761 this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
762 locale forms of this macro (the ones with C<LC> in their names) matched
763 precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
764 in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
765 (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
766 See the L<top of this section|/Character classification> for an explanation of
768 C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8_safe>,
769 C<isSPACE_LC>, C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8_safe>.
771 =for apidoc Am|bool|isPSXSPC|char ch
772 (short for Posix Space)
773 Starting in 5.18, this is identical in all its forms to the
774 corresponding C<isSPACE()> macros.
775 The locale forms of this macro are identical to their corresponding
776 C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
777 non-locale forms differ from their C<isSPACE()> forms only in that the
778 C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
779 Otherwise they are identical. Thus this macro is analogous to what
780 C<m/[[:space:]]/> matches in a regular expression.
781 See the L<top of this section|/Character classification> for an explanation of
782 variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8_safe>,
783 C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8_safe>.
785 =for apidoc Am|bool|isUPPER|char ch
786 Returns a boolean indicating whether the specified character is an
787 uppercase character, analogous to C<m/[[:upper:]]/>.
788 See the L<top of this section|/Character classification> for an explanation of
789 variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8_safe>,
790 C<isUPPER_LC>, C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8_safe>.
792 =for apidoc Am|bool|isPRINT|char ch
793 Returns a boolean indicating whether the specified character is a
794 printable character, analogous to C<m/[[:print:]]/>.
795 See the L<top of this section|/Character classification> for an explanation of
797 C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8_safe>,
798 C<isPRINT_LC>, C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8_safe>.
800 =for apidoc Am|bool|isWORDCHAR|char ch
801 Returns a boolean indicating whether the specified character is a character
802 that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
803 in a regular expression. A word character is an alphabetic character, a
804 decimal digit, a connecting punctuation character (such as an underscore), or
805 a "mark" character that attaches to one of those (like some sort of accent).
806 C<isALNUM()> is a synonym provided for backward compatibility, even though a
807 word character includes more than the standard C language meaning of
809 See the L<top of this section|/Character classification> for an explanation of
810 variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and
811 C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and
812 C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
813 include the platform's native underscore.
815 =for apidoc Am|bool|isXDIGIT|char ch
816 Returns a boolean indicating whether the specified character is a hexadecimal
817 digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
818 and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
819 See the L<top of this section|/Character classification> for an explanation of
821 C<isXDIGIT_uvchr>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>,
822 and C<isXDIGIT_LC_utf8_safe>.
824 =for apidoc Am|bool|isIDFIRST|char ch
825 Returns a boolean indicating whether the specified character can be the first
826 character of an identifier. This is very close to, but not quite the same as
827 the official Unicode property C<XID_Start>. The difference is that this
828 returns true only if the input character also matches L</isWORDCHAR>.
829 See the L<top of this section|/Character classification> for an explanation of
831 C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8_safe>,
832 C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8_safe>.
834 =for apidoc Am|bool|isIDCONT|char ch
835 Returns a boolean indicating whether the specified character can be the
836 second or succeeding character of an identifier. This is very close to, but
837 not quite the same as the official Unicode property C<XID_Continue>. The
838 difference is that this returns true only if the input character also matches
839 L</isWORDCHAR>. See the L<top of this section|/Character classification> for
841 explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
842 C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
843 C<isIDCONT_LC_utf8_safe>.
845 =head1 Miscellaneous Functions
847 =for apidoc Am|U8|READ_XDIGIT|char str*
848 Returns the value of an ASCII-range hex digit and advances the string pointer.
849 Behaviour is only well defined when isXDIGIT(*str) is true.
851 =head1 Character case changing
852 Perl uses "full" Unicode case mappings. This means that converting a single
853 character to another case may result in a sequence of more than one character.
854 For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
855 character sequence C<SS>. This presents some complications The lowercase of
856 all characters in the range 0..255 is a single character, and thus
857 C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
858 return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
859 an API that does allow every possible legal result to be returned.) Likewise
860 no other function that is crippled by not being able to give the correct
861 results for the full range of possible inputs has been implemented here.
863 =for apidoc Am|U8|toUPPER|U8 ch
864 Converts the specified character to uppercase. If the input is anything but an
865 ASCII lowercase character, that input character itself is returned. Variant
866 C<toUPPER_A> is equivalent.
868 =for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
869 Converts the code point C<cp> to its uppercase version, and
870 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
871 point is interpreted as native if less than 256; otherwise as Unicode. Note
872 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
873 bytes since the uppercase version may be longer than the original character.
875 The first code point of the uppercased version is returned
876 (but note, as explained at L<the top of this section|/Character case
877 changing>, that there may be more.)
879 =for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
880 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
881 extending no further than S<C<e - 1>> to its uppercase version, and
882 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
883 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
884 bytes since the uppercase version may be longer than the original character.
886 The first code point of the uppercased version is returned
887 (but note, as explained at L<the top of this section|/Character case
888 changing>, that there may be more).
890 The suffix C<_safe> in the function's name indicates that it will not attempt
891 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
892 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
893 input character is malformed in some way, the program may croak, or the
894 function may return the REPLACEMENT CHARACTER, at the discretion of the
895 implementation, and subject to change in future releases.
897 =for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
898 This is like C<L</toUPPER_utf8_safe>>, but doesn't have the C<e>
899 parameter The function therefore can't check if it is reading
900 beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
901 parameter, becoming a synonym for C<toUPPER_utf8_safe>. At that time every
902 program that uses it will have to be changed to successfully compile. In the
903 meantime, the first runtime call to C<toUPPER_utf8> from each call point in the
904 program will raise a deprecation warning, enabled by default. You can convert
905 your program now to use C<toUPPER_utf8_safe>, and avoid the warnings, and get an
906 extra measure of protection, or you can wait until v5.30, when you'll be forced
907 to add the C<e> parameter.
909 =for apidoc Am|U8|toFOLD|U8 ch
910 Converts the specified character to foldcase. If the input is anything but an
911 ASCII uppercase character, that input character itself is returned. Variant
912 C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
913 Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
915 =for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
916 Converts the code point C<cp> to its foldcase version, and
917 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
918 point is interpreted as native if less than 256; otherwise as Unicode. Note
919 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
920 bytes since the foldcase version may be longer than the original character.
922 The first code point of the foldcased version is returned
923 (but note, as explained at L<the top of this section|/Character case
924 changing>, that there may be more).
926 =for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
927 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
928 extending no further than S<C<e - 1>> to its foldcase version, and
929 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
930 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
931 bytes since the foldcase version may be longer than the original character.
933 The first code point of the foldcased version is returned
934 (but note, as explained at L<the top of this section|/Character case
935 changing>, that there may be more).
937 The suffix C<_safe> in the function's name indicates that it will not attempt
938 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
939 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
940 input character is malformed in some way, the program may croak, or the
941 function may return the REPLACEMENT CHARACTER, at the discretion of the
942 implementation, and subject to change in future releases.
944 =for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
945 This is like C<L</toFOLD_utf8_safe>>, but doesn't have the C<e>
946 parameter The function therefore can't check if it is reading
947 beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
948 parameter, becoming a synonym for C<toFOLD_utf8_safe>. At that time every
949 program that uses it will have to be changed to successfully compile. In the
950 meantime, the first runtime call to C<toFOLD_utf8> from each call point in the
951 program will raise a deprecation warning, enabled by default. You can convert
952 your program now to use C<toFOLD_utf8_safe>, and avoid the warnings, and get an
953 extra measure of protection, or you can wait until v5.30, when you'll be forced
954 to add the C<e> parameter.
