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 AmnU||Nullch
24 Null character pointer. (No longer available when C<PERL_CORE> is
27 =for apidoc AmnU||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
101 =for apidoc Am|bool|cBOOL|bool expr
103 Cast-to-bool. A simple S<C<(bool) I<expr>>> cast may not do the right thing:
104 if C<bool> is defined as C<char>, for example, then the cast from C<int> is
105 implementation-defined.
107 C<(bool)!!(cbool)> in a ternary triggers a bug in xlc on AIX
111 #define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
113 /* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
114 * XXX Should really be a Configure probe, with HAS__FUNCTION__
115 * and FUNCTION__ as results.
116 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
117 #if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
118 # define FUNCTION__ __func__
119 #elif (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
120 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
121 # define FUNCTION__ ""
123 # define FUNCTION__ __FUNCTION__ /* Common extension. */
126 /* XXX A note on the perl source internal type system. The
127 original intent was that I32 be *exactly* 32 bits.
129 Currently, we only guarantee that I32 is *at least* 32 bits.
130 Specifically, if int is 64 bits, then so is I32. (This is the case
131 for the Cray.) This has the advantage of meshing nicely with
132 standard library calls (where we pass an I32 and the library is
133 expecting an int), but the disadvantage that an I32 is not 32 bits.
134 Andy Dougherty August 1996
136 There is no guarantee that there is *any* integral type with
137 exactly 32 bits. It is perfectly legal for a system to have
138 sizeof(short) == sizeof(int) == sizeof(long) == 8.
140 Similarly, there is no guarantee that I16 and U16 have exactly 16
143 For dealing with issues that may arise from various 32/64-bit
144 systems, we will ask Configure to check out
146 SHORTSIZE == sizeof(short)
147 INTSIZE == sizeof(int)
148 LONGSIZE == sizeof(long)
149 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
150 PTRSIZE == sizeof(void *)
151 DOUBLESIZE == sizeof(double)
152 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
156 #ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
157 # include <inttypes.h>
158 # ifdef INT32_MIN_BROKEN
160 # define INT32_MIN (-2147483647-1)
162 # ifdef INT64_MIN_BROKEN
164 # define INT64_MIN (-9223372036854775807LL-1)
180 #if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
182 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
183 Please search CHAR_MAX in perl.h for further details. */
184 #define U8_MAX UINT8_MAX
185 #define U8_MIN UINT8_MIN
187 #define I16_MAX INT16_MAX
188 #define I16_MIN INT16_MIN
189 #define U16_MAX UINT16_MAX
190 #define U16_MIN UINT16_MIN
192 #define I32_MAX INT32_MAX
193 #define I32_MIN INT32_MIN
194 #ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
195 # define U32_MAX UINT32_MAX
197 # define U32_MAX 4294967295U
199 #define U32_MIN UINT32_MIN
203 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
204 Please search CHAR_MAX in perl.h for further details. */
205 #define U8_MAX PERL_UCHAR_MAX
206 #define U8_MIN PERL_UCHAR_MIN
208 #define I16_MAX PERL_SHORT_MAX
209 #define I16_MIN PERL_SHORT_MIN
210 #define U16_MAX PERL_USHORT_MAX
211 #define U16_MIN PERL_USHORT_MIN
214 # define I32_MAX PERL_INT_MAX
215 # define I32_MIN PERL_INT_MIN
216 # define U32_MAX PERL_UINT_MAX
217 # define U32_MIN PERL_UINT_MIN
219 # define I32_MAX PERL_LONG_MAX
220 # define I32_MIN PERL_LONG_MIN
221 # define U32_MAX PERL_ULONG_MAX
222 # define U32_MIN PERL_ULONG_MIN
227 /* These C99 typedefs are useful sometimes for, say, loop variables whose
228 * maximum values are small, but for which speed trumps size. If we have a C99
229 * compiler, use that. Otherwise, a plain 'int' should be good enough.
231 * Restrict these to core for now until we are more certain this is a good
233 #if defined(PERL_CORE) || defined(PERL_EXT)
235 typedef int_fast8_t PERL_INT_FAST8_T;
236 typedef uint_fast8_t PERL_UINT_FAST8_T;
237 typedef int_fast16_t PERL_INT_FAST16_T;
238 typedef uint_fast16_t PERL_UINT_FAST16_T;
240 typedef int PERL_INT_FAST8_T;
241 typedef unsigned int PERL_UINT_FAST8_T;
242 typedef int PERL_INT_FAST16_T;
243 typedef unsigned int PERL_UINT_FAST16_T;
247 /* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
248 * anyone is grepping for it */
249 #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
250 #define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
251 #define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
253 /* Unused by core; should be deprecated */
254 #define Ctl(ch) ((ch) & 037)
256 #if defined(PERL_CORE) || defined(PERL_EXT)
258 # define MIN(a,b) ((a) < (b) ? (a) : (b))
261 # define MAX(a,b) ((a) > (b) ? (a) : (b))
265 /* Returns a boolean as to whether the input unsigned number is a power of 2
266 * (2**0, 2**1, etc). In other words if it has just a single bit set.
267 * If not, subtracting 1 would leave the uppermost bit set, so the & would
269 #if defined(PERL_CORE) || defined(PERL_EXT)
270 # define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
274 =for apidoc Am|void|__ASSERT_|bool expr
276 This is a helper macro to avoid preprocessor issues, replaced by nothing
277 unless under DEBUGGING, where it expands to an assert of its argument,
278 followed by a comma (hence the comma operator). If we just used a straight
279 assert(), we would get a comma with nothing before it when not DEBUGGING.
283 We also use empty definition under Coverity since the __ASSERT__
284 checks often check for things that Really Cannot Happen, and Coverity
285 detects that and gets all excited. */
287 #if defined(DEBUGGING) && !defined(__COVERITY__)
288 # define __ASSERT_(statement) assert(statement),
290 # define __ASSERT_(statement)
294 =head1 SV Manipulation Functions
296 =for apidoc Ama|SV*|newSVpvs|"literal string"
297 Like C<newSVpvn>, but takes a literal string instead of a
300 =for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
301 Like C<newSVpvn_flags>, but takes a literal string instead of
302 a string/length pair.
304 =for apidoc Ama|SV*|newSVpvs_share|"literal string"
305 Like C<newSVpvn_share>, but takes a literal string instead of
306 a string/length pair and omits the hash parameter.
308 =for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
309 Like C<sv_catpvn_flags>, but takes a literal string instead
310 of a string/length pair.
312 =for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
313 Like C<sv_catpvn_nomg>, but takes a literal string instead of
314 a string/length pair.
316 =for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
317 Like C<sv_catpvn>, but takes a literal string instead of a
320 =for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
321 Like C<sv_catpvn_mg>, but takes a literal string instead of a
324 =for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
325 Like C<sv_setpvn>, but takes a literal string instead of a
328 =for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
329 Like C<sv_setpvn_mg>, but takes a literal string instead of a
332 =for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
333 Like C<sv_setref_pvn>, but takes a literal string instead of
334 a string/length pair.
336 =head1 Memory Management
338 =for apidoc Ama|char*|savepvs|"literal string"
339 Like C<savepvn>, but takes a literal string instead of a
342 =for apidoc Ama|char*|savesharedpvs|"literal string"
343 A version of C<savepvs()> which allocates the duplicate string in memory
344 which is shared between threads.
348 =for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
349 Like C<gv_stashpvn>, but takes a literal string instead of a
352 =head1 Hash Manipulation Functions
354 =for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
355 Like C<hv_fetch>, but takes a literal string instead of a
358 =for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
359 Like C<hv_store>, but takes a literal string instead of a
361 and omits the hash parameter.
363 =head1 Lexer interface
365 =for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
367 Like L</lex_stuff_pvn>, but takes a literal string instead of
368 a string/length pair.
376 =for apidoc Amu|pair|STR_WITH_LEN|"literal string"
378 Returns two comma separated tokens of the input literal string, and its length.
379 This is convenience macro which helps out in some API calls.
380 Note that it can't be used as an argument to macros or functions that under
381 some configurations might be macros, which means that it requires the full
382 Perl_xxx(aTHX_ ...) form for any API calls where it's used.
388 #define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
390 /* STR_WITH_LEN() shortcuts */
391 #define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
392 #define newSVpvs_flags(str,flags) \
393 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
394 #define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
395 #define sv_catpvs_flags(sv, str, flags) \
396 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
397 #define sv_catpvs_nomg(sv, str) \
398 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
399 #define sv_catpvs(sv, str) \
400 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
401 #define sv_catpvs_mg(sv, str) \
402 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
403 #define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
404 #define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
405 #define sv_setref_pvs(rv, classname, str) \
406 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
407 #define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
408 #define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
409 #define gv_stashpvs(str, create) \
410 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
411 #define gv_fetchpvs(namebeg, add, sv_type) \
412 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
413 #define gv_fetchpvn(namebeg, len, add, sv_type) \
414 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
415 #define sv_catxmlpvs(dsv, str, utf8) \
416 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
419 #define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
421 #define get_cvs(str, flags) \
422 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
425 =head1 Miscellaneous Functions
427 =for apidoc Am|bool|strNE|char* s1|char* s2
428 Test two C<NUL>-terminated strings to see if they are different. Returns true
431 =for apidoc Am|bool|strEQ|char* s1|char* s2
432 Test two C<NUL>-terminated strings to see if they are equal. Returns true or
435 =for apidoc Am|bool|strLT|char* s1|char* s2
436 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
437 second, C<s2>. Returns true or false.
439 =for apidoc Am|bool|strLE|char* s1|char* s2
440 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
441 equal to the second, C<s2>. Returns true or false.
443 =for apidoc Am|bool|strGT|char* s1|char* s2
444 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
445 the second, C<s2>. Returns true or false.
447 =for apidoc Am|bool|strGE|char* s1|char* s2
448 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
449 or equal to the second, C<s2>. Returns true or false.
451 =for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
452 Test two C<NUL>-terminated strings to see if they are different. The C<len>
453 parameter indicates the number of bytes to compare. Returns true or false. (A
454 wrapper for C<strncmp>).
456 =for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
457 Test two C<NUL>-terminated strings to see if they are equal. The C<len>
458 parameter indicates the number of bytes to compare. Returns true or false. (A
459 wrapper for C<strncmp>).
461 =for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
462 Test two buffers (which may contain embedded C<NUL> characters, to see if they
463 are equal. The C<len> parameter indicates the number of bytes to compare.
464 Returns zero if equal, or non-zero if non-equal.
466 =for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
467 Like L</memEQ>, but the second string is a literal enclosed in double quotes,
468 C<l1> gives the number of bytes in C<s1>.
469 Returns zero if equal, or non-zero if non-equal.
471 =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
472 Test two buffers (which may contain embedded C<NUL> characters, to see if they
473 are not equal. The C<len> parameter indicates the number of bytes to compare.
474 Returns zero if non-equal, or non-zero if equal.
476 =for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
477 Like L</memNE>, but the second string is a literal enclosed in double quotes,
478 C<l1> gives the number of bytes in C<s1>.
479 Returns zero if non-equal, or zero if non-equal.
483 New macros should use the following conventions for their names (which are
484 based on the underlying C library functions):
486 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
488 Each has two main parameters, string-like operands that are compared
489 against each other, as specified by the macro name. Some macros may
490 additionally have one or potentially even two length parameters. If a length
491 parameter applies to both string parameters, it will be positioned third;
492 otherwise any length parameter immediately follows the string parameter it
495 If the prefix to the name is 'str', the string parameter is a pointer to a C
496 language string. Such a string does not contain embedded NUL bytes; its
497 length may be unknown, but can be calculated by C<strlen()>, since it is
498 terminated by a NUL, which isn't included in its length.
500 The optional 'n' following 'str' means that that there is a third parameter,
501 giving the maximum number of bytes to look at in each string. Even if both
502 strings are longer than the length parameter, those extra bytes will be
505 The 's' suffix means that the 2nd byte string parameter is a literal C
506 double-quoted string. Its length will automatically be calculated by the
507 macro, so no length parameter will ever be needed for it.
509 If the prefix is 'mem', the string parameters don't have to be C strings;
510 they may contain embedded NUL bytes, do not necessarily have a terminating
511 NUL, and their lengths can be known only through other means, which in
512 practice are additional parameter(s) passed to the function. All 'mem'
513 functions have at least one length parameter. Barring any 'l' or 's' suffix,
514 there is a single length parameter, in position 3, which applies to both
515 string parameters. The 's' suffix means, as described above, that the 2nd
516 string is a literal double-quoted C string (hence its length is calculated by
517 the macro, and the length parameter to the function applies just to the first
518 string parameter, and hence is positioned just after it). An 'l' suffix
519 means that the 2nd string parameter has its own length parameter, and the
520 signature will look like memFOOl(s1, l1, s2, l2).
522 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
523 substring of the 1st string. 'P' if present indicates that the substring
524 must be a "proper" one in tha mathematical sense that the first one must be
525 strictly larger than the 2nd.
530 #define strNE(s1,s2) (strcmp(s1,s2) != 0)
531 #define strEQ(s1,s2) (strcmp(s1,s2) == 0)
532 #define strLT(s1,s2) (strcmp(s1,s2) < 0)
533 #define strLE(s1,s2) (strcmp(s1,s2) <= 0)
534 #define strGT(s1,s2) (strcmp(s1,s2) > 0)
535 #define strGE(s1,s2) (strcmp(s1,s2) >= 0)
537 #define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
538 #define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
540 #define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
541 #define memNE(s1,s2,l) (! memEQ(s1,s2,l))
543 /* memEQ and memNE where second comparand is a string constant */
544 #define memEQs(s1, l, s2) \
545 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
546 #define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
548 /* Keep these private until we decide it was a good idea */
549 #if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
551 #define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
553 #define memBEGINs(s1, l, s2) \
554 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
555 && memEQ(s1, "" s2 "", sizeof(s2)-1))
556 #define memBEGINPs(s1, l, s2) \
557 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
558 && memEQ(s1, "" s2 "", sizeof(s2)-1))
559 #define memENDs(s1, l, s2) \
560 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
561 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
562 #define memENDPs(s1, l, s2) \
563 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
564 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
565 #endif /* End of making macros private */
567 #define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
568 #define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
569 #define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
570 #define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
575 * Unfortunately, the introduction of locales means that we
576 * can't trust isupper(), etc. to tell the truth. And when
577 * it comes to /\w+/ with tainting enabled, we *must* be able
578 * to trust our character classes.
580 * Therefore, the default tests in the text of Perl will be
581 * independent of locale. Any code that wants to depend on
582 * the current locale will use the tests that begin with "lc".
585 #ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
593 =head1 Character classification
594 This section is about functions (really macros) that classify characters
595 into types, such as punctuation versus alphabetic, etc. Most of these are
596 analogous to regular expression character classes. (See
597 L<perlrecharclass/POSIX Character Classes>.) There are several variants for
598 each class. (Not all macros have all variants; each item below lists the
599 ones valid for it.) None are affected by C<use bytes>, and only the ones
600 with C<LC> in the name are affected by the current locale.
602 The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
603 treating it as a code point, and returns a boolean as to whether or not the
604 character represented by it is (or on non-ASCII platforms, corresponds to) an
605 ASCII character in the named class based on platform, Unicode, and Perl rules.
606 If the input is a number that doesn't fit in an octet, FALSE is returned.
608 Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
609 with no suffix C<"_A">. This variant is used to emphasize by its name that
610 only ASCII-range characters can return TRUE.
612 Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
613 onto the platform. That is, the code points that are ASCII are unaffected,
614 since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
615 as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
616 true when called with the code point 0xDF, which is a word character in both
617 ASCII and EBCDIC (though it represents different characters in each).
618 If the input is a number that doesn't fit in an octet, FALSE is returned.
619 (Perl's documentation uses a colloquial definition of Latin-1, to include all
620 code points below 256.)
622 Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
623 inputs below 256, but if the code point is larger than 255, Unicode rules are
624 used to determine if it is in the character class. For example,
625 C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
626 WITH MACRON in Unicode, and is a word character.
628 Variant C<isI<FOO>_utf8_safe> is like C<isI<FOO>_uvchr>, but is used for UTF-8
629 encoded strings. Each call classifies the first character of the string. This
630 variant takes two parameters. The first, C<p>, is a
631 pointer to the first byte of the character to be classified. (Recall that it
632 may take more than one byte to represent a character in UTF-8 strings.) The
633 second parameter, C<e>, points to anywhere in the string beyond the first
634 character, up to one byte past the end of the entire string. The suffix
635 C<_safe> in the function's name indicates that it will not attempt to read
636 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
637 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
638 character is malformed in some way, the program may croak, or the function may
639 return FALSE, at the discretion of the implementation, and subject to change in
642 Variant C<isI<FOO>_utf8> is like C<isI<FOO>_utf8_safe>, but takes just a single
643 parameter, C<p>, which has the same meaning as the corresponding parameter does
644 in C<isI<FOO>_utf8_safe>. The function therefore can't check if it is reading
645 beyond the end of the string. Starting in Perl v5.32, it will take a second
646 parameter, becoming a synonym for C<isI<FOO>_utf8_safe>. At that time every
647 program that uses it will have to be changed to successfully compile. In the
648 meantime, the first runtime call to C<isI<FOO>_utf8> from each call point in the
649 program will raise a deprecation warning, enabled by default. You can convert
650 your program now to use C<isI<FOO>_utf8_safe>, and avoid the warnings, and get an
651 extra measure of protection, or you can wait until v5.32, when you'll be forced
652 to add the C<e> parameter.
654 Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
655 but the result is based on the current locale, which is what C<LC> in the name
656 stands for. If Perl can determine that the current locale is a UTF-8 locale,
657 it uses the published Unicode rules; otherwise, it uses the C library function
658 that gives the named classification. For example, C<isDIGIT_LC()> when not in
659 a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
660 returned if the input won't fit into an octet. On some platforms where the C
661 library function is known to be defective, Perl changes its result to follow
662 the POSIX standard's rules.
664 Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
665 than 256, but for larger ones it returns the Unicode classification of the code
668 Variant C<isI<FOO>_LC_utf8_safe> is like C<isI<FOO>_LC_uvchr>, but is used for UTF-8
669 encoded strings. Each call classifies the first character of the string. This
670 variant takes two parameters. The first, C<p>, is a pointer to the first byte
671 of the character to be classified. (Recall that it may take more than one byte
672 to represent a character in UTF-8 strings.) The second parameter, C<e>,
673 points to anywhere in the string beyond the first character, up to one byte
674 past the end of the entire string. The suffix C<_safe> in the function's name
675 indicates that it will not attempt to read beyond S<C<e - 1>>, provided that
676 the constraint S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING>
677 builds). If the UTF-8 for the input character is malformed in some way, the
678 program may croak, or the function may return FALSE, at the discretion of the
679 implementation, and subject to change in future releases.
681 Variant C<isI<FOO>_LC_utf8> is like C<isI<FOO>_LC_utf8_safe>, but takes just a single
682 parameter, C<p>, which has the same meaning as the corresponding parameter does
683 in C<isI<FOO>_LC_utf8_safe>. The function therefore can't check if it is reading
684 beyond the end of the string. Starting in Perl v5.32, it will take a second
685 parameter, becoming a synonym for C<isI<FOO>_LC_utf8_safe>. At that time every
686 program that uses it will have to be changed to successfully compile. In the
687 meantime, the first runtime call to C<isI<FOO>_LC_utf8> from each call point in
688 the program will raise a deprecation warning, enabled by default. You can
689 convert your program now to use C<isI<FOO>_LC_utf8_safe>, and avoid the warnings,
690 and get an extra measure of protection, or you can wait until v5.32, when
691 you'll be forced to add the C<e> parameter.
693 =for apidoc Am|bool|isALPHA|int ch
694 Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
695 analogous to C<m/[[:alpha:]]/>.
696 See the L<top of this section|/Character classification> for an explanation of
698 C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8_safe>,
699 C<isALPHA_LC>, C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8_safe>.
703 Here and below, we add the protoypes of these macros for downstream programs
704 that would be interested in them, such as Devel::PPPort
706 =for apidoc Amh|bool|isALPHA_A|int ch
707 =for apidoc Amh|bool|isALPHA_L1|int ch
708 =for apidoc Amh|bool|isALPHA_uvchr|int ch
709 =for apidoc Amh|bool|isALPHA_utf8_safe|U8 * s|U8 * end
710 =for apidoc Amh|bool|isALPHA_utf8|U8 * s
711 =for apidoc Amh|bool|isALPHA_LC|int ch
712 =for apidoc Amh|bool|isALPHA_LC_uvchr|int ch
713 =for apidoc Amh|bool|isALPHA_LC_utf8_safe|U8 * s| U8 *end
715 =for apidoc Am|bool|isALPHANUMERIC|int ch
716 Returns a boolean indicating whether the specified character is one of
717 C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
718 See the L<top of this section|/Character classification> for an explanation of
720 C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
721 C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>,
722 and C<isALPHANUMERIC_LC_utf8_safe>.
724 A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
725 this corresponds to the C language alphanumeric definition). Also
726 there are the variants
727 C<isALNUMC_A>, C<isALNUMC_L1>
728 C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
730 =for apidoc Amh|bool|isALPHANUMERIC_A|int ch
731 =for apidoc Amh|bool|isALPHANUMERIC_L1|int ch
732 =for apidoc Amh|bool|isALPHANUMERIC_uvchr|int ch
733 =for apidoc Amh|bool|isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
734 =for apidoc Amh|bool|isALPHANUMERIC_utf8|U8 * s
735 =for apidoc Amh|bool|isALPHANUMERIC_LC|int ch
736 =for apidoc Amh|bool|isALPHANUMERIC_LC_uvchr|int ch
737 =for apidoc Amh|bool|isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
738 =for apidoc Amh|bool|isALNUMC|int ch
739 =for apidoc Amh|bool|isALNUMC_A|int ch
740 =for apidoc Amh|bool|isALNUMC_L1|int ch
741 =for apidoc Amh|bool|isALNUMC_LC|int ch
742 =for apidoc Amh|bool|isALNUMC_LC_uvchr|int ch
744 =for apidoc Am|bool|isASCII|int ch
745 Returns a boolean indicating whether the specified character is one of the 128
746 characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
747 On non-ASCII platforms, it returns TRUE iff this
748 character corresponds to an ASCII character. Variants C<isASCII_A()> and
749 C<isASCII_L1()> are identical to C<isASCII()>.
750 See the L<top of this section|/Character classification> for an explanation of
752 C<isASCII_uvchr>, C<isASCII_utf8_safe>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
753 C<isASCII_LC_utf8_safe>. Note, however, that some platforms do not have the C
754 library routine C<isascii()>. In these cases, the variants whose names contain
755 C<LC> are the same as the corresponding ones without.
757 =for apidoc Amh|bool|isASCII_A|int ch
758 =for apidoc Amh|bool|isASCII_L1|int ch
759 =for apidoc Amh|bool|isASCII_uvchr|int ch
760 =for apidoc Amh|bool|isASCII_utf8_safe|U8 * s|U8 * end
761 =for apidoc Amh|bool|isASCII_utf8|U8 * s
762 =for apidoc Amh|bool|isASCII_LC|int ch
763 =for apidoc Amh|bool|isASCII_LC_uvchr|int ch
764 =for apidoc Amh|bool|isASCII_LC_utf8_safe|U8 * s| U8 *end
766 Also note, that because all ASCII characters are UTF-8 invariant (meaning they
767 have the exact same representation (always a single byte) whether encoded in
768 UTF-8 or not), C<isASCII> will give the correct results when called with any
769 byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8_safe>
770 will work properly on any string encoded or not in UTF-8.
772 =for apidoc Am|bool|isBLANK|char ch
773 Returns a boolean indicating whether the specified character is a
774 character considered to be a blank, analogous to C<m/[[:blank:]]/>.