956 =for apidoc Am|U8|toLOWER|U8 ch
957 Converts the specified character to lowercase. If the input is anything but an
958 ASCII uppercase character, that input character itself is returned. Variant
959 C<toLOWER_A> is equivalent.
961 =for apidoc Am|U8|toLOWER_L1|U8 ch
962 Converts the specified Latin1 character to lowercase. The results are
963 undefined if the input doesn't fit in a byte.
965 =for apidoc Am|U8|toLOWER_LC|U8 ch
966 Converts the specified character to lowercase using the current locale's rules,
967 if possible; otherwise returns the input character itself.
969 =for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
970 Converts the code point C<cp> to its lowercase version, and
971 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
972 point is interpreted as native if less than 256; otherwise as Unicode. Note
973 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
974 bytes since the lowercase version may be longer than the original character.
976 The first code point of the lowercased version is returned
977 (but note, as explained at L<the top of this section|/Character case
978 changing>, that there may be more).
981 =for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
982 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
983 extending no further than S<C<e - 1>> to its lowercase version, and
984 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
985 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
986 bytes since the lowercase version may be longer than the original character.
988 The first code point of the lowercased version is returned
989 (but note, as explained at L<the top of this section|/Character case
990 changing>, that there may be more).
992 The suffix C<_safe> in the function's name indicates that it will not attempt
993 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
994 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
995 input character is malformed in some way, the program may croak, or the
996 function may return the REPLACEMENT CHARACTER, at the discretion of the
997 implementation, and subject to change in future releases.
999 =for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
1000 This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1001 parameter The function therefore can't check if it is reading
1002 beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
1003 parameter, becoming a synonym for C<toLOWER_utf8_safe>. At that time every
1004 program that uses it will have to be changed to successfully compile. In the
1005 meantime, the first runtime call to C<toLOWER_utf8> from each call point in the
1006 program will raise a deprecation warning, enabled by default. You can convert
1007 your program now to use C<toLOWER_utf8_safe>, and avoid the warnings, and get an
1008 extra measure of protection, or you can wait until v5.30, when you'll be forced
1009 to add the C<e> parameter.
1011 =for apidoc Am|U8|toTITLE|U8 ch
1012 Converts the specified character to titlecase. If the input is anything but an
1013 ASCII lowercase character, that input character itself is returned. Variant
1014 C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
1015 range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
1016 not a concept used in locale handling, so there is no functionality for that.)
1018 =for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1019 Converts the code point C<cp> to its titlecase version, and
1020 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1021 point is interpreted as native if less than 256; otherwise as Unicode. Note
1022 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1023 bytes since the titlecase version may be longer than the original character.
1025 The first code point of the titlecased version is returned
1026 (but note, as explained at L<the top of this section|/Character case
1027 changing>, that there may be more).
1029 =for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1030 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1031 extending no further than S<C<e - 1>> to its titlecase version, and
1032 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1033 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1034 bytes since the titlecase version may be longer than the original character.
1036 The first code point of the titlecased version is returned
1037 (but note, as explained at L<the top of this section|/Character case
1038 changing>, that there may be more).
1040 The suffix C<_safe> in the function's name indicates that it will not attempt
1041 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1042 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1043 input character is malformed in some way, the program may croak, or the
1044 function may return the REPLACEMENT CHARACTER, at the discretion of the
1045 implementation, and subject to change in future releases.
1047 =for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
1048 This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1049 parameter The function therefore can't check if it is reading
1050 beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
1051 parameter, becoming a synonym for C<toTITLE_utf8_safe>. At that time every
1052 program that uses it will have to be changed to successfully compile. In the
1053 meantime, the first runtime call to C<toTITLE_utf8> from each call point in the
1054 program will raise a deprecation warning, enabled by default. You can convert
1055 your program now to use C<toTITLE_utf8_safe>, and avoid the warnings, and get an
1056 extra measure of protection, or you can wait until v5.30, when you'll be forced
1057 to add the C<e> parameter.
1061 XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1062 really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1063 and aren't general purpose as they don't work on U+DF, and assert against that.
1065 Note that these macros are repeated in Devel::PPPort, so should also be
1066 patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1070 /* Specify the widest unsigned type on the platform. */
1072 # define WIDEST_UTYPE U64
1074 # define WIDEST_UTYPE U32
1077 /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1078 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1079 * bits of information are lost even on a 64-bit machine, but to get the
1080 * compiler to optimize it out if possible. This is because Configure makes
1081 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1082 * sizeof() guarantees that this evaluates to a constant true at compile time.
1084 * For Coverity, be always true, because otherwise Coverity thinks
1085 * it finds several expressions that are always true, independent
1086 * of operands. Well, they are, but that is kind of the point.
1088 #ifndef __COVERITY__
1089 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1091 #define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
1092 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
1094 #define FITS_IN_8_BITS(c) (1)
1097 /* Returns true if c is in the range l..u
1098 * Written with the cast so it only needs one conditional test
1100 #define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
1101 ((WIDEST_UTYPE) (((c) - (l)) | 0) <= ((WIDEST_UTYPE) ((u) - (l)))))
1104 # ifndef _ALL_SOURCE
1105 /* The native libc isascii() et.al. functions return the wrong results
1106 * on at least z/OS unless this is defined. */
1107 # error _ALL_SOURCE should probably be defined
1110 /* There is a simple definition of ASCII for ASCII platforms. But the
1111 * EBCDIC one isn't so simple, so is defined using table look-up like the
1112 * other macros below.
1114 * The cast here is used instead of '(c) >= 0', because some compilers emit
1115 * a warning that that test is always true when the parameter is an
1116 * unsigned type. khw supposes that it could be written as
1117 * && ((c) == '\0' || (c) > 0)
1118 * to avoid the message, but the cast will likely avoid extra branches even
1119 * with stupid compilers.