775 See the L<top of this section|/Character classification> for an explanation of
777 C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8_safe>,
778 C<isBLANK_LC>, C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8_safe>. Note,
779 however, that some platforms do not have the C library routine
780 C<isblank()>. In these cases, the variants whose names contain C<LC> are
781 the same as the corresponding ones without.
783 =for apidoc Amh|bool|isBLANK_A|int ch
784 =for apidoc Amh|bool|isBLANK_L1|int ch
785 =for apidoc Amh|bool|isBLANK_uvchr|int ch
786 =for apidoc Amh|bool|isBLANK_utf8_safe|U8 * s|U8 * end
787 =for apidoc Amh|bool|isBLANK_utf8|U8 * s
788 =for apidoc Amh|bool|isBLANK_LC|int ch
789 =for apidoc Amh|bool|isBLANK_LC_uvchr|int ch
790 =for apidoc Amh|bool|isBLANK_LC_utf8_safe|U8 * s| U8 *end
792 =for apidoc Am|bool|isCNTRL|char ch
793 Returns a boolean indicating whether the specified character is a
794 control character, analogous to C<m/[[:cntrl:]]/>.
795 See the L<top of this section|/Character classification> for an explanation of
797 C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8_safe>,
798 C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8_safe> On EBCDIC
799 platforms, you almost always want to use the C<isCNTRL_L1> variant.
801 =for apidoc Amh|bool|isCNTRL_A|int ch
802 =for apidoc Amh|bool|isCNTRL_L1|int ch
803 =for apidoc Amh|bool|isCNTRL_uvchr|int ch
804 =for apidoc Amh|bool|isCNTRL_utf8_safe|U8 * s|U8 * end
805 =for apidoc Amh|bool|isCNTRL_utf8|U8 * s
806 =for apidoc Amh|bool|isCNTRL_LC|int ch
807 =for apidoc Amh|bool|isCNTRL_LC_uvchr|int ch
808 =for apidoc Amh|bool|isCNTRL_LC_utf8_safe|U8 * s| U8 *end
810 =for apidoc Am|bool|isDIGIT|char ch
811 Returns a boolean indicating whether the specified character is a
812 digit, analogous to C<m/[[:digit:]]/>.
813 Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
814 See the L<top of this section|/Character classification> for an explanation of
816 C<isDIGIT_uvchr>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
817 C<isDIGIT_LC_utf8_safe>.
819 =for apidoc Amh|bool|isDIGIT_A|int ch
820 =for apidoc Amh|bool|isDIGIT_L1|int ch
821 =for apidoc Amh|bool|isDIGIT_uvchr|int ch
822 =for apidoc Amh|bool|isDIGIT_utf8_safe|U8 * s|U8 * end
823 =for apidoc Amh|bool|isDIGIT_utf8|U8 * s
824 =for apidoc Amh|bool|isDIGIT_LC|int ch
825 =for apidoc Amh|bool|isDIGIT_LC_uvchr|int ch
826 =for apidoc Amh|bool|isDIGIT_LC_utf8_safe|U8 * s| U8 *end
828 =for apidoc Am|bool|isGRAPH|char ch
829 Returns a boolean indicating whether the specified character is a
830 graphic character, analogous to C<m/[[:graph:]]/>.
831 See the L<top of this section|/Character classification> for an explanation of
832 variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8_safe>,
833 C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8_safe>.
835 =for apidoc Amh|bool|isGRAPH_A|int ch
836 =for apidoc Amh|bool|isGRAPH_L1|int ch
837 =for apidoc Amh|bool|isGRAPH_uvchr|int ch
838 =for apidoc Amh|bool|isGRAPH_utf8_safe|U8 * s|U8 * end
839 =for apidoc Amh|bool|isGRAPH_utf8|U8 * s
840 =for apidoc Amh|bool|isGRAPH_LC|int ch
841 =for apidoc Amh|bool|isGRAPH_LC_uvchr|int ch
842 =for apidoc Amh|bool|isGRAPH_LC_utf8_safe|U8 * s| U8 *end
844 =for apidoc Am|bool|isLOWER|char ch
845 Returns a boolean indicating whether the specified character is a
846 lowercase character, analogous to C<m/[[:lower:]]/>.
847 See the L<top of this section|/Character classification> for an explanation of
849 C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8_safe>,
850 C<isLOWER_LC>, C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8_safe>.
852 =for apidoc Amh|bool|isLOWER_A|int ch
853 =for apidoc Amh|bool|isLOWER_L1|int ch
854 =for apidoc Amh|bool|isLOWER_uvchr|int ch
855 =for apidoc Amh|bool|isLOWER_utf8_safe|U8 * s|U8 * end
856 =for apidoc Amh|bool|isLOWER_utf8|U8 * s
857 =for apidoc Amh|bool|isLOWER_LC|int ch
858 =for apidoc Amh|bool|isLOWER_LC_uvchr|int ch
859 =for apidoc Amh|bool|isLOWER_LC_utf8_safe|U8 * s| U8 *end
861 =for apidoc Am|bool|isOCTAL|char ch
862 Returns a boolean indicating whether the specified character is an
864 The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
867 =for apidoc Amh|bool|isOCTAL_A|int ch
868 =for apidoc Amh|bool|isOCTAL_L1|int ch
870 =for apidoc Am|bool|isPUNCT|char ch
871 Returns a boolean indicating whether the specified character is a
872 punctuation character, analogous to C<m/[[:punct:]]/>.
873 Note that the definition of what is punctuation isn't as
874 straightforward as one might desire. See L<perlrecharclass/POSIX Character
875 Classes> for details.
876 See the L<top of this section|/Character classification> for an explanation of
877 variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8_safe>,
878 C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8_safe>.
880 =for apidoc Amh|bool|isPUNCT_A|int ch
881 =for apidoc Amh|bool|isPUNCT_L1|int ch
882 =for apidoc Amh|bool|isPUNCT_uvchr|int ch
883 =for apidoc Amh|bool|isPUNCT_utf8_safe|U8 * s|U8 * end
884 =for apidoc Amh|bool|isPUNCT_utf8|U8 * s
885 =for apidoc Amh|bool|isPUNCT_LC|int ch
886 =for apidoc Amh|bool|isPUNCT_LC_uvchr|int ch
887 =for apidoc Amh|bool|isPUNCT_LC_utf8_safe|U8 * s| U8 *end
889 =for apidoc Am|bool|isSPACE|char ch
890 Returns a boolean indicating whether the specified character is a
891 whitespace character. This is analogous
892 to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
893 this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
894 locale forms of this macro (the ones with C<LC> in their names) matched
895 precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
896 in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
897 (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
898 See the L<top of this section|/Character classification> for an explanation of
900 C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8_safe>,
901 C<isSPACE_LC>, C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8_safe>.
903 =for apidoc Amh|bool|isSPACE_A|int ch
904 =for apidoc Amh|bool|isSPACE_L1|int ch
905 =for apidoc Amh|bool|isSPACE_uvchr|int ch
906 =for apidoc Amh|bool|isSPACE_utf8_safe|U8 * s|U8 * end
907 =for apidoc Amh|bool|isSPACE_utf8|U8 * s
908 =for apidoc Amh|bool|isSPACE_LC|int ch
909 =for apidoc Amh|bool|isSPACE_LC_uvchr|int ch
910 =for apidoc Amh|bool|isSPACE_LC_utf8_safe|U8 * s| U8 *end
912 =for apidoc Am|bool|isPSXSPC|char ch
913 (short for Posix Space)
914 Starting in 5.18, this is identical in all its forms to the
915 corresponding C<isSPACE()> macros.
916 The locale forms of this macro are identical to their corresponding
917 C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
918 non-locale forms differ from their C<isSPACE()> forms only in that the
919 C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
920 Otherwise they are identical. Thus this macro is analogous to what
921 C<m/[[:space:]]/> matches in a regular expression.
922 See the L<top of this section|/Character classification> for an explanation of
923 variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8_safe>,
924 C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8_safe>.
926 =for apidoc Amh|bool|isPSXSPC_A|int ch
927 =for apidoc Amh|bool|isPSXSPC_L1|int ch
928 =for apidoc Amh|bool|isPSXSPC_uvchr|int ch
929 =for apidoc Amh|bool|isPSXSPC_utf8_safe|U8 * s|U8 * end
930 =for apidoc Amh|bool|isPSXSPC_utf8|U8 * s
931 =for apidoc Amh|bool|isPSXSPC_LC|int ch
932 =for apidoc Amh|bool|isPSXSPC_LC_uvchr|int ch
933 =for apidoc Amh|bool|isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
935 =for apidoc Am|bool|isUPPER|char ch
936 Returns a boolean indicating whether the specified character is an
937 uppercase character, analogous to C<m/[[:upper:]]/>.
938 See the L<top of this section|/Character classification> for an explanation of
939 variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8_safe>,
940 C<isUPPER_LC>, C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8_safe>.
942 =for apidoc Amh|bool|isUPPER_A|int ch
943 =for apidoc Amh|bool|isUPPER_L1|int ch
944 =for apidoc Amh|bool|isUPPER_uvchr|int ch
945 =for apidoc Amh|bool|isUPPER_utf8_safe|U8 * s|U8 * end
946 =for apidoc Amh|bool|isUPPER_utf8|U8 * s
947 =for apidoc Amh|bool|isUPPER_LC|int ch
948 =for apidoc Amh|bool|isUPPER_LC_uvchr|int ch
949 =for apidoc Amh|bool|isUPPER_LC_utf8_safe|U8 * s| U8 *end
951 =for apidoc Am|bool|isPRINT|char ch
952 Returns a boolean indicating whether the specified character is a
953 printable character, analogous to C<m/[[:print:]]/>.
954 See the L<top of this section|/Character classification> for an explanation of
956 C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8_safe>,
957 C<isPRINT_LC>, C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8_safe>.
959 =for apidoc Amh|bool|isPRINT_A|int ch
960 =for apidoc Amh|bool|isPRINT_L1|int ch
961 =for apidoc Amh|bool|isPRINT_uvchr|int ch
962 =for apidoc Amh|bool|isPRINT_utf8_safe|U8 * s|U8 * end
963 =for apidoc Amh|bool|isPRINT_utf8|U8 * s
964 =for apidoc Amh|bool|isPRINT_LC|int ch
965 =for apidoc Amh|bool|isPRINT_LC_uvchr|int ch
966 =for apidoc Amh|bool|isPRINT_LC_utf8_safe|U8 * s| U8 *end
968 =for apidoc Am|bool|isWORDCHAR|char ch
969 Returns a boolean indicating whether the specified character is a character
970 that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
971 in a regular expression. A word character is an alphabetic character, a
972 decimal digit, a connecting punctuation character (such as an underscore), or
973 a "mark" character that attaches to one of those (like some sort of accent).
974 C<isALNUM()> is a synonym provided for backward compatibility, even though a
975 word character includes more than the standard C language meaning of
977 See the L<top of this section|/Character classification> for an explanation of
978 variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and
979 C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and
980 C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
981 include the platform's native underscore.
983 =for apidoc Amh|bool|isWORDCHAR_A|int ch
984 =for apidoc Amh|bool|isWORDCHAR_L1|int ch
985 =for apidoc Amh|bool|isWORDCHAR_uvchr|int ch
986 =for apidoc Amh|bool|isWORDCHAR_utf8_safe|U8 * s|U8 * end
987 =for apidoc Amh|bool|isWORDCHAR_utf8|U8 * s
988 =for apidoc Amh|bool|isWORDCHAR_LC|int ch
989 =for apidoc Amh|bool|isWORDCHAR_LC_uvchr|int ch
990 =for apidoc Amh|bool|isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
991 =for apidoc Amh|bool|isALNUM|int ch
992 =for apidoc Amh|bool|isALNUM_A|int ch
993 =for apidoc Amh|bool|isALNUM_LC|int ch
994 =for apidoc Amh|bool|isALNUM_LC_uvchr|int ch
996 =for apidoc Am|bool|isXDIGIT|char ch
997 Returns a boolean indicating whether the specified character is a hexadecimal
998 digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
999 and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
1000 See the L<top of this section|/Character classification> for an explanation of
1002 C<isXDIGIT_uvchr>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>,
1003 and C<isXDIGIT_LC_utf8_safe>.
1005 =for apidoc Amh|bool|isXDIGIT_A|int ch
1006 =for apidoc Amh|bool|isXDIGIT_L1|int ch
1007 =for apidoc Amh|bool|isXDIGIT_uvchr|int ch
1008 =for apidoc Amh|bool|isXDIGIT_utf8_safe|U8 * s|U8 * end
1009 =for apidoc Amh|bool|isXDIGIT_utf8|U8 * s
1010 =for apidoc Amh|bool|isXDIGIT_LC|int ch
1011 =for apidoc Amh|bool|isXDIGIT_LC_uvchr|int ch
1012 =for apidoc Amh|bool|isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1014 =for apidoc Am|bool|isIDFIRST|char ch
1015 Returns a boolean indicating whether the specified character can be the first
1016 character of an identifier. This is very close to, but not quite the same as
1017 the official Unicode property C<XID_Start>. The difference is that this
1018 returns true only if the input character also matches L</isWORDCHAR>.
1019 See the L<top of this section|/Character classification> for an explanation of
1021 C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8_safe>,
1022 C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8_safe>.
1024 =for apidoc Amh|bool|isIDFIRST_A|int ch
1025 =for apidoc Amh|bool|isIDFIRST_L1|int ch
1026 =for apidoc Amh|bool|isIDFIRST_uvchr|int ch
1027 =for apidoc Amh|bool|isIDFIRST_utf8_safe|U8 * s|U8 * end
1028 =for apidoc Amh|bool|isIDFIRST_utf8|U8 * s
1029 =for apidoc Amh|bool|isIDFIRST_LC|int ch
1030 =for apidoc Amh|bool|isIDFIRST_LC_uvchr|int ch
1031 =for apidoc Amh|bool|isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1033 =for apidoc Am|bool|isIDCONT|char ch
1034 Returns a boolean indicating whether the specified character can be the
1035 second or succeeding character of an identifier. This is very close to, but
1036 not quite the same as the official Unicode property C<XID_Continue>. The
1037 difference is that this returns true only if the input character also matches
1038 L</isWORDCHAR>. See the L<top of this section|/Character classification> for
1040 explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
1041 C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
1042 C<isIDCONT_LC_utf8_safe>.
1044 =for apidoc Amh|bool|isIDCONT_A|int ch
1045 =for apidoc Amh|bool|isIDCONT_L1|int ch
1046 =for apidoc Amh|bool|isIDCONT_uvchr|int ch
1047 =for apidoc Amh|bool|isIDCONT_utf8_safe|U8 * s|U8 * end
1048 =for apidoc Amh|bool|isIDCONT_utf8|U8 * s
1049 =for apidoc Amh|bool|isIDCONT_LC|int ch
1050 =for apidoc Amh|bool|isIDCONT_LC_uvchr|int ch
1051 =for apidoc Amh|bool|isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1053 =head1 Miscellaneous Functions
1055 =for apidoc Am|U8|READ_XDIGIT|char str*
1056 Returns the value of an ASCII-range hex digit and advances the string pointer.
1057 Behaviour is only well defined when isXDIGIT(*str) is true.
1059 =head1 Character case changing
1060 Perl uses "full" Unicode case mappings. This means that converting a single
1061 character to another case may result in a sequence of more than one character.
1062 For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1063 character sequence C<SS>. This presents some complications The lowercase of
1064 all characters in the range 0..255 is a single character, and thus
1065 C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1066 return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1067 an API that does allow every possible legal result to be returned.) Likewise
1068 no other function that is crippled by not being able to give the correct
1069 results for the full range of possible inputs has been implemented here.
1071 =for apidoc Am|U8|toUPPER|int ch
1072 Converts the specified character to uppercase. If the input is anything but an
1073 ASCII lowercase character, that input character itself is returned. Variant
1074 C<toUPPER_A> is equivalent.
1076 =for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1077 Converts the code point C<cp> to its uppercase version, and
1078 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1079 point is interpreted as native if less than 256; otherwise as Unicode. Note
1080 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1081 bytes since the uppercase version may be longer than the original character.
1083 The first code point of the uppercased version is returned
1084 (but note, as explained at L<the top of this section|/Character case
1085 changing>, that there may be more.)
1087 =for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1088 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1089 extending no further than S<C<e - 1>> to its uppercase version, and
1090 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1091 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1092 bytes since the uppercase version may be longer than the original character.
1094 The first code point of the uppercased version is returned
1095 (but note, as explained at L<the top of this section|/Character case
1096 changing>, that there may be more).
1098 The suffix C<_safe> in the function's name indicates that it will not attempt
1099 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1100 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1101 input character is malformed in some way, the program may croak, or the
1102 function may return the REPLACEMENT CHARACTER, at the discretion of the
1103 implementation, and subject to change in future releases.
1105 =for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
1106 This is like C<L</toUPPER_utf8_safe>>, but doesn't have the C<e>
1107 parameter The function therefore can't check if it is reading
1108 beyond the end of the string. Starting in Perl v5.32, it will take the C<e>
1109 parameter, becoming a synonym for C<toUPPER_utf8_safe>. At that time every
1110 program that uses it will have to be changed to successfully compile. In the
1111 meantime, the first runtime call to C<toUPPER_utf8> from each call point in the
1112 program will raise a deprecation warning, enabled by default. You can convert
1113 your program now to use C<toUPPER_utf8_safe>, and avoid the warnings, and get an
1114 extra measure of protection, or you can wait until v5.32, when you'll be forced
1115 to add the C<e> parameter.
1117 =for apidoc Am|U8|toFOLD|U8 ch
1118 Converts the specified character to foldcase. If the input is anything but an
1119 ASCII uppercase character, that input character itself is returned. Variant
1120 C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
1121 Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
1123 =for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1124 Converts the code point C<cp> to its foldcase version, and
1125 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1126 point is interpreted as native if less than 256; otherwise as Unicode. Note
1127 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1128 bytes since the foldcase version may be longer than the original character.
1130 The first code point of the foldcased version is returned
1131 (but note, as explained at L<the top of this section|/Character case
1132 changing>, that there may be more).
1134 =for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1135 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1136 extending no further than S<C<e - 1>> to its foldcase version, and
1137 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1138 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1139 bytes since the foldcase version may be longer than the original character.
1141 The first code point of the foldcased version is returned
1142 (but note, as explained at L<the top of this section|/Character case
1143 changing>, that there may be more).
1145 The suffix C<_safe> in the function's name indicates that it will not attempt
1146 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1147 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1148 input character is malformed in some way, the program may croak, or the
1149 function may return the REPLACEMENT CHARACTER, at the discretion of the
1150 implementation, and subject to change in future releases.
1152 =for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
1153 This is like C<L</toFOLD_utf8_safe>>, but doesn't have the C<e>
1154 parameter The function therefore can't check if it is reading
1155 beyond the end of the string. Starting in Perl v5.32, it will take the C<e>
1156 parameter, becoming a synonym for C<toFOLD_utf8_safe>. At that time every
1157 program that uses it will have to be changed to successfully compile. In the
1158 meantime, the first runtime call to C<toFOLD_utf8> from each call point in the
1159 program will raise a deprecation warning, enabled by default. You can convert
1160 your program now to use C<toFOLD_utf8_safe>, and avoid the warnings, and get an
1161 extra measure of protection, or you can wait until v5.32, when you'll be forced
1162 to add the C<e> parameter.
1164 =for apidoc Am|U8|toLOWER|U8 ch
1165 Converts the specified character to lowercase. If the input is anything but an
1166 ASCII uppercase character, that input character itself is returned. Variant
1167 C<toLOWER_A> is equivalent.
1169 =for apidoc Am|U8|toLOWER_L1|U8 ch
1170 Converts the specified Latin1 character to lowercase. The results are
1171 undefined if the input doesn't fit in a byte.
1173 =for apidoc Am|U8|toLOWER_LC|U8 ch
1174 Converts the specified character to lowercase using the current locale's rules,
1175 if possible; otherwise returns the input character itself.
1177 =for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1178 Converts the code point C<cp> to its lowercase version, and
1179 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1180 point is interpreted as native if less than 256; otherwise as Unicode. Note
1181 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1182 bytes since the lowercase version may be longer than the original character.
1184 The first code point of the lowercased version is returned
1185 (but note, as explained at L<the top of this section|/Character case
1186 changing>, that there may be more).
1189 =for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1190 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1191 extending no further than S<C<e - 1>> to its lowercase version, and
1192 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1193 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1194 bytes since the lowercase version may be longer than the original character.
1196 The first code point of the lowercased version is returned
1197 (but note, as explained at L<the top of this section|/Character case
1198 changing>, that there may be more).
1200 The suffix C<_safe> in the function's name indicates that it will not attempt
1201 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1202 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1203 input character is malformed in some way, the program may croak, or the
1204 function may return the REPLACEMENT CHARACTER, at the discretion of the
1205 implementation, and subject to change in future releases.
1207 =for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
1208 This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1209 parameter The function therefore can't check if it is reading
1210 beyond the end of the string. Starting in Perl v5.32, it will take the C<e>
1211 parameter, becoming a synonym for C<toLOWER_utf8_safe>. At that time every
1212 program that uses it will have to be changed to successfully compile. In the
1213 meantime, the first runtime call to C<toLOWER_utf8> from each call point in the
1214 program will raise a deprecation warning, enabled by default. You can convert
1215 your program now to use C<toLOWER_utf8_safe>, and avoid the warnings, and get an
1216 extra measure of protection, or you can wait until v5.32, when you'll be forced
1217 to add the C<e> parameter.
1219 =for apidoc Am|U8|toTITLE|U8 ch
1220 Converts the specified character to titlecase. If the input is anything but an
1221 ASCII lowercase character, that input character itself is returned. Variant
1222 C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
1223 range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
1224 not a concept used in locale handling, so there is no functionality for that.)
1226 =for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1227 Converts the code point C<cp> to its titlecase version, and
1228 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1229 point is interpreted as native if less than 256; otherwise as Unicode. Note
1230 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1231 bytes since the titlecase version may be longer than the original character.
1233 The first code point of the titlecased version is returned
1234 (but note, as explained at L<the top of this section|/Character case
1235 changing>, that there may be more).
1237 =for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1238 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1239 extending no further than S<C<e - 1>> to its titlecase version, and
1240 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1241 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1242 bytes since the titlecase version may be longer than the original character.
1244 The first code point of the titlecased version is returned
1245 (but note, as explained at L<the top of this section|/Character case
1246 changing>, that there may be more).
1248 The suffix C<_safe> in the function's name indicates that it will not attempt
1249 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1250 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1251 input character is malformed in some way, the program may croak, or the
1252 function may return the REPLACEMENT CHARACTER, at the discretion of the
1253 implementation, and subject to change in future releases.
1255 =for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
1256 This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1257 parameter The function therefore can't check if it is reading
1258 beyond the end of the string. Starting in Perl v5.32, it will take the C<e>
1259 parameter, becoming a synonym for C<toTITLE_utf8_safe>. At that time every
1260 program that uses it will have to be changed to successfully compile. In the
1261 meantime, the first runtime call to C<toTITLE_utf8> from each call point in the
1262 program will raise a deprecation warning, enabled by default. You can convert
1263 your program now to use C<toTITLE_utf8_safe>, and avoid the warnings, and get an
1264 extra measure of protection, or you can wait until v5.32, when you'll be forced
1265 to add the C<e> parameter.
1269 XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1270 really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1271 and aren't general purpose as they don't work on U+DF, and assert against that.
1273 Note that these macros are repeated in Devel::PPPort, so should also be
1274 patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1279 void below because that's the best fit, and works for Devel::PPPort
1280 =for apidoc AmnU|void|WIDEST_UTYPE
1282 Yields the widest unsigned integer type on the platform, currently either
1283 C<U32> or C<64>. This can be used in declarations such as
1289 my_uv = (WIDEST_UTYPE) val;
1295 # define WIDEST_UTYPE U64
1297 # define WIDEST_UTYPE U32
1300 /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1301 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1302 * bits of information are lost even on a 64-bit machine, but to get the
1303 * compiler to optimize it out if possible. This is because Configure makes
1304 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1305 * sizeof() guarantees that this evaluates to a constant true at compile time.