1121 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1123 # define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
1126 /* Take the eight possible bit patterns of the lower 3 bits and you get the
1127 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1128 * can be ignored. If the rest match '0', we have an octal */
1129 #define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
1131 #ifdef H_PERL /* If have access to perl.h, lookup in its table */
1133 /* Character class numbers. For internal core Perl use only. The ones less
1134 * than 32 are used in PL_charclass[] and the ones up through the one that
1135 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1136 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1137 * their actual definitions are here. If that file has a name not used here,
1140 * The first group of these is ordered in what I (khw) estimate to be the
1141 * frequency of their use. This gives a slight edge to exiting a loop earlier
1142 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1143 * real Posix character class, and some (small) inefficiencies in regular
1144 * expression handling would be introduced by putting it in the middle of those
1145 * that are. Also, cntrl and ascii come after the others as it may be useful
1146 * to group these which have no members that match above Latin1, (or above
1147 * ASCII in the latter case) */
1149 # define _CC_WORDCHAR 0 /* \w and [:word:] */
1150 # define _CC_DIGIT 1 /* \d and [:digit:] */
1151 # define _CC_ALPHA 2 /* [:alpha:] */
1152 # define _CC_LOWER 3 /* [:lower:] */
1153 # define _CC_UPPER 4 /* [:upper:] */
1154 # define _CC_PUNCT 5 /* [:punct:] */
1155 # define _CC_PRINT 6 /* [:print:] */
1156 # define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1157 # define _CC_GRAPH 8 /* [:graph:] */
1158 # define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1159 # define _CC_SPACE 10 /* \s, [:space:] */
1160 # define _CC_PSXSPC _CC_SPACE /* XXX Temporary, can be removed
1161 when the deprecated isFOO_utf8()
1162 functions are removed */
1163 # define _CC_BLANK 11 /* [:blank:] */
1164 # define _CC_XDIGIT 12 /* [:xdigit:] */
1165 # define _CC_CNTRL 13 /* [:cntrl:] */
1166 # define _CC_ASCII 14 /* [:ascii:] */
1167 # define _CC_VERTSPACE 15 /* \v */
1169 # define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1171 /* The members of the third group below do not need to be coordinated with data
1172 * structures in regcomp.[ch] and regexec.c. */
1173 # define _CC_IDFIRST 16
1174 # define _CC_CHARNAME_CONT 17
1175 # define _CC_NONLATIN1_FOLD 18
1176 # define _CC_NONLATIN1_SIMPLE_FOLD 19
1177 # define _CC_QUOTEMETA 20
1178 # define _CC_NON_FINAL_FOLD 21
1179 # define _CC_IS_IN_SOME_FOLD 22
1180 # define _CC_MNEMONIC_CNTRL 23
1182 # define _CC_IDCONT 24 /* XXX Temporary, can be removed when the deprecated
1183 isFOO_utf8() functions are removed */
1185 /* This next group is only used on EBCDIC platforms, so theoretically could be
1186 * shared with something entirely different that's only on ASCII platforms */
1187 # define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1188 # define _CC_UTF8_IS_START 29
1189 # define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1190 # define _CC_UTF8_IS_CONTINUATION 31
1192 * If more bits are needed, one could add a second word for non-64bit
1193 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
1194 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1195 * is used only for optimization (as of this writing), and differs in the
1196 * Latin1 range from the ALPHA bit only in two relatively unimportant
1197 * characters: the masculine and feminine ordinal indicators, so removing it
1198 * would just cause /i regexes which match them to run less efficiently.
1199 * Similarly the EBCDIC-only bits are used just for speed, and could be
1200 * replaced by other means */
1202 #if defined(PERL_CORE) || defined(PERL_EXT)
1203 /* An enum version of the character class numbers, to help compilers
1206 _CC_ENUM_ALPHA = _CC_ALPHA,
1207 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1208 _CC_ENUM_ASCII = _CC_ASCII,
1209 _CC_ENUM_BLANK = _CC_BLANK,
1210 _CC_ENUM_CASED = _CC_CASED,
1211 _CC_ENUM_CNTRL = _CC_CNTRL,
1212 _CC_ENUM_DIGIT = _CC_DIGIT,
1213 _CC_ENUM_GRAPH = _CC_GRAPH,
1214 _CC_ENUM_LOWER = _CC_LOWER,
1215 _CC_ENUM_PRINT = _CC_PRINT,
1216 _CC_ENUM_PUNCT = _CC_PUNCT,
1217 _CC_ENUM_SPACE = _CC_SPACE,
1218 _CC_ENUM_UPPER = _CC_UPPER,
1219 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
1220 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
1221 _CC_ENUM_XDIGIT = _CC_XDIGIT
1222 } _char_class_number;
1225 #define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
1229 EXTCONST U32 PL_charclass[] = {
1230 # include "l1_char_class_tab.h"
1233 # else /* ! DOINIT */
1234 EXTCONST U32 PL_charclass[];
1238 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
1239 # define _CC_mask(classnum) (1U << (classnum))
1241 /* For internal core Perl use only: the base macro for defining macros like
1243 # define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
1244 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
1246 /* The mask for the _A versions of the macros; it just adds in the bit for
1248 # define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1250 /* For internal core Perl use only: the base macro for defining macros like
1251 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1252 * the ASCII bit are present */
1253 # define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1254 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1255 == _CC_mask_A(classnum)))
1257 # define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1258 # define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
1259 # define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1260 # define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
1261 # define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
1262 # define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1263 # define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1264 # define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1265 # define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1266 # define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1267 # define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1268 # define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
1269 # define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1271 # define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
1272 # define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1273 # define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1274 # define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1276 /* continuation character for legal NAME in \N{NAME} */
1277 # define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1279 # define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1280 # define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1281 # define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1282 # define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1283 # define isPSXSPC_L1(c) isSPACE_L1(c)
1284 # define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1285 # define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1286 # define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1287 # define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1288 # define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
1291 # define isASCII(c) _generic_isCC(c, _CC_ASCII)
1294 /* Participates in a single-character fold with a character above 255 */
1295 # 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)))
1297 /* Like the above, but also can be part of a multi-char fold */
1298 # 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)))
1300 # define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
1301 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1302 _generic_isCC(c, _CC_NON_FINAL_FOLD)
1303 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1304 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1305 # define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1306 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1307 #else /* else we don't have perl.h H_PERL */
1309 /* If we don't have perl.h, we are compiling a utility program. Below we
1310 * hard-code various macro definitions that wouldn't otherwise be available
1311 * to it. Most are coded based on first principles. These are written to
1312 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
1313 # define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1314 # define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1315 # define isSPACE_A(c) (isBLANK_A(c) \
1320 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1321 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1322 * performance (if optimization isn't on), and allow the same code to be
1323 * used for both platform types */
1324 # define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1326 || ((c) >= 'j' && (c) <= 'r') \
1328 # define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1330 || ((c) >= 'J' && (c) <= 'R') \
1332 # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1333 # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
1334 # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
1335 # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1336 # define isXDIGIT_A(c) (isDIGIT_A(c) \
1337 || ((c) >= 'a' && (c) <= 'f') \
1338 || ((c) <= 'F' && (c) >= 'A'))
1339 # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1340 || (c) == '#' || (c) == '$' || (c) == '%' \
1341 || (c) == '&' || (c) == '\'' || (c) == '(' \
1342 || (c) == ')' || (c) == '*' || (c) == '+' \
1343 || (c) == ',' || (c) == '.' || (c) == '/' \
1344 || (c) == ':' || (c) == ';' || (c) == '<' \
1345 || (c) == '=' || (c) == '>' || (c) == '?' \
1346 || (c) == '@' || (c) == '[' || (c) == '\\' \
1347 || (c) == ']' || (c) == '^' || (c) == '_' \
1348 || (c) == '`' || (c) == '{' || (c) == '|' \
1349 || (c) == '}' || (c) == '~')
1350 # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1351 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1354 /* The below is accurate for the 3 EBCDIC code pages traditionally
1355 * supported by perl. The only difference between them in the controls
1356 * is the position of \n, and that is represented symbolically below */
1357 # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1358 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1359 || (c) == '\t' || (c) == '\v' \
1360 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1361 || (c) == 7 /* U+7F DEL */ \
1362 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1363 /* DLE, DC[1-3] */ \
1364 || (c) == 0x18 /* U+18 CAN */ \
1365 || (c) == 0x19 /* U+19 EOM */ \
1366 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1367 || (c) == 0x26 /* U+17 ETB */ \
1368 || (c) == 0x27 /* U+1B ESC */ \
1369 || (c) == 0x2D /* U+05 ENQ */ \
1370 || (c) == 0x2E /* U+06 ACK */ \
1371 || (c) == 0x32 /* U+16 SYN */ \
1372 || (c) == 0x37 /* U+04 EOT */ \
1373 || (c) == 0x3C /* U+14 DC4 */ \
1374 || (c) == 0x3D /* U+15 NAK */ \
1375 || (c) == 0x3F)/* U+1A SUB */
1376 # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1377 # else /* isASCII is already defined for ASCII platforms, so can use that to
1379 # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
1382 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
1383 * during debugging, and it seems best to keep them. We may be called
1384 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1385 * do anything anyway, so make it not a problem */
1386 # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1387 # define NATIVE_TO_LATIN1(ch) (ch)
1389 # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1390 # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1391 # define isBLANK_L1(c) (isBLANK_A(c) \
1392 || (FITS_IN_8_BITS(c) \
1393 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1394 # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1395 # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1396 # define isLOWER_L1(c) (isLOWER_A(c) \
1397 || (FITS_IN_8_BITS(c) \
1398 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1399 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1400 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1401 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1402 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1403 # define isPRINT_L1(c) (isPRINT_A(c) \
1404 || (FITS_IN_8_BITS(c) \
1405 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1406 # define isPUNCT_L1(c) (isPUNCT_A(c) \
1407 || (FITS_IN_8_BITS(c) \
1408 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1409 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1410 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1411 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1412 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1413 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1414 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1415 # define isSPACE_L1(c) (isSPACE_A(c) \
1416 || (FITS_IN_8_BITS(c) \
1417 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
1418 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1419 # define isUPPER_L1(c) (isUPPER_A(c) \
1420 || (FITS_IN_8_BITS(c) \
1421 && ( NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1422 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1423 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1424 # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1425 # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1426 # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1431 /* The following are not fully accurate in the above-ASCII range. I (khw)
1432 * don't think it's necessary to be so for the purposes where this gets
1434 # define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1435 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1437 /* And these aren't accurate at all. They are useful only for above
1438 * Latin1, which utilities and bootstrapping don't deal with */
1439 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
1440 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1441 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1443 /* Many of the macros later in this file are defined in terms of these. By
1444 * implementing them with a function, which converts the class number into
1445 * a call to the desired macro, all of the later ones work. However, that
1446 * function won't be actually defined when building a utility program (no
1447 * perl.h), and so a compiler error will be generated if one is attempted
1448 * to be used. And the above-Latin1 code points require Unicode tables to
1449 * be present, something unlikely to be the case when bootstrapping */
1450 # define _generic_isCC(c, classnum) \
1451 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1452 # define _generic_isCC_A(c, classnum) \
1453 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
1454 #endif /* End of no perl.h H_PERL */
1456 #define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1457 #define isALPHA(c) isALPHA_A(c)
1458 #define isASCII_A(c) isASCII(c)
1459 #define isASCII_L1(c) isASCII(c)
1460 #define isBLANK(c) isBLANK_A(c)
1461 #define isCNTRL(c) isCNTRL_A(c)
1462 #define isDIGIT(c) isDIGIT_A(c)
1463 #define isGRAPH(c) isGRAPH_A(c)
1464 #define isIDFIRST(c) isIDFIRST_A(c)
1465 #define isLOWER(c) isLOWER_A(c)
1466 #define isPRINT(c) isPRINT_A(c)
1467 #define isPSXSPC_A(c) isSPACE_A(c)
1468 #define isPSXSPC(c) isPSXSPC_A(c)
1469 #define isPSXSPC_L1(c) isSPACE_L1(c)
1470 #define isPUNCT(c) isPUNCT_A(c)
1471 #define isSPACE(c) isSPACE_A(c)
1472 #define isUPPER(c) isUPPER_A(c)
1473 #define isWORDCHAR(c) isWORDCHAR_A(c)
1474 #define isXDIGIT(c) isXDIGIT_A(c)
1476 /* ASCII casing. These could also be written as
1477 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1478 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1479 which uses table lookup and mask instead of subtraction. (This would
1480 work because the _MOD does not apply in the ASCII range).
1482 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1483 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1484 there non-ASCII invariants, and all of them are controls.) */
1485 #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1486 #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
1488 /* In the ASCII range, these are equivalent to what they're here defined to be.
1489 * But by creating these definitions, other code doesn't have to be aware of
1490 * this detail. Actually this works for all UTF-8 invariants, not just the
1491 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
1492 #define toFOLD(c) toLOWER(c)
1493 #define toTITLE(c) toUPPER(c)
1495 #define toLOWER_A(c) toLOWER(c)
1496 #define toUPPER_A(c) toUPPER(c)
1497 #define toFOLD_A(c) toFOLD(c)
1498 #define toTITLE_A(c) toTITLE(c)
1500 /* Use table lookup for speed; returns the input itself if is out-of-range */
1501 #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
1503 : PL_latin1_lc[ (U8) (c) ])
1504 #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1506 /* Modified uc. Is correct uc except for three non-ascii chars which are
1507 * all mapped to one of them, and these need special handling; returns the
1508 * input itself if is out-of-range */
1509 #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
1511 : PL_mod_latin1_uc[ (U8) (c) ])
1512 #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
1514 /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1516 /* For internal core Perl use only: the base macro for defining macros like
1517 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
1518 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1519 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1520 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1521 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1522 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1523 * code points under locale rules, use the macros based on _generic_LC_uvchr
1524 * instead of this. */
1525 #define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1526 (! FITS_IN_8_BITS(c) \
1528 : IN_UTF8_CTYPE_LOCALE \
1529 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
1532 /* For internal core Perl use only: a helper macro for defining macros like
1533 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1534 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1535 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1536 * _generic_LC_base for more info */
1537 #define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1538 _generic_LC_base(c,utf8_locale_classnum, \
1539 non_utf8_func( (_LC_CAST) (c)))
1541 /* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1542 * 'c' is the platform's native underscore character */
1543 #define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1544 _generic_LC_base(c, utf8_locale_classnum, \
1545 (non_utf8_func( (_LC_CAST) (c)) \
1546 || (char)(c) == '_'))
1548 /* These next three are also for internal core Perl use only: case-change
1549 * helper macros. The reason for using the PL_latin arrays is in case the
1550 * system function is defective; it ensures uniform results that conform to the
1551 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
1552 #define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1554 : (IN_UTF8_CTYPE_LOCALE) \
1555 ? PL_latin1_lc[ (U8) (c) ] \
1556 : (cast)function((cast)(c)))
1558 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1559 * returns a single value, so can't adequately return the upper case of LATIN
1560 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1561 * values "SS"); instead it asserts against that under DEBUGGING, and
1562 * otherwise returns its input. It does not handle the anomalies in UTF-8
1563 * Turkic locales. */
1564 #define _generic_toUPPER_LC(c, function, cast) \
1565 (! FITS_IN_8_BITS(c) \
1567 : ((! IN_UTF8_CTYPE_LOCALE) \
1568 ? (cast)function((cast)(c)) \
1569 : ((((U8)(c)) == MICRO_SIGN) \
1570 ? GREEK_CAPITAL_LETTER_MU \
1571 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1572 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1573 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1574 ? (__ASSERT_(0) (c)) \
1575 : PL_mod_latin1_uc[ (U8) (c) ])))))
1577 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1578 * returns a single value, so can't adequately return the fold case of LATIN
1579 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1580 * values "ss"); instead it asserts against that under DEBUGGING, and
1581 * otherwise returns its input. It does not handle the anomalies in UTF-8
1583 #define _generic_toFOLD_LC(c, function, cast) \
1584 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
1585 ? GREEK_SMALL_LETTER_MU \
1586 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1587 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1588 _generic_toLOWER_LC(c, function, cast)))
1590 /* Use the libc versions for these if available. */
1591 #if defined(HAS_ISASCII)
1592 # define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1594 # define isASCII_LC(c) isASCII(c)
1597 #if defined(HAS_ISBLANK)
1598 # define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
1599 #else /* Unlike isASCII, varies if in a UTF-8 locale */
1600 # define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
1606 /* The Windows functions don't bother to follow the POSIX standard, which
1607 * for example says that something can't both be a printable and a control.