1307 * For Coverity, be always true, because otherwise Coverity thinks
1308 * it finds several expressions that are always true, independent
1309 * of operands. Well, they are, but that is kind of the point.
1311 #ifndef __COVERITY__
1312 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1314 #define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
1315 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
1317 #define FITS_IN_8_BITS(c) (1)
1320 /* Returns true if l <= c <= l + n, where 'l' and 'n' are non-negative
1321 * Written this way so that after optimization, only one conditional test is
1323 #define withinCOUNT(c, l, n) (__ASSERT_((l) >= 0) __ASSERT_((n) >= (0)) \
1324 (((WIDEST_UTYPE) (((c) | 0) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
1326 /* Returns true if c is in the range l..u, where 'l' is non-negative
1327 * Written this way so that after optimization, only one conditional test is
1329 #define inRANGE(c, l, u) (__ASSERT_((l) >= 0) __ASSERT_((u) >= (l)) \
1330 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
1331 : (sizeof(c) == sizeof(U16)) ? withinCOUNT(((U16) (c)), (l), ((u) - (l))) \
1332 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1333 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
1334 withinCOUNT(( (c)), (l), ((u) - (l))))))
1337 # ifndef _ALL_SOURCE
1338 /* The native libc isascii() et.al. functions return the wrong results
1339 * on at least z/OS unless this is defined. */
1340 # error _ALL_SOURCE should probably be defined
1343 /* There is a simple definition of ASCII for ASCII platforms. But the
1344 * EBCDIC one isn't so simple, so is defined using table look-up like the
1345 * other macros below.
1347 * The cast here is used instead of '(c) >= 0', because some compilers emit
1348 * a warning that that test is always true when the parameter is an
1349 * unsigned type. khw supposes that it could be written as
1350 * && ((c) == '\0' || (c) > 0)
1351 * to avoid the message, but the cast will likely avoid extra branches even
1352 * with stupid compilers.
1354 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1356 # define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
1359 /* Take the eight possible bit patterns of the lower 3 bits and you get the
1360 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1361 * can be ignored. If the rest match '0', we have an octal */
1362 #define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
1364 #ifdef H_PERL /* If have access to perl.h, lookup in its table */
1366 /* Character class numbers. For internal core Perl use only. The ones less
1367 * than 32 are used in PL_charclass[] and the ones up through the one that
1368 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1369 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1370 * their actual definitions are here. If that file has a name not used here,
1373 * The first group of these is ordered in what I (khw) estimate to be the
1374 * frequency of their use. This gives a slight edge to exiting a loop earlier
1375 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1376 * real Posix character class, and some (small) inefficiencies in regular
1377 * expression handling would be introduced by putting it in the middle of those
1378 * that are. Also, cntrl and ascii come after the others as it may be useful
1379 * to group these which have no members that match above Latin1, (or above
1380 * ASCII in the latter case) */
1382 # define _CC_WORDCHAR 0 /* \w and [:word:] */
1383 # define _CC_DIGIT 1 /* \d and [:digit:] */
1384 # define _CC_ALPHA 2 /* [:alpha:] */
1385 # define _CC_LOWER 3 /* [:lower:] */
1386 # define _CC_UPPER 4 /* [:upper:] */
1387 # define _CC_PUNCT 5 /* [:punct:] */
1388 # define _CC_PRINT 6 /* [:print:] */
1389 # define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1390 # define _CC_GRAPH 8 /* [:graph:] */
1391 # define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1392 # define _CC_SPACE 10 /* \s, [:space:] */
1393 # define _CC_PSXSPC _CC_SPACE /* XXX Temporary, can be removed
1394 when the deprecated isFOO_utf8()
1395 functions are removed */
1396 # define _CC_BLANK 11 /* [:blank:] */
1397 # define _CC_XDIGIT 12 /* [:xdigit:] */
1398 # define _CC_CNTRL 13 /* [:cntrl:] */
1399 # define _CC_ASCII 14 /* [:ascii:] */
1400 # define _CC_VERTSPACE 15 /* \v */
1402 # define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1404 /* The members of the third group below do not need to be coordinated with data
1405 * structures in regcomp.[ch] and regexec.c. */
1406 # define _CC_IDFIRST 16
1407 # define _CC_CHARNAME_CONT 17
1408 # define _CC_NONLATIN1_FOLD 18
1409 # define _CC_NONLATIN1_SIMPLE_FOLD 19
1410 # define _CC_QUOTEMETA 20
1411 # define _CC_NON_FINAL_FOLD 21
1412 # define _CC_IS_IN_SOME_FOLD 22
1413 # define _CC_MNEMONIC_CNTRL 23
1415 # define _CC_IDCONT 24 /* XXX Temporary, can be removed when the deprecated
1416 isFOO_utf8() functions are removed */
1418 /* This next group is only used on EBCDIC platforms, so theoretically could be
1419 * shared with something entirely different that's only on ASCII platforms */
1420 # define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1421 # define _CC_UTF8_IS_START 29
1422 # define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1423 # define _CC_UTF8_IS_CONTINUATION 31
1425 * If more bits are needed, one could add a second word for non-64bit
1426 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
1427 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1428 * is used only for optimization (as of this writing), and differs in the
1429 * Latin1 range from the ALPHA bit only in two relatively unimportant
1430 * characters: the masculine and feminine ordinal indicators, so removing it
1431 * would just cause /i regexes which match them to run less efficiently.
1432 * Similarly the EBCDIC-only bits are used just for speed, and could be
1433 * replaced by other means */
1435 #if defined(PERL_CORE) || defined(PERL_EXT)
1436 /* An enum version of the character class numbers, to help compilers
1439 _CC_ENUM_ALPHA = _CC_ALPHA,
1440 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1441 _CC_ENUM_ASCII = _CC_ASCII,
1442 _CC_ENUM_BLANK = _CC_BLANK,
1443 _CC_ENUM_CASED = _CC_CASED,
1444 _CC_ENUM_CNTRL = _CC_CNTRL,
1445 _CC_ENUM_DIGIT = _CC_DIGIT,
1446 _CC_ENUM_GRAPH = _CC_GRAPH,
1447 _CC_ENUM_LOWER = _CC_LOWER,
1448 _CC_ENUM_PRINT = _CC_PRINT,
1449 _CC_ENUM_PUNCT = _CC_PUNCT,
1450 _CC_ENUM_SPACE = _CC_SPACE,
1451 _CC_ENUM_UPPER = _CC_UPPER,
1452 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
1453 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
1454 _CC_ENUM_XDIGIT = _CC_XDIGIT
1455 } _char_class_number;
1458 #define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
1462 EXTCONST U32 PL_charclass[] = {
1463 # include "l1_char_class_tab.h"
1466 # else /* ! DOINIT */
1467 EXTCONST U32 PL_charclass[];
1471 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
1472 # define _CC_mask(classnum) (1U << (classnum))
1474 /* For internal core Perl use only: the base macro for defining macros like
1476 # define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
1477 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
1479 /* The mask for the _A versions of the macros; it just adds in the bit for
1481 # define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1483 /* For internal core Perl use only: the base macro for defining macros like
1484 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1485 * the ASCII bit are present */
1486 # define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1487 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1488 == _CC_mask_A(classnum)))
1490 /* On ASCII platforms certain classes form a single range. It's faster to
1491 * special case these. isDIGIT is a single range on all platforms */
1493 # define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1494 # define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1495 # define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1496 # define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1497 # define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1499 /* By folding the upper and lowercase, we can use a single range */
1500 # define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
1501 # define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
1502 # define isLOWER_A(c) inRANGE(c, 'a', 'z')
1503 # define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1504 # define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1506 # define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
1507 # define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1508 # define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
1509 # define isDIGIT_A(c) inRANGE(c, '0', '9')
1510 # define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1511 # define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1512 # define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
1513 # define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1515 # define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
1516 # define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1517 # define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1518 # define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1520 /* continuation character for legal NAME in \N{NAME} */
1521 # define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1523 # define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1524 # define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1525 # define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1526 # define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1527 # define isPSXSPC_L1(c) isSPACE_L1(c)
1528 # define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1529 # define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1530 # define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1531 # define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1532 # define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
1535 # define isASCII(c) _generic_isCC(c, _CC_ASCII)
1538 /* Participates in a single-character fold with a character above 255 */
1539 # 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)))
1541 /* Like the above, but also can be part of a multi-char fold */
1542 # 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)))
1544 # define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
1545 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1546 _generic_isCC(c, _CC_NON_FINAL_FOLD)
1547 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1548 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1549 # define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1550 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1551 #else /* else we don't have perl.h H_PERL */
1553 /* If we don't have perl.h, we are compiling a utility program. Below we
1554 * hard-code various macro definitions that wouldn't otherwise be available
1555 * to it. Most are coded based on first principles. These are written to
1556 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
1557 # define isDIGIT_A(c) inRANGE(c, '0', '9')
1558 # define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1559 # define isSPACE_A(c) (isBLANK_A(c) \
1564 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1565 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1566 * performance (if optimization isn't on), and allow the same code to be
1567 * used for both platform types */
1568 # define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1569 || inRANGE((c), 'j', 'r') \
1570 || inRANGE((c), 's', 'z')
1571 # define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1572 || inRANGE((c), 'J', 'R') \
1573 || inRANGE((c), 'S', 'Z')
1574 # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1575 # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
1576 # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
1577 # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1578 # define isXDIGIT_A(c) ( isDIGIT_A(c) \
1579 || inRANGE((c), 'a', 'f') \
1580 || inRANGE((c), 'A', 'F')
1581 # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1582 || (c) == '#' || (c) == '$' || (c) == '%' \
1583 || (c) == '&' || (c) == '\'' || (c) == '(' \
1584 || (c) == ')' || (c) == '*' || (c) == '+' \
1585 || (c) == ',' || (c) == '.' || (c) == '/' \
1586 || (c) == ':' || (c) == ';' || (c) == '<' \
1587 || (c) == '=' || (c) == '>' || (c) == '?' \
1588 || (c) == '@' || (c) == '[' || (c) == '\\' \
1589 || (c) == ']' || (c) == '^' || (c) == '_' \
1590 || (c) == '`' || (c) == '{' || (c) == '|' \
1591 || (c) == '}' || (c) == '~')
1592 # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1593 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1596 /* The below is accurate for the 3 EBCDIC code pages traditionally
1597 * supported by perl. The only difference between them in the controls
1598 * is the position of \n, and that is represented symbolically below */
1599 # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1600 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1601 || (c) == '\t' || (c) == '\v' \
1602 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
1603 || (c) == 7 /* U+7F DEL */ \
1604 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1606 || (c) == 0x18 /* U+18 CAN */ \
1607 || (c) == 0x19 /* U+19 EOM */ \
1608 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
1609 || (c) == 0x26 /* U+17 ETB */ \
1610 || (c) == 0x27 /* U+1B ESC */ \
1611 || (c) == 0x2D /* U+05 ENQ */ \
1612 || (c) == 0x2E /* U+06 ACK */ \
1613 || (c) == 0x32 /* U+16 SYN */ \
1614 || (c) == 0x37 /* U+04 EOT */ \
1615 || (c) == 0x3C /* U+14 DC4 */ \
1616 || (c) == 0x3D /* U+15 NAK */ \
1617 || (c) == 0x3F)/* U+1A SUB */
1618 # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1619 # else /* isASCII is already defined for ASCII platforms, so can use that to
1621 # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
1624 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
1625 * during debugging, and it seems best to keep them. We may be called
1626 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1627 * do anything anyway, so make it not a problem */
1628 # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1629 # define NATIVE_TO_LATIN1(ch) (ch)
1631 # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1632 # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1633 # define isBLANK_L1(c) (isBLANK_A(c) \
1634 || (FITS_IN_8_BITS(c) \
1635 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1636 # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1637 # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1638 # define isLOWER_L1(c) (isLOWER_A(c) \
1639 || (FITS_IN_8_BITS(c) \
1640 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1641 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1642 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1643 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1644 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1645 # define isPRINT_L1(c) (isPRINT_A(c) \
1646 || (FITS_IN_8_BITS(c) \
1647 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1648 # define isPUNCT_L1(c) (isPUNCT_A(c) \
1649 || (FITS_IN_8_BITS(c) \
1650 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1651 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1652 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1653 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1654 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1655 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1656 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1657 # define isSPACE_L1(c) (isSPACE_A(c) \
1658 || (FITS_IN_8_BITS(c) \
1659 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
1660 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1661 # define isUPPER_L1(c) (isUPPER_A(c) \
1662 || (FITS_IN_8_BITS(c) \
1663 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1665 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1666 # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1667 # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1668 # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1673 /* The following are not fully accurate in the above-ASCII range. I (khw)
1674 * don't think it's necessary to be so for the purposes where this gets
1676 # define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1677 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1679 /* And these aren't accurate at all. They are useful only for above
1680 * Latin1, which utilities and bootstrapping don't deal with */
1681 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
1682 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1683 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1685 /* Many of the macros later in this file are defined in terms of these. By
1686 * implementing them with a function, which converts the class number into
1687 * a call to the desired macro, all of the later ones work. However, that
1688 * function won't be actually defined when building a utility program (no
1689 * perl.h), and so a compiler error will be generated if one is attempted
1690 * to be used. And the above-Latin1 code points require Unicode tables to
1691 * be present, something unlikely to be the case when bootstrapping */
1692 # define _generic_isCC(c, classnum) \
1693 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1694 # define _generic_isCC_A(c, classnum) \
1695 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
1696 #endif /* End of no perl.h H_PERL */
1698 #define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1699 #define isALPHA(c) isALPHA_A(c)
1700 #define isASCII_A(c) isASCII(c)
1701 #define isASCII_L1(c) isASCII(c)
1702 #define isBLANK(c) isBLANK_A(c)
1703 #define isCNTRL(c) isCNTRL_A(c)
1704 #define isDIGIT(c) isDIGIT_A(c)
1705 #define isGRAPH(c) isGRAPH_A(c)
1706 #define isIDFIRST(c) isIDFIRST_A(c)
1707 #define isLOWER(c) isLOWER_A(c)
1708 #define isPRINT(c) isPRINT_A(c)
1709 #define isPSXSPC_A(c) isSPACE_A(c)
1710 #define isPSXSPC(c) isPSXSPC_A(c)
1711 #define isPSXSPC_L1(c) isSPACE_L1(c)
1712 #define isPUNCT(c) isPUNCT_A(c)
1713 #define isSPACE(c) isSPACE_A(c)
1714 #define isUPPER(c) isUPPER_A(c)
1715 #define isWORDCHAR(c) isWORDCHAR_A(c)
1716 #define isXDIGIT(c) isXDIGIT_A(c)
1718 /* ASCII casing. These could also be written as
1719 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1720 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1721 which uses table lookup and mask instead of subtraction. (This would
1722 work because the _MOD does not apply in the ASCII range).