1608 * But Windows treats the \t control as a printable, and does such things
1609 * as making superscripts into both digits and punctuation. This tames
1610 * these flaws by assuming that the definitions of both controls and space
1611 * are correct, and then making sure that other definitions don't have
1612 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1613 * Not all possible weirdnesses are checked for, just the ones that were
1614 * detected on actual Microsoft code pages */
1616 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1617 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1619 # define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1620 && isALPHANUMERIC_LC(c))
1621 # define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1623 # define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1624 isALPHANUMERIC_LC(c))
1625 # define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1626 # define isIDFIRST_LC(c) (((c) == '_') \
1627 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1628 # define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1629 # define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1630 # define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1631 # define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1632 # define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1633 # define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1634 && isALPHANUMERIC_LC(c))
1636 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1637 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1638 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1640 #elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
1641 /* For most other platforms */
1643 # define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1644 # define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1645 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1646 # define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1647 # define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1648 # define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1649 # define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1650 # define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1651 # define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1652 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1653 # define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1654 # define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1655 # define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1658 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1659 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1660 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1662 #else /* The final fallback position */
1664 # define isALPHA_LC(c) (isascii(c) && isalpha(c))
1665 # define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1666 # define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1667 # define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1668 # define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1669 # define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1670 # define isLOWER_LC(c) (isascii(c) && islower(c))
1671 # define isPRINT_LC(c) (isascii(c) && isprint(c))
1672 # define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1673 # define isSPACE_LC(c) (isascii(c) && isspace(c))
1674 # define isUPPER_LC(c) (isascii(c) && isupper(c))
1675 # define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1676 # define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1678 # define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1679 # define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1680 # define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
1684 #define isIDCONT(c) isWORDCHAR(c)
1685 #define isIDCONT_A(c) isWORDCHAR_A(c)
1686 #define isIDCONT_L1(c) isWORDCHAR_L1(c)
1687 #define isIDCONT_LC(c) isWORDCHAR_LC(c)
1688 #define isPSXSPC_LC(c) isSPACE_LC(c)
1690 /* For internal core Perl use only: the base macros for defining macros like
1691 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1692 * number defined earlier in this file. _generic_uvchr() is used for POSIX
1693 * classes where there is a macro or function 'above_latin1' that takes the
1694 * single argument 'c' and returns the desired value. These exist for those
1695 * classes which have simple definitions, avoiding the overhead of a hash
1696 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
1697 * for classes where that overhead is faster than a direct lookup.
1698 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
1699 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1700 * have duplicate checks here, so could create versions of the macros that
1701 * don't, but experiments show that gcc optimizes them out anyway. */
1703 /* Note that all ignore 'use bytes' */
1704 #define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1705 ? _generic_isCC(c, classnum) \
1707 #define _generic_swash_uvchr(classnum, c) ((c) < 256 \
1708 ? _generic_isCC(c, classnum) \
1709 : _is_uni_FOO(classnum, c))
1710 #define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1711 #define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1712 #define isASCII_uvchr(c) isASCII(c)
1713 #define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1714 #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1715 #define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1716 #define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1717 #define isIDCONT_uvchr(c) \
1718 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1719 #define isIDFIRST_uvchr(c) \
1720 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
1721 #define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1722 #define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1724 #define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1725 #define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1726 #define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1728 #define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1729 #define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1730 #define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1731 #define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1733 #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1734 #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1735 #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1736 #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1738 /* For backwards compatibility, even though '_uni' should mean official Unicode
1739 * code points, in Perl it means native for those below 256 */
1740 #define isALPHA_uni(c) isALPHA_uvchr(c)
1741 #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1742 #define isASCII_uni(c) isASCII_uvchr(c)
1743 #define isBLANK_uni(c) isBLANK_uvchr(c)
1744 #define isCNTRL_uni(c) isCNTRL_uvchr(c)
1745 #define isDIGIT_uni(c) isDIGIT_uvchr(c)
1746 #define isGRAPH_uni(c) isGRAPH_uvchr(c)
1747 #define isIDCONT_uni(c) isIDCONT_uvchr(c)
1748 #define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1749 #define isLOWER_uni(c) isLOWER_uvchr(c)
1750 #define isPRINT_uni(c) isPRINT_uvchr(c)
1751 #define isPUNCT_uni(c) isPUNCT_uvchr(c)
1752 #define isSPACE_uni(c) isSPACE_uvchr(c)
1753 #define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1754 #define isUPPER_uni(c) isUPPER_uvchr(c)
1755 #define isVERTWS_uni(c) isVERTWS_uvchr(c)
1756 #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1757 #define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1758 #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1759 #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1760 #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1761 #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
1763 /* For internal core Perl use only: the base macros for defining macros like
1764 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1765 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
1766 * simple class definitions; the other for more complex. These are like
1767 * _generic_uvchr, so see it for more info. */
1768 #define _generic_LC_uvchr(latin1, above_latin1, c) \
1769 (c < 256 ? latin1(c) : above_latin1(c))
1770 #define _generic_LC_swash_uvchr(latin1, classnum, c) \
1771 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1773 #define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1774 #define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1775 _CC_ALPHANUMERIC, c)
1776 #define isASCII_LC_uvchr(c) isASCII_LC(c)
1777 #define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1778 is_HORIZWS_cp_high, c)
1779 #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
1780 #define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1781 #define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
1782 #define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
1783 _is_uni_perl_idcont, c)
1784 #define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
1785 _is_uni_perl_idstart, c)
1786 #define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1787 #define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
1788 #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
1789 #define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
1790 #define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
1791 is_XPERLSPACE_cp_high, c)
1792 #define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
1793 #define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
1795 #define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1796 is_XDIGIT_cp_high, c)
1798 #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
1800 /* For internal core Perl use only: the base macros for defining macros like
1801 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1802 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1803 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1804 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1805 * same representation whether utf8 or not. Note that it assumes that the utf8
1806 * has been validated, and ignores 'use bytes' */
1807 #define _base_generic_utf8(enum_name, name, p, use_locale ) \
1808 _is_utf8_FOO(CAT2(_CC_, enum_name), \
1810 "is" STRINGIFY(name) "_utf8", \
1811 "is" STRINGIFY(name) "_utf8_safe", \
1812 1, use_locale, __FILE__,__LINE__)
1814 #define _generic_utf8(name, p) _base_generic_utf8(name, name, p, 0)
1816 /* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
1817 * they don't otherwise go out of their way to look for malformed UTF-8. If
1818 * they can return accurate results without knowing if the input is otherwise
1819 * malformed, they do so. For example isASCII is accurate in spite of any
1820 * non-length malformations because it looks only at a single byte. Likewise
1821 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
1822 * variant ones return FALSE. But, if the input has to be well-formed in order
1823 * for the results to be accurate, the macros will test and if malformed will
1824 * call a routine to die
1826 * Except for toke.c, the macros do assume that e > p, asserting that on
1827 * DEBUGGING builds. Much code that calls these depends on this being true,
1828 * for other reasons. toke.c is treated specially as using the regular
1829 * assertion breaks it in many ways. All strings that these operate on there
1830 * are supposed to have an extra NUL character at the end, so that *e = \0. A
1831 * bunch of code in toke.c assumes that this is true, so the assertion allows
1833 #ifdef PERL_IN_TOKE_C
1834 # define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
1836 # define _utf8_safe_assert(p,e) ((e) > (p))
1839 #define _generic_utf8_safe(classnum, p, e, above_latin1) \
1840 (__ASSERT_(_utf8_safe_assert(p, e)) \
1841 (UTF8_IS_INVARIANT(*(p))) \
1842 ? _generic_isCC(*(p), classnum) \
1843 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1844 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1845 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
1847 : (_force_out_malformed_utf8_message( \
1848 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1850 /* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1851 * 'above_latin1' can be a macro */
1852 #define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
1853 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
1854 #define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1855 _generic_utf8_safe(classnum, p, e, \
1856 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1857 ? (_force_out_malformed_utf8_message( \
1858 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1860 /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
1861 * 'above_latin1' parameter */
1862 #define _generic_swash_utf8_safe(classnum, p, e) \
1863 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e))
1865 /* Like the above, but should be used only when it is known that there are no
1866 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1867 * class is TRUE for. Hence it can skip the tests for this range.