1724 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1725 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1726 there non-ASCII invariants, and all of them are controls.) */
1727 #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1728 #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
1730 /* In the ASCII range, these are equivalent to what they're here defined to be.
1731 * But by creating these definitions, other code doesn't have to be aware of
1732 * this detail. Actually this works for all UTF-8 invariants, not just the
1733 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
1734 #define toFOLD(c) toLOWER(c)
1735 #define toTITLE(c) toUPPER(c)
1737 #define toLOWER_A(c) toLOWER(c)
1738 #define toUPPER_A(c) toUPPER(c)
1739 #define toFOLD_A(c) toFOLD(c)
1740 #define toTITLE_A(c) toTITLE(c)
1742 /* Use table lookup for speed; returns the input itself if is out-of-range */
1743 #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
1745 : PL_latin1_lc[ (U8) (c) ])
1746 #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1748 /* Modified uc. Is correct uc except for three non-ascii chars which are
1749 * all mapped to one of them, and these need special handling; returns the
1750 * input itself if is out-of-range */
1751 #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
1753 : PL_mod_latin1_uc[ (U8) (c) ])
1754 #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
1756 /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1758 /* For internal core Perl use only: the base macro for defining macros like
1759 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
1760 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1761 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1762 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1763 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1764 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1765 * code points under locale rules, use the macros based on _generic_LC_uvchr
1766 * instead of this. */
1767 #define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1768 (! FITS_IN_8_BITS(c) \
1770 : IN_UTF8_CTYPE_LOCALE \
1771 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
1774 /* For internal core Perl use only: a helper macro for defining macros like
1775 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1776 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1777 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1778 * _generic_LC_base for more info */
1779 #define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1780 _generic_LC_base(c,utf8_locale_classnum, \
1781 non_utf8_func( (_LC_CAST) (c)))
1783 /* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1784 * 'c' is the platform's native underscore character */
1785 #define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1786 _generic_LC_base(c, utf8_locale_classnum, \
1787 (non_utf8_func( (_LC_CAST) (c)) \
1788 || (char)(c) == '_'))
1790 /* These next three are also for internal core Perl use only: case-change
1791 * helper macros. The reason for using the PL_latin arrays is in case the
1792 * system function is defective; it ensures uniform results that conform to the
1793 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
1794 #define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1796 : (IN_UTF8_CTYPE_LOCALE) \
1797 ? PL_latin1_lc[ (U8) (c) ] \
1798 : (cast)function((cast)(c)))
1800 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1801 * returns a single value, so can't adequately return the upper case of LATIN
1802 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1803 * values "SS"); instead it asserts against that under DEBUGGING, and
1804 * otherwise returns its input. It does not handle the anomalies in UTF-8
1805 * Turkic locales. */
1806 #define _generic_toUPPER_LC(c, function, cast) \
1807 (! FITS_IN_8_BITS(c) \
1809 : ((! IN_UTF8_CTYPE_LOCALE) \
1810 ? (cast)function((cast)(c)) \
1811 : ((((U8)(c)) == MICRO_SIGN) \
1812 ? GREEK_CAPITAL_LETTER_MU \
1813 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1814 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1815 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1816 ? (__ASSERT_(0) (c)) \
1817 : PL_mod_latin1_uc[ (U8) (c) ])))))
1819 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1820 * returns a single value, so can't adequately return the fold case of LATIN
1821 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1822 * values "ss"); instead it asserts against that under DEBUGGING, and
1823 * otherwise returns its input. It does not handle the anomalies in UTF-8
1825 #define _generic_toFOLD_LC(c, function, cast) \
1826 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
1827 ? GREEK_SMALL_LETTER_MU \
1828 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1829 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1830 _generic_toLOWER_LC(c, function, cast)))
1832 /* Use the libc versions for these if available. */
1833 #if defined(HAS_ISASCII)
1834 # define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1836 # define isASCII_LC(c) isASCII(c)
1839 #if defined(HAS_ISBLANK)
1840 # define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
1841 #else /* Unlike isASCII, varies if in a UTF-8 locale */
1842 # define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
1848 /* The Windows functions don't bother to follow the POSIX standard, which
1849 * for example says that something can't both be a printable and a control.
1850 * But Windows treats the \t control as a printable, and does such things
1851 * as making superscripts into both digits and punctuation. This tames
1852 * these flaws by assuming that the definitions of both controls and space
1853 * are correct, and then making sure that other definitions don't have
1854 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1855 * Not all possible weirdnesses are checked for, just the ones that were
1856 * detected on actual Microsoft code pages */
1858 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1859 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1861 # define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1862 && isALPHANUMERIC_LC(c))
1863 # define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1865 # define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1866 isALPHANUMERIC_LC(c))
1867 # define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1868 # define isIDFIRST_LC(c) (((c) == '_') \
1869 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1870 # define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1871 # define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1872 # define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1873 # define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1874 # define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1875 # define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1876 && isALPHANUMERIC_LC(c))
1878 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1879 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1880 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1882 #elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
1883 /* For most other platforms */
1885 # define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1886 # define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1887 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1888 # define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1889 # define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1890 # define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1891 # define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1892 # define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1893 # define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1894 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1895 # define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1896 # define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1897 # define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1900 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1901 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1902 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1904 #else /* The final fallback position */
1906 # define isALPHA_LC(c) (isascii(c) && isalpha(c))
1907 # define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1908 # define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1909 # define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1910 # define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1911 # define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1912 # define isLOWER_LC(c) (isascii(c) && islower(c))
1913 # define isPRINT_LC(c) (isascii(c) && isprint(c))
1914 # define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1915 # define isSPACE_LC(c) (isascii(c) && isspace(c))
1916 # define isUPPER_LC(c) (isascii(c) && isupper(c))
1917 # define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1918 # define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1920 # define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1921 # define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1922 # define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
1926 #define isIDCONT(c) isWORDCHAR(c)
1927 #define isIDCONT_A(c) isWORDCHAR_A(c)
1928 #define isIDCONT_L1(c) isWORDCHAR_L1(c)
1929 #define isIDCONT_LC(c) isWORDCHAR_LC(c)
1930 #define isPSXSPC_LC(c) isSPACE_LC(c)
1932 /* For internal core Perl use only: the base macros for defining macros like
1933 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1934 * number defined earlier in this file. _generic_uvchr() is used for POSIX
1935 * classes where there is a macro or function 'above_latin1' that takes the
1936 * single argument 'c' and returns the desired value. These exist for those
1937 * classes which have simple definitions, avoiding the overhead of a hash
1938 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
1939 * for classes where that overhead is faster than a direct lookup.