1868 * 'above_latin1' should include its arguments */
1869 #define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
1870 (__ASSERT_(_utf8_safe_assert(p, e)) \
1871 (UTF8_IS_INVARIANT(*(p))) \
1872 ? _generic_isCC(*(p), classnum) \
1873 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1874 ? 0 /* Note that doesn't check validity for latin1 */ \
1878 #define isALPHA_utf8(p) _generic_utf8(ALPHA, p)
1879 #define isALPHANUMERIC_utf8(p) _generic_utf8(ALPHANUMERIC, p)
1880 #define isASCII_utf8(p) _generic_utf8(ASCII, p)
1881 #define isBLANK_utf8(p) _generic_utf8(BLANK, p)
1882 #define isCNTRL_utf8(p) _generic_utf8(CNTRL, p)
1883 #define isDIGIT_utf8(p) _generic_utf8(DIGIT, p)
1884 #define isGRAPH_utf8(p) _generic_utf8(GRAPH, p)
1885 #define isIDCONT_utf8(p) _generic_utf8(IDCONT, p)
1886 #define isIDFIRST_utf8(p) _generic_utf8(IDFIRST, p)
1887 #define isLOWER_utf8(p) _generic_utf8(LOWER, p)
1888 #define isPRINT_utf8(p) _generic_utf8(PRINT, p)
1889 #define isPSXSPC_utf8(p) _generic_utf8(PSXSPC, p)
1890 #define isPUNCT_utf8(p) _generic_utf8(PUNCT, p)
1891 #define isSPACE_utf8(p) _generic_utf8(SPACE, p)
1892 #define isUPPER_utf8(p) _generic_utf8(UPPER, p)
1893 #define isVERTWS_utf8(p) _generic_utf8(VERTSPACE, p)
1894 #define isWORDCHAR_utf8(p) _generic_utf8(WORDCHAR, p)
1895 #define isXDIGIT_utf8(p) _generic_utf8(XDIGIT, p)
1897 #define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e)
1898 #define isALPHANUMERIC_utf8_safe(p, e) \
1899 _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e)
1900 #define isASCII_utf8_safe(p, e) \
1901 /* Because ASCII is invariant under utf8, the non-utf8 macro \
1903 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
1904 #define isBLANK_utf8_safe(p, e) \
1905 _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
1908 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1909 * more efficient macro instead of the more general one */
1910 # define isCNTRL_utf8_safe(p, e) \
1911 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
1913 # define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
1916 #define isDIGIT_utf8_safe(p, e) \
1917 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
1918 _is_utf8_FOO_with_len(_CC_DIGIT, p, e))
1919 #define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e)
1920 #define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
1921 _is_utf8_perl_idcont_with_len, p, e)
1923 /* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1924 * IDFIRST is an alnum. See
1925 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
1926 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1927 * This used to be not the XID version, but we decided to go with the more
1928 * modern Unicode definition */
1929 #define isIDFIRST_utf8_safe(p, e) \
1930 _generic_func_utf8_safe(_CC_IDFIRST, \
1931 _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e))
1933 #define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e)
1934 #define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e)
1935 #define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
1936 #define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e)
1937 #define isSPACE_utf8_safe(p, e) \
1938 _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
1939 #define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e)
1940 #define isVERTWS_utf8_safe(p, e) \
1941 _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
1942 #define isWORDCHAR_utf8_safe(p, e) \
1943 _generic_swash_utf8_safe(_CC_WORDCHAR, p, e)
1944 #define isXDIGIT_utf8_safe(p, e) \
1945 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
1946 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1947 ? (_force_out_malformed_utf8_message( \
1948 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1949 : is_XDIGIT_high(p)))
1951 #define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
1952 #define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1953 #define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
1954 #define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
1956 /* For internal core use only, subject to change */
1957 #define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f, "", 0)
1958 #define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f, "", 0)
1959 #define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f, "", 0)
1960 #define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f, "", 0)
1962 #define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
1963 #define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
1964 #define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
1965 #define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
1967 /* For internal core Perl use only: the base macros for defining macros like
1968 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
1969 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
1970 * 'macro' parameter */
1971 #define _generic_LC_utf8(name, p) _base_generic_utf8(name, name, p, 1)
1973 #define isALPHA_LC_utf8(p) _generic_LC_utf8(ALPHA, p)
1974 #define isALPHANUMERIC_LC_utf8(p) _generic_LC_utf8(ALPHANUMERIC, p)
1975 #define isASCII_LC_utf8(p) _generic_LC_utf8(ASCII, p)
1976 #define isBLANK_LC_utf8(p) _generic_LC_utf8(BLANK, p)
1977 #define isCNTRL_LC_utf8(p) _generic_LC_utf8(CNTRL, p)
1978 #define isDIGIT_LC_utf8(p) _generic_LC_utf8(DIGIT, p)
1979 #define isGRAPH_LC_utf8(p) _generic_LC_utf8(GRAPH, p)
1980 #define isIDCONT_LC_utf8(p) _generic_LC_utf8(IDCONT, p)
1981 #define isIDFIRST_LC_utf8(p) _generic_LC_utf8(IDFIRST, p)
1982 #define isLOWER_LC_utf8(p) _generic_LC_utf8(LOWER, p)
1983 #define isPRINT_LC_utf8(p) _generic_LC_utf8(PRINT, p)
1984 #define isPSXSPC_LC_utf8(p) _generic_LC_utf8(PSXSPC, p)
1985 #define isPUNCT_LC_utf8(p) _generic_LC_utf8(PUNCT, p)
1986 #define isSPACE_LC_utf8(p) _generic_LC_utf8(SPACE, p)
1987 #define isUPPER_LC_utf8(p) _generic_LC_utf8(UPPER, p)
1988 #define isWORDCHAR_LC_utf8(p) _generic_LC_utf8(WORDCHAR, p)
1989 #define isXDIGIT_LC_utf8(p) _generic_LC_utf8(XDIGIT, p)
1991 /* For internal core Perl use only: the base macros for defining macros like
1992 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
1993 * point in 'p' is within the 0-255 range, it uses locale rules from the
1994 * passed-in 'macro' parameter */
1995 #define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
1996 (__ASSERT_(_utf8_safe_assert(p, e)) \
1997 (UTF8_IS_INVARIANT(*(p))) \
1999 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2000 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2001 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2002 : (_force_out_malformed_utf8_message( \
2003 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2006 #define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \
2007 _generic_LC_utf8_safe(macro, p, e, \
2008 _is_utf8_FOO_with_len(classnum, p, e))
2010 #define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2011 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2013 #define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \
2014 _generic_LC_utf8_safe(classnum, p, e, \
2015 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2016 ? (_force_out_malformed_utf8_message( \
2017 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2020 #define isALPHANUMERIC_LC_utf8_safe(p, e) \
2021 _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \
2022 _CC_ALPHANUMERIC, p, e)
2023 #define isALPHA_LC_utf8_safe(p, e) \
2024 _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2025 #define isASCII_LC_utf8_safe(p, e) \
2026 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2027 #define isBLANK_LC_utf8_safe(p, e) \
2028 _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2029 #define isCNTRL_LC_utf8_safe(p, e) \
2030 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2031 #define isDIGIT_LC_utf8_safe(p, e) \
2032 _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2033 #define isGRAPH_LC_utf8_safe(p, e) \
2034 _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2035 #define isIDCONT_LC_utf8_safe(p, e) \
2036 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2037 _is_utf8_perl_idcont_with_len, p, e)