1940 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
1941 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1942 * have duplicate checks here, so could create versions of the macros that
1943 * don't, but experiments show that gcc optimizes them out anyway. */
1945 /* Note that all ignore 'use bytes' */
1946 #define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1947 ? _generic_isCC(c, classnum) \
1949 #define _generic_swash_uvchr(classnum, c) ((c) < 256 \
1950 ? _generic_isCC(c, classnum) \
1951 : _is_uni_FOO(classnum, c))
1952 #define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1953 #define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1954 #define isASCII_uvchr(c) isASCII(c)
1955 #define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1956 #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1957 #define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1958 #define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1959 #define isIDCONT_uvchr(c) \
1960 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1961 #define isIDFIRST_uvchr(c) \
1962 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
1963 #define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1964 #define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1966 #define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1967 #define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1968 #define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1970 #define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1971 #define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1972 #define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1973 #define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1975 #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1976 #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1977 #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1978 #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1980 /* For backwards compatibility, even though '_uni' should mean official Unicode
1981 * code points, in Perl it means native for those below 256 */
1982 #define isALPHA_uni(c) isALPHA_uvchr(c)
1983 #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1984 #define isASCII_uni(c) isASCII_uvchr(c)
1985 #define isBLANK_uni(c) isBLANK_uvchr(c)
1986 #define isCNTRL_uni(c) isCNTRL_uvchr(c)
1987 #define isDIGIT_uni(c) isDIGIT_uvchr(c)
1988 #define isGRAPH_uni(c) isGRAPH_uvchr(c)
1989 #define isIDCONT_uni(c) isIDCONT_uvchr(c)
1990 #define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1991 #define isLOWER_uni(c) isLOWER_uvchr(c)
1992 #define isPRINT_uni(c) isPRINT_uvchr(c)
1993 #define isPUNCT_uni(c) isPUNCT_uvchr(c)
1994 #define isSPACE_uni(c) isSPACE_uvchr(c)
1995 #define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1996 #define isUPPER_uni(c) isUPPER_uvchr(c)
1997 #define isVERTWS_uni(c) isVERTWS_uvchr(c)
1998 #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1999 #define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2000 #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2001 #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2002 #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2003 #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
2005 /* For internal core Perl use only: the base macros for defining macros like
2006 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
2007 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
2008 * simple class definitions; the other for more complex. These are like
2009 * _generic_uvchr, so see it for more info. */
2010 #define _generic_LC_uvchr(latin1, above_latin1, c) \
2011 (c < 256 ? latin1(c) : above_latin1(c))
2012 #define _generic_LC_swash_uvchr(latin1, classnum, c) \
2013 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2015 #define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
2016 #define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
2017 _CC_ALPHANUMERIC, c)
2018 #define isASCII_LC_uvchr(c) isASCII_LC(c)
2019 #define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2020 is_HORIZWS_cp_high, c)
2021 #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2022 #define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2023 #define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
2024 #define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
2025 _is_uni_perl_idcont, c)
2026 #define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
2027 _is_uni_perl_idstart, c)
2028 #define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
2029 #define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
2030 #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2031 #define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
2032 #define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
2033 is_XPERLSPACE_cp_high, c)
2034 #define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
2035 #define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
2037 #define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2038 is_XDIGIT_cp_high, c)
2040 #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
2042 /* For internal core Perl use only: the base macros for defining macros like
2043 * isALPHA_utf8. These are like the earlier defined macros, but take an input
2044 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
2045 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
2046 * 'utf8' parameter. This relies on the fact that ASCII characters have the
2047 * same representation whether utf8 or not. Note that it assumes that the utf8
2048 * has been validated, and ignores 'use bytes' */
2049 #define _base_generic_utf8(enum_name, name, p, use_locale ) \
2050 _is_utf8_FOO(CAT2(_CC_, enum_name), \
2052 "is" STRINGIFY(name) "_utf8", \
2053 "is" STRINGIFY(name) "_utf8_safe", \
2054 1, use_locale, __FILE__,__LINE__)
2056 #define _generic_utf8(name, p) _base_generic_utf8(name, name, p, 0)
2058 /* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2059 * they don't otherwise go out of their way to look for malformed UTF-8. If
2060 * they can return accurate results without knowing if the input is otherwise
2061 * malformed, they do so. For example isASCII is accurate in spite of any
2062 * non-length malformations because it looks only at a single byte. Likewise
2063 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2064 * variant ones return FALSE. But, if the input has to be well-formed in order
2065 * for the results to be accurate, the macros will test and if malformed will
2066 * call a routine to die
2068 * Except for toke.c, the macros do assume that e > p, asserting that on
2069 * DEBUGGING builds. Much code that calls these depends on this being true,
2070 * for other reasons. toke.c is treated specially as using the regular
2071 * assertion breaks it in many ways. All strings that these operate on there
2072 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2073 * bunch of code in toke.c assumes that this is true, so the assertion allows
2075 #ifdef PERL_IN_TOKE_C
2076 # define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2078 # define _utf8_safe_assert(p,e) ((e) > (p))
2081 #define _generic_utf8_safe(classnum, p, e, above_latin1) \
2082 (__ASSERT_(_utf8_safe_assert(p, e)) \
2083 (UTF8_IS_INVARIANT(*(p))) \
2084 ? _generic_isCC(*(p), classnum) \
2085 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2086 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2087 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2089 : (_force_out_malformed_utf8_message( \
2090 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2092 /* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2093 * 'above_latin1' can be a macro */
2094 #define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2095 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2096 #define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \
2097 _generic_utf8_safe(classnum, p, e, \
2098 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2099 ? (_force_out_malformed_utf8_message( \
2100 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2102 /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2103 * 'above_latin1' parameter */
2104 #define _generic_swash_utf8_safe(classnum, p, e) \
2105 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e))
2107 /* Like the above, but should be used only when it is known that there are no
2108 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2109 * class is TRUE for. Hence it can skip the tests for this range.
2110 * 'above_latin1' should include its arguments */
2111 #define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2112 (__ASSERT_(_utf8_safe_assert(p, e)) \
2113 (UTF8_IS_INVARIANT(*(p))) \
2114 ? _generic_isCC(*(p), classnum) \
2115 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2116 ? 0 /* Note that doesn't check validity for latin1 */ \
2120 #define isALPHA_utf8(p) _generic_utf8(ALPHA, p)
2121 #define isALPHANUMERIC_utf8(p) _generic_utf8(ALPHANUMERIC, p)
2122 #define isASCII_utf8(p) _generic_utf8(ASCII, p)
2123 #define isBLANK_utf8(p) _generic_utf8(BLANK, p)
2124 #define isCNTRL_utf8(p) _generic_utf8(CNTRL, p)
2125 #define isDIGIT_utf8(p) _generic_utf8(DIGIT, p)
2126 #define isGRAPH_utf8(p) _generic_utf8(GRAPH, p)
2127 #define isIDCONT_utf8(p) _generic_utf8(IDCONT, p)
2128 #define isIDFIRST_utf8(p) _generic_utf8(IDFIRST, p)
2129 #define isLOWER_utf8(p) _generic_utf8(LOWER, p)
2130 #define isPRINT_utf8(p) _generic_utf8(PRINT, p)
2131 #define isPSXSPC_utf8(p) _generic_utf8(PSXSPC, p)
2132 #define isPUNCT_utf8(p) _generic_utf8(PUNCT, p)
2133 #define isSPACE_utf8(p) _generic_utf8(SPACE, p)
2134 #define isUPPER_utf8(p) _generic_utf8(UPPER, p)
2135 #define isVERTWS_utf8(p) _generic_utf8(VERTSPACE, p)
2136 #define isWORDCHAR_utf8(p) _generic_utf8(WORDCHAR, p)
2137 #define isXDIGIT_utf8(p) _generic_utf8(XDIGIT, p)
2139 #define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e)
2140 #define isALPHANUMERIC_utf8_safe(p, e) \
2141 _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e)
2142 #define isASCII_utf8_safe(p, e) \
2143 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2145 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2146 #define isBLANK_utf8_safe(p, e) \
2147 _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
2150 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2151 * more efficient macro instead of the more general one */
2152 # define isCNTRL_utf8_safe(p, e) \
2153 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
2155 # define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
2158 #define isDIGIT_utf8_safe(p, e) \
2159 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2160 _is_utf8_FOO_with_len(_CC_DIGIT, p, e))
2161 #define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e)
2162 #define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
2163 _is_utf8_perl_idcont_with_len, p, e)
2165 /* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2166 * IDFIRST is an alnum. See
2167 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
2168 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2169 * This used to be not the XID version, but we decided to go with the more
2170 * modern Unicode definition */
2171 #define isIDFIRST_utf8_safe(p, e) \
2172 _generic_func_utf8_safe(_CC_IDFIRST, \
2173 _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e))
2175 #define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e)
2176 #define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e)
2177 #define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2178 #define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e)
2179 #define isSPACE_utf8_safe(p, e) \
2180 _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2181 #define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e)
2182 #define isVERTWS_utf8_safe(p, e) \
2183 _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
2184 #define isWORDCHAR_utf8_safe(p, e) \
2185 _generic_swash_utf8_safe(_CC_WORDCHAR, p, e)
2186 #define isXDIGIT_utf8_safe(p, e) \
2187 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2188 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2189 ? (_force_out_malformed_utf8_message( \
2190 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2191 : is_XDIGIT_high(p)))
2193 #define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
2194 #define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
2195 #define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
2196 #define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2198 /* For internal core use only, subject to change */
2199 #define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f, "", 0)
2200 #define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f, "", 0)
2201 #define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f, "", 0)
2202 #define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f, "", 0)
2204 #define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2205 #define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2206 #define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2207 #define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
2209 /* For internal core Perl use only: the base macros for defining macros like
2210 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
2211 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
2212 * 'macro' parameter */
2213 #define _generic_LC_utf8(name, p) _base_generic_utf8(name, name, p, 1)
2215 #define isALPHA_LC_utf8(p) _generic_LC_utf8(ALPHA, p)
2216 #define isALPHANUMERIC_LC_utf8(p) _generic_LC_utf8(ALPHANUMERIC, p)
2217 #define isASCII_LC_utf8(p) _generic_LC_utf8(ASCII, p)
2218 #define isBLANK_LC_utf8(p) _generic_LC_utf8(BLANK, p)
2219 #define isCNTRL_LC_utf8(p) _generic_LC_utf8(CNTRL, p)
2220 #define isDIGIT_LC_utf8(p) _generic_LC_utf8(DIGIT, p)
2221 #define isGRAPH_LC_utf8(p) _generic_LC_utf8(GRAPH, p)
2222 #define isIDCONT_LC_utf8(p) _generic_LC_utf8(IDCONT, p)
2223 #define isIDFIRST_LC_utf8(p) _generic_LC_utf8(IDFIRST, p)
2224 #define isLOWER_LC_utf8(p) _generic_LC_utf8(LOWER, p)
2225 #define isPRINT_LC_utf8(p) _generic_LC_utf8(PRINT, p)
2226 #define isPSXSPC_LC_utf8(p) _generic_LC_utf8(PSXSPC, p)
2227 #define isPUNCT_LC_utf8(p) _generic_LC_utf8(PUNCT, p)
2228 #define isSPACE_LC_utf8(p) _generic_LC_utf8(SPACE, p)
2229 #define isUPPER_LC_utf8(p) _generic_LC_utf8(UPPER, p)
2230 #define isWORDCHAR_LC_utf8(p) _generic_LC_utf8(WORDCHAR, p)
2231 #define isXDIGIT_LC_utf8(p) _generic_LC_utf8(XDIGIT, p)
2233 /* For internal core Perl use only: the base macros for defining macros like
2234 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2235 * point in 'p' is within the 0-255 range, it uses locale rules from the
2236 * passed-in 'macro' parameter */
2237 #define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2238 (__ASSERT_(_utf8_safe_assert(p, e)) \
2239 (UTF8_IS_INVARIANT(*(p))) \
2241 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2242 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2243 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2244 : (_force_out_malformed_utf8_message( \
2245 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2248 #define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \
2249 _generic_LC_utf8_safe(macro, p, e, \
2250 _is_utf8_FOO_with_len(classnum, p, e))
2252 #define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2253 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2255 #define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \
2256 _generic_LC_utf8_safe(classnum, p, e, \
2257 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2258 ? (_force_out_malformed_utf8_message( \
2259 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2262 #define isALPHANUMERIC_LC_utf8_safe(p, e) \
2263 _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \
2264 _CC_ALPHANUMERIC, p, e)
2265 #define isALPHA_LC_utf8_safe(p, e) \
2266 _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2267 #define isASCII_LC_utf8_safe(p, e) \
2268 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2269 #define isBLANK_LC_utf8_safe(p, e) \
2270 _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2271 #define isCNTRL_LC_utf8_safe(p, e) \
2272 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2273 #define isDIGIT_LC_utf8_safe(p, e) \
2274 _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2275 #define isGRAPH_LC_utf8_safe(p, e) \
2276 _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2277 #define isIDCONT_LC_utf8_safe(p, e) \
2278 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2279 _is_utf8_perl_idcont_with_len, p, e)
2280 #define isIDFIRST_LC_utf8_safe(p, e) \
2281 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2282 _is_utf8_perl_idstart_with_len, p, e)
2283 #define isLOWER_LC_utf8_safe(p, e) \
2284 _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2285 #define isPRINT_LC_utf8_safe(p, e) \
2286 _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2287 #define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2288 #define isPUNCT_LC_utf8_safe(p, e) \
2289 _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2290 #define isSPACE_LC_utf8_safe(p, e) \
2291 _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2292 #define isUPPER_LC_utf8_safe(p, e) \
2293 _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2294 #define isWORDCHAR_LC_utf8_safe(p, e) \
2295 _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2296 #define isXDIGIT_LC_utf8_safe(p, e) \
2297 _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
2299 /* Macros for backwards compatibility and for completeness when the ASCII and
2300 * Latin1 values are identical */
2301 #define isALPHAU(c) isALPHA_L1(c)
2302 #define isDIGIT_L1(c) isDIGIT_A(c)
2303 #define isOCTAL(c) isOCTAL_A(c)
2304 #define isOCTAL_L1(c) isOCTAL_A(c)
2305 #define isXDIGIT_L1(c) isXDIGIT_A(c)
2306 #define isALNUM(c) isWORDCHAR(c)
2307 #define isALNUM_A(c) isALNUM(c)
2308 #define isALNUMU(c) isWORDCHAR_L1(c)
2309 #define isALNUM_LC(c) isWORDCHAR_LC(c)
2310 #define isALNUM_uni(c) isWORDCHAR_uni(c)
2311 #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
2312 #define isALNUM_utf8(p) isWORDCHAR_utf8(p)
2313 #define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
2314 #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2315 #define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2316 #define isALNUMC(c) isALPHANUMERIC(c)
2317 #define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2318 #define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
2319 #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
2320 #define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
2321 #define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
2323 /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2324 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2325 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2326 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2327 * the outlier from the block that contains the other controls, just like
2328 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2329 * block. If it weren't special cased, it would yield a non-control.