2038 #define isIDFIRST_LC_utf8_safe(p, e) \
2039 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2040 _is_utf8_perl_idstart_with_len, p, e)
2041 #define isLOWER_LC_utf8_safe(p, e) \
2042 _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2043 #define isPRINT_LC_utf8_safe(p, e) \
2044 _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2045 #define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2046 #define isPUNCT_LC_utf8_safe(p, e) \
2047 _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2048 #define isSPACE_LC_utf8_safe(p, e) \
2049 _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2050 #define isUPPER_LC_utf8_safe(p, e) \
2051 _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2052 #define isWORDCHAR_LC_utf8_safe(p, e) \
2053 _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2054 #define isXDIGIT_LC_utf8_safe(p, e) \
2055 _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
2057 /* Macros for backwards compatibility and for completeness when the ASCII and
2058 * Latin1 values are identical */
2059 #define isALPHAU(c) isALPHA_L1(c)
2060 #define isDIGIT_L1(c) isDIGIT_A(c)
2061 #define isOCTAL(c) isOCTAL_A(c)
2062 #define isOCTAL_L1(c) isOCTAL_A(c)
2063 #define isXDIGIT_L1(c) isXDIGIT_A(c)
2064 #define isALNUM(c) isWORDCHAR(c)
2065 #define isALNUMU(c) isWORDCHAR_L1(c)
2066 #define isALNUM_LC(c) isWORDCHAR_LC(c)
2067 #define isALNUM_uni(c) isWORDCHAR_uni(c)
2068 #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
2069 #define isALNUM_utf8(p) isWORDCHAR_utf8(p)
2070 #define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
2071 #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2072 #define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2073 #define isALNUMC(c) isALPHANUMERIC(c)
2074 #define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2075 #define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
2076 #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
2077 #define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
2078 #define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
2080 /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2081 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2082 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2083 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2084 * the outlier from the block that contains the other controls, just like
2085 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2086 * block. If it weren't special cased, it would yield a non-control.
2087 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2090 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2092 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2094 ? (UNLIKELY((c) == '?') \
2095 ? QUESTION_MARK_CTRL \
2096 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2097 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2099 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2102 /* Line numbers are unsigned, 32 bits. */
2104 #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
2106 /* Helpful alias for version prescan */
2107 #define is_LAX_VERSION(a,b) \
2108 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2110 #define is_STRICT_VERSION(a,b) \
2111 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2113 #define BADVERSION(a,b,c) \
2119 /* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2120 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2121 * which is advanced. The input is validated only by an assert() in DEBUGGING
2122 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2123 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2124 #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2127 #define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2131 /* Converts a character known to represent an octal digit (0-7) to its numeric
2132 * value. The input is validated only by an assert() in DEBUGGING builds. In
2133 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2134 #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2136 /* Efficiently returns a boolean as to if two native characters are equivalent
2137 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2138 * the ALPHA in the name is to remind you of that. This is asserted() in
2139 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2140 * works (on valid input) for both non- and UTF-8-encoded bytes.
2142 * When one of the inputs is a compile-time constant and gets folded by the
2143 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2144 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2145 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
2146 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2147 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2148 * just a single 0, in the bit position where the upper- and lowercase differ.
2150 #define isALPHA_FOLD_EQ(c1, c2) \
2151 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2152 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2153 #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2156 =head1 Memory Management
2158 =for apidoc Am|void|Newx|void* ptr|int nitems|type
2159 The XSUB-writer's interface to the C C<malloc> function.
2161 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2163 In 5.9.3, Newx() and friends replace the older New() API, and drops
2164 the first parameter, I<x>, a debug aid which allowed callers to identify
2165 themselves. This aid has been superseded by a new build option,
2166 PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
2167 there for use in XS modules supporting older perls.
2169 =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
2170 The XSUB-writer's interface to the C C<malloc> function, with
2171 cast. See also C<L</Newx>>.
2173 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2175 =for apidoc Am|void|Newxz|void* ptr|int nitems|type
2176 The XSUB-writer's interface to the C C<malloc> function. The allocated
2177 memory is zeroed with C<memzero>. See also C<L</Newx>>.
2179 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2181 =for apidoc Am|void|Renew|void* ptr|int nitems|type
2182 The XSUB-writer's interface to the C C<realloc> function.
2184 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2186 =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2187 The XSUB-writer's interface to the C C<realloc> function, with
2190 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2192 =for apidoc Am|void|Safefree|void* ptr
2193 The XSUB-writer's interface to the C C<free> function.
2195 This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
2197 =for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2198 The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
2199 source, C<dest> is the destination, C<nitems> is the number of items, and
2200 C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
2202 =for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
2203 Like C<Move> but returns C<dest>. Useful
2204 for encouraging compilers to tail-call
2207 =for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2208 The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
2209 source, C<dest> is the destination, C<nitems> is the number of items, and
2210 C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
2212 =for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2214 Like C<Copy> but returns C<dest>. Useful
2215 for encouraging compilers to tail-call
2218 =for apidoc Am|void|Zero|void* dest|int nitems|type
2220 The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2221 destination, C<nitems> is the number of items, and C<type> is the type.
2223 =for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2225 Like C<Zero> but returns dest. Useful
2226 for encouraging compilers to tail-call
2229 =for apidoc Am|void|StructCopy|type *src|type *dest|type
2230 This is an architecture-independent macro to copy one structure to another.
2232 =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2234 Fill up memory with a byte pattern (a byte repeated over and over
2235 again) that hopefully catches attempts to access uninitialized memory.
2237 =for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2239 PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2241 =for apidoc Am|void|PoisonFree|void* dest|int nitems|type
2243 PoisonWith(0xEF) for catching access to freed memory.
2245 =for apidoc Am|void|Poison|void* dest|int nitems|type
2247 PoisonWith(0xEF) for catching access to freed memory.
2251 /* Maintained for backwards-compatibility only. Use newSV() instead. */
2253 #define NEWSV(x,len) newSV(len)
2256 #define MEM_SIZE_MAX ((MEM_SIZE)-1)
2258 #define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
2260 #ifdef PERL_MALLOC_WRAP
2262 /* This expression will be constant-folded at compile time. It checks
2263 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2264 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2265 * As well as avoiding the need for a run-time check in some cases, it's
2266 * designed to avoid compiler warnings like:
2267 * comparison is always false due to limited range of data type
2268 * It's mathematically equivalent to
2269 * max(n) * sizeof(t) > MEM_SIZE_MAX
2272 # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
2273 ( sizeof(MEM_SIZE) < sizeof(n) \
2274 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
2276 /* This is written in a slightly odd way to avoid various spurious
2277 * compiler warnings. We *want* to write the expression as
2278 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2279 * (for some compile-time constant C), but even when the LHS
2280 * constant-folds to false at compile-time, g++ insists on emitting
2281 * warnings about the RHS (e.g. "comparison is always false"), so instead
2284 * (cond ? n : X) > C
2286 * where X is a constant with X > C always false. Choosing a value for X
2287 * is tricky. If 0, some compilers will complain about 0 > C always being
2288 * false; if 1, Coverity complains when n happens to be the constant value
2289 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2290 * for X and hope that nothing else whines.
2293 # define _MEM_WRAP_WILL_WRAP(n,t) \
2294 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2295 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
2297 # define MEM_WRAP_CHECK(n,t) \
2298 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2299 && (croak_memory_wrap(),0))
2301 # define MEM_WRAP_CHECK_1(n,t,a) \
2302 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2303 && (Perl_croak_nocontext("%s",(a)),0))
2305 /* "a" arg must be a string literal */
2306 # define MEM_WRAP_CHECK_s(n,t,a) \
2307 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2308 && (Perl_croak_nocontext("" a ""),0))
2310 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
2312 #define PERL_STRLEN_ROUNDUP(n) ((void)(((n) > MEM_SIZE_MAX - 2 * PERL_STRLEN_ROUNDUP_QUANTUM) ? (croak_memory_wrap(),0) : 0), _PERL_STRLEN_ROUNDUP_UNCHECKED(n))
2315 #define MEM_WRAP_CHECK(n,t)
2316 #define MEM_WRAP_CHECK_1(n,t,a)
2317 #define MEM_WRAP_CHECK_s(n,t,a)
2318 #define MEM_WRAP_CHECK_(n,t)
2320 #define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
2326 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
2327 * go through functions, which are handy for debugging breakpoints, but
2328 * which more importantly get the immediate calling environment (file and
2329 * line number, and C function name if available) passed in. This info can
2330 * then be used for logging the calls, for which one gets a sample
2331 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
2334 * - not all memory allocs get logged, only those
2335 * that go through Newx() and derivatives (while all
2336 * Safefrees do get logged)
2337 * - __FILE__ and __LINE__ do not work everywhere
2338 * - __func__ or __FUNCTION__ even less so
2339 * - I think more goes on after the perlio frees but
2340 * the thing is that STDERR gets closed (as do all
2341 * the file descriptors)
2342 * - no deeper calling stack than the caller of the Newx()
2343 * or the kind, but do I look like a C reflection/introspection
2345 * - the function prototypes for the logging functions
2346 * probably should maybe be somewhere else than handy.h
2347 * - one could consider inlining (macrofying) the logging
2348 * for speed, but I am too lazy
2349 * - one could imagine recording the allocations in a hash,
2350 * (keyed by the allocation address?), and maintain that
2351 * through reallocs and frees, but how to do that without
2352 * any News() happening...?
2353 * - lots of -Ddefines to get useful/controllable output
2354 * - lots of ENV reads
2358 # ifndef PERL_MEM_LOG_NOIMPL
2367 # if defined(PERL_IN_SV_C) /* those are only used in sv.c */
2368 void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2369 void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2376 #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2377 #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
2378 #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
2381 #ifndef MEM_LOG_ALLOC
2382 #define MEM_LOG_ALLOC(n,t,a) (a)
2384 #ifndef MEM_LOG_REALLOC
2385 #define MEM_LOG_REALLOC(n,t,v,a) (a)
2387 #ifndef MEM_LOG_FREE
2388 #define MEM_LOG_FREE(a) (a)
2391 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2392 #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2393 #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
2396 /* pre 5.9.x compatibility */
2397 #define New(x,v,n,t) Newx(v,n,t)
2398 #define Newc(x,v,n,t,c) Newxc(v,n,t,c)
2399 #define Newz(x,v,n,t) Newxz(v,n,t)
2402 #define Renew(v,n,t) \
2403 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2404 #define Renewc(v,n,t,c) \
2405 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2408 #define Safefree(d) \
2409 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
2411 #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
2414 /* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2415 * as it handles cases like constant string arguments without throwing warnings *
2416 * the cast is required, as is the inequality check, to avoid warnings */
2417 #define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
2420 #define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2421 #define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2422 #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2424 /* Like above, but returns a pointer to 'd' */
2425 #define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), memmove((char*)(d),(const char*)(s), (n) * sizeof(t)))
2426 #define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), memcpy((char*)(d),(const char*)(s), (n) * sizeof(t)))
2427 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
2429 #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2430 #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2431 #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2432 #define Poison(d,n,t) PoisonFree(d,n,t)
2435 # define PERL_POISON_EXPR(x) x
2437 # define PERL_POISON_EXPR(x)
2440 #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2442 /* C_ARRAY_LENGTH is the number of elements in the C array (so you
2443 * want your zero-based indices to be less than but not equal to).
2445 * C_ARRAY_END is one past the last: half-open/half-closed range,
2446 * not last-inclusive range. */
2447 #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
2448 #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
2452 # define Perl_va_copy(s, d) va_copy(d, s)
2453 # elif defined(__va_copy)
2454 # define Perl_va_copy(s, d) __va_copy(d, s)
2456 # define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2460 /* convenience debug macros */
2462 #define pTHX_FORMAT "Perl interpreter: 0x%p"
2463 #define pTHX__FORMAT ", Perl interpreter: 0x%p"
2464 #define pTHX_VALUE_ (void *)my_perl,
2465 #define pTHX_VALUE (void *)my_perl
2466 #define pTHX__VALUE_ ,(void *)my_perl,
2467 #define pTHX__VALUE ,(void *)my_perl
2470 #define pTHX__FORMAT
2473 #define pTHX__VALUE_
2475 #endif /* USE_ITHREADS */
2477 /* Perl_deprecate was not part of the public API, and did not have a deprecate()
2478 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2479 CPAN::Unpack show any users outside the core. */
2481 # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2482 "Use of " s " is deprecated")
2483 # define deprecate_disappears_in(when,message) \
2484 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2485 message ", and will disappear in Perl " when)
2486 # define deprecate_fatal_in(when,message) \
2487 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2488 message ". Its use will be fatal in Perl " when)
2491 /* Internal macros to deal with gids and uids */
2494 # if Uid_t_size > IVSIZE
2495 # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2496 # define SvUID(sv) SvNV(sv)
2497 # elif Uid_t_sign <= 0
2498 # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2499 # define SvUID(sv) SvIV(sv)
2501 # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2502 # define SvUID(sv) SvUV(sv)
2503 # endif /* Uid_t_size */
2505 # if Gid_t_size > IVSIZE
2506 # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2507 # define SvGID(sv) SvNV(sv)
2508 # elif Gid_t_sign <= 0
2509 # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2510 # define SvGID(sv) SvIV(sv)
2512 # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2513 # define SvGID(sv) SvUV(sv)
2514 # endif /* Gid_t_size */
2518 #endif /* PERL_HANDY_H_ */
2521 * ex: set ts=8 sts=4 sw=4 et:
https://perl5.git.perl.org / perl5.git / blob