2330 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2333 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2335 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2337 ? (UNLIKELY((c) == '?') \
2338 ? QUESTION_MARK_CTRL \
2339 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2340 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2342 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2345 /* Line numbers are unsigned, 32 bits. */
2347 #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
2349 /* Helpful alias for version prescan */
2350 #define is_LAX_VERSION(a,b) \
2351 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2353 #define is_STRICT_VERSION(a,b) \
2354 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2356 #define BADVERSION(a,b,c) \
2362 /* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2363 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2364 * which is advanced. The input is validated only by an assert() in DEBUGGING
2365 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2366 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2367 #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2370 #define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2374 /* Converts a character known to represent an octal digit (0-7) to its numeric
2375 * value. The input is validated only by an assert() in DEBUGGING builds. In
2376 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2377 #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2379 /* Efficiently returns a boolean as to if two native characters are equivalent
2380 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2381 * the ALPHA in the name is to remind you of that. This is asserted() in
2382 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2383 * works (on valid input) for both non- and UTF-8-encoded bytes.
2385 * When one of the inputs is a compile-time constant and gets folded by the
2386 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2387 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2388 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
2389 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2390 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2391 * just a single 0, in the bit position where the upper- and lowercase differ.
2393 #define isALPHA_FOLD_EQ(c1, c2) \
2394 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2395 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2396 #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2399 =head1 Memory Management
2401 =for apidoc Am|void|Newx|void* ptr|int nitems|type
2402 The XSUB-writer's interface to the C C<malloc> function.
2404 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2406 In 5.9.3, Newx() and friends replace the older New() API, and drops
2407 the first parameter, I<x>, a debug aid which allowed callers to identify
2408 themselves. This aid has been superseded by a new build option,
2409 PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
2410 there for use in XS modules supporting older perls.
2412 =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
2413 The XSUB-writer's interface to the C C<malloc> function, with
2414 cast. See also C<L</Newx>>.
2416 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2418 =for apidoc Am|void|Newxz|void* ptr|int nitems|type
2419 The XSUB-writer's interface to the C C<malloc> function. The allocated
2420 memory is zeroed with C<memzero>. See also C<L</Newx>>.
2422 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2424 =for apidoc Am|void|Renew|void* ptr|int nitems|type
2425 The XSUB-writer's interface to the C C<realloc> function.
2427 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2429 =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2430 The XSUB-writer's interface to the C C<realloc> function, with
2433 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2435 =for apidoc Am|void|Safefree|void* ptr
2436 The XSUB-writer's interface to the C C<free> function.
2438 This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
2440 =for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2441 The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
2442 source, C<dest> is the destination, C<nitems> is the number of items, and
2443 C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
2445 =for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
2446 Like C<Move> but returns C<dest>. Useful
2447 for encouraging compilers to tail-call
2450 =for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2451 The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
2452 source, C<dest> is the destination, C<nitems> is the number of items, and
2453 C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
2455 =for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2457 Like C<Copy> but returns C<dest>. Useful
2458 for encouraging compilers to tail-call
2461 =for apidoc Am|void|Zero|void* dest|int nitems|type
2463 The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2464 destination, C<nitems> is the number of items, and C<type> is the type.
2466 =for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2468 Like C<Zero> but returns dest. Useful
2469 for encouraging compilers to tail-call
2472 =for apidoc Am|void|StructCopy|type *src|type *dest|type
2473 This is an architecture-independent macro to copy one structure to another.
2475 =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2477 Fill up memory with a byte pattern (a byte repeated over and over
2478 again) that hopefully catches attempts to access uninitialized memory.
2480 =for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2482 PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2484 =for apidoc Am|void|PoisonFree|void* dest|int nitems|type
2486 PoisonWith(0xEF) for catching access to freed memory.
2488 =for apidoc Am|void|Poison|void* dest|int nitems|type
2490 PoisonWith(0xEF) for catching access to freed memory.
2494 /* Maintained for backwards-compatibility only. Use newSV() instead. */
2496 #define NEWSV(x,len) newSV(len)
2499 #define MEM_SIZE_MAX ((MEM_SIZE)-1)
2501 #define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
2503 #ifdef PERL_MALLOC_WRAP
2505 /* This expression will be constant-folded at compile time. It checks
2506 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2507 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2508 * As well as avoiding the need for a run-time check in some cases, it's
2509 * designed to avoid compiler warnings like:
2510 * comparison is always false due to limited range of data type
2511 * It's mathematically equivalent to
2512 * max(n) * sizeof(t) > MEM_SIZE_MAX
2515 # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
2516 ( sizeof(MEM_SIZE) < sizeof(n) \
2517 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
2519 /* This is written in a slightly odd way to avoid various spurious
2520 * compiler warnings. We *want* to write the expression as
2521 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2522 * (for some compile-time constant C), but even when the LHS
2523 * constant-folds to false at compile-time, g++ insists on emitting
2524 * warnings about the RHS (e.g. "comparison is always false"), so instead
2527 * (cond ? n : X) > C
2529 * where X is a constant with X > C always false. Choosing a value for X
2530 * is tricky. If 0, some compilers will complain about 0 > C always being
2531 * false; if 1, Coverity complains when n happens to be the constant value
2532 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2533 * for X and hope that nothing else whines.
2536 # define _MEM_WRAP_WILL_WRAP(n,t) \
2537 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2538 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
2540 # define MEM_WRAP_CHECK(n,t) \
2541 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2542 && (croak_memory_wrap(),0))
2544 # define MEM_WRAP_CHECK_1(n,t,a) \
2545 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2546 && (Perl_croak_nocontext("%s",(a)),0))
2548 /* "a" arg must be a string literal */
2549 # define MEM_WRAP_CHECK_s(n,t,a) \
2550 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2551 && (Perl_croak_nocontext("" a ""),0))
2553 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
2555 #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))
2558 #define MEM_WRAP_CHECK(n,t)
2559 #define MEM_WRAP_CHECK_1(n,t,a)
2560 #define MEM_WRAP_CHECK_s(n,t,a)
2561 #define MEM_WRAP_CHECK_(n,t)
2563 #define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
2569 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
2570 * go through functions, which are handy for debugging breakpoints, but
2571 * which more importantly get the immediate calling environment (file and
2572 * line number, and C function name if available) passed in. This info can
2573 * then be used for logging the calls, for which one gets a sample
2574 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
2577 * - not all memory allocs get logged, only those
2578 * that go through Newx() and derivatives (while all
2579 * Safefrees do get logged)
2580 * - __FILE__ and __LINE__ do not work everywhere
2581 * - __func__ or __FUNCTION__ even less so
2582 * - I think more goes on after the perlio frees but
2583 * the thing is that STDERR gets closed (as do all
2584 * the file descriptors)
2585 * - no deeper calling stack than the caller of the Newx()
2586 * or the kind, but do I look like a C reflection/introspection
2588 * - the function prototypes for the logging functions
2589 * probably should maybe be somewhere else than handy.h
2590 * - one could consider inlining (macrofying) the logging
2591 * for speed, but I am too lazy
2592 * - one could imagine recording the allocations in a hash,
2593 * (keyed by the allocation address?), and maintain that
2594 * through reallocs and frees, but how to do that without
2595 * any News() happening...?
2596 * - lots of -Ddefines to get useful/controllable output
2597 * - lots of ENV reads
2601 # ifndef PERL_MEM_LOG_NOIMPL
2610 # if defined(PERL_IN_SV_C) /* those are only used in sv.c */
2611 void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2612 void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2619 #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2620 #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
2621 #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
2624 #ifndef MEM_LOG_ALLOC
2625 #define MEM_LOG_ALLOC(n,t,a) (a)
2627 #ifndef MEM_LOG_REALLOC
2628 #define MEM_LOG_REALLOC(n,t,v,a) (a)
2630 #ifndef MEM_LOG_FREE
2631 #define MEM_LOG_FREE(a) (a)
2634 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2635 #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2636 #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
2639 /* pre 5.9.x compatibility */
2640 #define New(x,v,n,t) Newx(v,n,t)
2641 #define Newc(x,v,n,t,c) Newxc(v,n,t,c)
2642 #define Newz(x,v,n,t) Newxz(v,n,t)
2645 #define Renew(v,n,t) \
2646 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2647 #define Renewc(v,n,t,c) \
2648 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2651 #define Safefree(d) \
2652 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
2654 #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
2657 /* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2658 * as it handles cases like constant string arguments without throwing warnings *
2659 * the cast is required, as is the inequality check, to avoid warnings */
2660 #define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
2663 #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)))
2664 #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)))
2665 #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2667 /* Like above, but returns a pointer to 'd' */
2668 #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)))
2669 #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)))
2670 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
2672 #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2673 #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2674 #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2675 #define Poison(d,n,t) PoisonFree(d,n,t)
2678 # define PERL_POISON_EXPR(x) x
2680 # define PERL_POISON_EXPR(x)
2683 #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2688 =for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2690 Returns the number of elements in the input C array (so you want your
2691 zero-based indices to be less than but not equal to).
2693 =for apidoc Am|void *|C_ARRAY_END|void *a
2695 Returns a pointer to one element past the final element of the input C array.
2699 C_ARRAY_END is one past the last: half-open/half-closed range, not
2700 last-inclusive range.
2702 #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
2703 #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
2707 # define Perl_va_copy(s, d) va_copy(d, s)
2708 # elif defined(__va_copy)
2709 # define Perl_va_copy(s, d) __va_copy(d, s)
2711 # define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2715 /* convenience debug macros */
2717 #define pTHX_FORMAT "Perl interpreter: 0x%p"
2718 #define pTHX__FORMAT ", Perl interpreter: 0x%p"
2719 #define pTHX_VALUE_ (void *)my_perl,
2720 #define pTHX_VALUE (void *)my_perl
2721 #define pTHX__VALUE_ ,(void *)my_perl,
2722 #define pTHX__VALUE ,(void *)my_perl
2725 #define pTHX__FORMAT
2728 #define pTHX__VALUE_
2730 #endif /* USE_ITHREADS */
2732 /* Perl_deprecate was not part of the public API, and did not have a deprecate()
2733 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2734 CPAN::Unpack show any users outside the core. */
2736 # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2737 "Use of " s " is deprecated")
2738 # define deprecate_disappears_in(when,message) \
2739 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2740 message ", and will disappear in Perl " when)
2741 # define deprecate_fatal_in(when,message) \
2742 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2743 message ". Its use will be fatal in Perl " when)
2746 /* Internal macros to deal with gids and uids */
2749 # if Uid_t_size > IVSIZE
2750 # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2751 # define SvUID(sv) SvNV(sv)
2752 # elif Uid_t_sign <= 0
2753 # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2754 # define SvUID(sv) SvIV(sv)
2756 # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2757 # define SvUID(sv) SvUV(sv)
2758 # endif /* Uid_t_size */
2760 # if Gid_t_size > IVSIZE
2761 # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2762 # define SvGID(sv) SvNV(sv)
2763 # elif Gid_t_sign <= 0
2764 # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2765 # define SvGID(sv) SvIV(sv)
2767 # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2768 # define SvGID(sv) SvUV(sv)
2769 # endif /* Gid_t_size */
2773 #endif /* PERL_HANDY_H_ */
2776 * ex: set ts=8 sts=4 sw=4 et: