3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
11 /* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
14 #ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */
18 # define Null(type) ((type)NULL)
23 =for apidoc AmU||Nullch
24 Null character pointer. (No longer available when C<PERL_CORE> is
27 =for apidoc AmU||Nullsv
28 Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
33 # define Nullch Null(char*)
34 # define Nullfp Null(PerlIO*)
35 # define Nullsv Null(SV*)
47 /* The MUTABLE_*() macros cast pointers to the types shown, in such a way
48 * (compiler permitting) that casting away const-ness will give a warning;
52 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
53 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
56 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
57 # define MUTABLE_PTR(p) ({ void *_p = (p); _p; })
59 # define MUTABLE_PTR(p) ((void *) (p))
62 #define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
63 #define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
64 #define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
65 #define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
66 #define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
67 #define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
69 #if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
76 /* bool is built-in for g++-2.6.3 and later, which might be used
77 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
78 be sure _G_config.h will be included before this file. _G_config.h
79 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
80 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
81 g++ can be identified by __GNUG__.
82 Andy Dougherty February 2000
84 #ifdef __GNUG__ /* GNU g++ has bool built-in */
85 # ifndef PERL_BOOL_AS_CHAR
100 /* cast-to-bool. A simple (bool) cast may not do the right thing: if bool is
101 * defined as char for example, then the cast from int is
102 * implementation-defined (bool)!!(cbool) in a ternary triggers a bug in xlc on
104 #define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
106 /* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
107 * XXX Should really be a Configure probe, with HAS__FUNCTION__
108 * and FUNCTION__ as results.
109 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
110 #if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
111 # define FUNCTION__ __func__
112 #elif (defined(USING_MSVC6)) || /* MSVC6 has neither __func__ nor __FUNCTION and no good workarounds, either. */ \
113 (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
114 # define FUNCTION__ ""
116 # define FUNCTION__ __FUNCTION__ /* Common extension. */
119 /* XXX A note on the perl source internal type system. The
120 original intent was that I32 be *exactly* 32 bits.
122 Currently, we only guarantee that I32 is *at least* 32 bits.
123 Specifically, if int is 64 bits, then so is I32. (This is the case
124 for the Cray.) This has the advantage of meshing nicely with
125 standard library calls (where we pass an I32 and the library is
126 expecting an int), but the disadvantage that an I32 is not 32 bits.
127 Andy Dougherty August 1996
129 There is no guarantee that there is *any* integral type with
130 exactly 32 bits. It is perfectly legal for a system to have
131 sizeof(short) == sizeof(int) == sizeof(long) == 8.
133 Similarly, there is no guarantee that I16 and U16 have exactly 16
136 For dealing with issues that may arise from various 32/64-bit
137 systems, we will ask Configure to check out
139 SHORTSIZE == sizeof(short)
140 INTSIZE == sizeof(int)
141 LONGSIZE == sizeof(long)
142 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
143 PTRSIZE == sizeof(void *)
144 DOUBLESIZE == sizeof(double)
145 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
149 #ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
150 # include <inttypes.h>
151 # ifdef INT32_MIN_BROKEN
153 # define INT32_MIN (-2147483647-1)
155 # ifdef INT64_MIN_BROKEN
157 # define INT64_MIN (-9223372036854775807LL-1)
173 /* INT64_C/UINT64_C are C99 from <stdint.h> (so they will not be
174 * available in strict C89 mode), but they are nice, so let's define
175 * them if necessary. */
177 /* N.B. We would like to say HAS_QUAD here, but that doesn't actually mean what
178 * it has always been documented to mean (see RT #119753) and is explicitly turned
179 * off in perl.h outside of core with dire warnings about removing the undef.
182 #if defined(QUADKIND)
184 # undef PeRl_UINT64_C
185 /* Prefer the native integer types (int and long) over long long
186 * (which is not C89) and Win32-specific __int64. */
187 # if QUADKIND == QUAD_IS_INT && INTSIZE == 8
188 # define PeRl_INT64_C(c) (c)
189 # define PeRl_UINT64_C(c) CAT2(c,U)
191 # if QUADKIND == QUAD_IS_LONG && LONGSIZE == 8
192 # define PeRl_INT64_C(c) CAT2(c,L)
193 # define PeRl_UINT64_C(c) CAT2(c,UL)
195 # if QUADKIND == QUAD_IS_LONG_LONG && defined(HAS_LONG_LONG)
196 # define PeRl_INT64_C(c) CAT2(c,LL)
197 # define PeRl_UINT64_C(c) CAT2(c,ULL)
199 # if QUADKIND == QUAD_IS___INT64
200 # define PeRl_INT64_C(c) CAT2(c,I64)
201 # define PeRl_UINT64_C(c) CAT2(c,UI64)
203 # ifndef PeRl_INT64_C
204 # define PeRl_INT64_C(c) ((I64)(c)) /* last resort */
205 # define PeRl_UINT64_C(c) ((U64)(c))
207 /* In OS X the INT64_C/UINT64_C are defined with LL/ULL, which will
208 * not fly with C89-pedantic gcc, so let's undefine them first so that
209 * we can redefine them with our native integer preferring versions. */
210 # if defined(PERL_DARWIN) && defined(PERL_GCC_PEDANTIC)
215 # define INT64_C(c) PeRl_INT64_C(c)
218 # define UINT64_C(c) PeRl_UINT64_C(c)
222 #if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
224 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
225 Please search CHAR_MAX in perl.h for further details. */
226 #define U8_MAX UINT8_MAX
227 #define U8_MIN UINT8_MIN
229 #define I16_MAX INT16_MAX
230 #define I16_MIN INT16_MIN
231 #define U16_MAX UINT16_MAX
232 #define U16_MIN UINT16_MIN
234 #define I32_MAX INT32_MAX
235 #define I32_MIN INT32_MIN
236 #ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
237 # define U32_MAX UINT32_MAX
239 # define U32_MAX 4294967295U
241 #define U32_MIN UINT32_MIN
245 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
246 Please search CHAR_MAX in perl.h for further details. */
247 #define U8_MAX PERL_UCHAR_MAX
248 #define U8_MIN PERL_UCHAR_MIN
250 #define I16_MAX PERL_SHORT_MAX
251 #define I16_MIN PERL_SHORT_MIN
252 #define U16_MAX PERL_USHORT_MAX
253 #define U16_MIN PERL_USHORT_MIN
256 # define I32_MAX PERL_INT_MAX
257 # define I32_MIN PERL_INT_MIN
258 # define U32_MAX PERL_UINT_MAX
259 # define U32_MIN PERL_UINT_MIN
261 # define I32_MAX PERL_LONG_MAX
262 # define I32_MIN PERL_LONG_MIN
263 # define U32_MAX PERL_ULONG_MAX
264 # define U32_MIN PERL_ULONG_MIN
269 /* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
270 * anyone is grepping for it */
271 #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
272 #define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
273 #define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
275 /* Unused by core; should be deprecated */
276 #define Ctl(ch) ((ch) & 037)
278 #if defined(PERL_CORE) || defined(PERL_EXT)
280 # define MIN(a,b) ((a) < (b) ? (a) : (b))
283 # define MAX(a,b) ((a) > (b) ? (a) : (b))
287 /* This is a helper macro to avoid preprocessor issues, replaced by nothing
288 * unless under DEBUGGING, where it expands to an assert of its argument,
289 * followed by a comma (hence the comma operator). If we just used a straight
290 * assert(), we would get a comma with nothing before it when not DEBUGGING.
292 * We also use empty definition under Coverity since the __ASSERT__
293 * checks often check for things that Really Cannot Happen, and Coverity
294 * detects that and gets all excited. */
296 #if defined(DEBUGGING) && !defined(__COVERITY__)
297 # define __ASSERT_(statement) assert(statement),
299 # define __ASSERT_(statement)
303 =head1 SV-Body Allocation
305 =for apidoc Ama|SV*|newSVpvs|"literal string" s
306 Like C<newSVpvn>, but takes a literal string instead of a
309 =for apidoc Ama|SV*|newSVpvs_flags|"literal string" s|U32 flags
310 Like C<newSVpvn_flags>, but takes a literal string instead of
311 a string/length pair.
313 =for apidoc Ama|SV*|newSVpvs_share|"literal string" s
314 Like C<newSVpvn_share>, but takes a literal string instead of
315 a string/length pair and omits the hash parameter.
317 =for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string" s|I32 flags
318 Like C<sv_catpvn_flags>, but takes a literal string instead
319 of a string/length pair.
321 =for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string" s
322 Like C<sv_catpvn_nomg>, but takes a literal string instead of
323 a string/length pair.
325 =for apidoc Am|void|sv_catpvs|SV* sv|"literal string" s
326 Like C<sv_catpvn>, but takes a literal string instead of a
329 =for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string" s
330 Like C<sv_catpvn_mg>, but takes a literal string instead of a
333 =for apidoc Am|void|sv_setpvs|SV* sv|"literal string" s
334 Like C<sv_setpvn>, but takes a literal string instead of a
337 =for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string" s
338 Like C<sv_setpvn_mg>, but takes a literal string instead of a
341 =for apidoc Am|SV *|sv_setref_pvs|"literal string" s
342 Like C<sv_setref_pvn>, but takes a literal string instead of
343 a string/length pair.
345 =head1 Memory Management
347 =for apidoc Ama|char*|savepvs|"literal string" s
348 Like C<savepvn>, but takes a literal string instead of a
351 =for apidoc Ama|char*|savesharedpvs|"literal string" s
352 A version of C<savepvs()> which allocates the duplicate string in memory
353 which is shared between threads.
357 =for apidoc Am|HV*|gv_stashpvs|"literal string" name|I32 create
358 Like C<gv_stashpvn>, but takes a literal string instead of a
361 =head1 Hash Manipulation Functions
363 =for apidoc Am|SV**|hv_fetchs|HV* tb|"literal string" key|I32 lval
364 Like C<hv_fetch>, but takes a literal string instead of a
367 =for apidoc Am|SV**|hv_stores|HV* tb|"literal string" key|SV* val
368 Like C<hv_store>, but takes a literal string instead of a
370 and omits the hash parameter.
372 =head1 Lexer interface
374 =for apidoc Amx|void|lex_stuff_pvs|"literal string" pv|U32 flags
376 Like L</lex_stuff_pvn>, but takes a literal string instead of
377 a string/length pair.
382 /* concatenating with "" ensures that only literal strings are accepted as
384 #define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
386 /* note that STR_WITH_LEN() can't be used as argument to macros or functions
387 * that under some configurations might be macros, which means that it requires
388 * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used.
391 /* STR_WITH_LEN() shortcuts */
392 #define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
393 #define newSVpvs_flags(str,flags) \
394 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
395 #define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
396 #define sv_catpvs_flags(sv, str, flags) \
397 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
398 #define sv_catpvs_nomg(sv, str) \
399 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
400 #define sv_catpvs(sv, str) \
401 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
402 #define sv_catpvs_mg(sv, str) \
403 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
404 #define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
405 #define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
406 #define sv_setref_pvs(rv, classname, str) \
407 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
408 #define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
409 #define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
410 #define gv_stashpvs(str, create) \
411 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
412 #define gv_fetchpvs(namebeg, add, sv_type) \
413 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
414 #define gv_fetchpvn(namebeg, len, add, sv_type) \
415 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
416 #define sv_catxmlpvs(dsv, str, utf8) \
417 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
420 #define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
422 #define get_cvs(str, flags) \
423 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
426 =head1 Miscellaneous Functions
428 =for apidoc Am|bool|strNE|char* s1|char* s2
429 Test two C<NUL>-terminated strings to see if they are different. Returns true
432 =for apidoc Am|bool|strEQ|char* s1|char* s2
433 Test two C<NUL>-terminated strings to see if they are equal. Returns true or
436 =for apidoc Am|bool|strLT|char* s1|char* s2
437 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
438 second, C<s2>. Returns true or false.
440 =for apidoc Am|bool|strLE|char* s1|char* s2
441 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
442 equal to the second, C<s2>. Returns true or false.
444 =for apidoc Am|bool|strGT|char* s1|char* s2
445 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
446 the second, C<s2>. Returns true or false.
448 =for apidoc Am|bool|strGE|char* s1|char* s2
449 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
450 or equal to the second, C<s2>. Returns true or false.
452 =for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
453 Test two C<NUL>-terminated strings to see if they are different. The C<len>
454 parameter indicates the number of bytes to compare. Returns true or false. (A
455 wrapper for C<strncmp>).
457 =for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
458 Test two C<NUL>-terminated strings to see if they are equal. The C<len>
459 parameter indicates the number of bytes to compare. Returns true or false. (A
460 wrapper for C<strncmp>).
462 =for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
463 Test two buffers (which may contain embedded C<NUL> characters, to see if they
464 are equal. The C<len> parameter indicates the number of bytes to compare.
465 Returns zero if equal, or non-zero if non-equal.
467 =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
468 Test two buffers (which may contain embedded C<NUL> characters, to see if they
469 are not equal. The C<len> parameter indicates the number of bytes to compare.
470 Returns zero if non-equal, or non-zero if equal.
474 New macros should use the following conventions for their names (which are
475 based on the underlying C library functions):
477 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
479 Each has two main parameters, string-like operands that are compared
480 against each other, as specified by the macro name. Some macros may
481 additionally have one or potentially even two length parameters. If a length
482 parameter applies to both string parameters, it will be positioned third;
483 otherwise any length parameter immediately follows the string parameter it
486 If the prefix to the name is 'str', the string parameter is a pointer to a C
487 language string. Such a string does not contain embedded NUL bytes; its
488 length may be unknown, but can be calculated by C<strlen()>, since it is
489 terminated by a NUL, which isn't included in its length.
491 The optional 'n' following 'str' means that that there is a third parameter,
492 giving the maximum number of bytes to look at in each string. Even if both
493 strings are longer than the length parameter, those extra bytes will be
496 The 's' suffix means that the 2nd byte string parameter is a literal C
497 double-quoted string. Its length will automatically be calculated by the
498 macro, so no length parameter will ever be needed for it.
500 If the prefix is 'mem', the string parameters don't have to be C strings;
501 they may contain embedded NUL bytes, do not necessarily have a terminating
502 NUL, and their lengths can be known only through other means, which in
503 practice are additional parameter(s) passed to the function. All 'mem'
504 functions have at least one length parameter. Barring any 'l' or 's' suffix,
505 there is a single length parameter, in position 3, which applies to both
506 string parameters. The 's' suffix means, as described above, that the 2nd
507 string is a literal double-quoted C string (hence its length is calculated by
508 the macro, and the length parameter to the function applies just to the first
509 string parameter, and hence is positioned just after it). An 'l' suffix
510 means that the 2nd string parameter has its own length parameter, and the
511 signature will look like memFOOl(s1, l1, s2, l2).
513 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
514 substring of the 1st string. 'P' if present indicates that the substring
515 must be a "proper" one in tha mathematical sense that the first one must be
516 strictly larger than the 2nd.
521 #define strNE(s1,s2) (strcmp(s1,s2) != 0)
522 #define strEQ(s1,s2) (strcmp(s1,s2) == 0)
523 #define strLT(s1,s2) (strcmp(s1,s2) < 0)
524 #define strLE(s1,s2) (strcmp(s1,s2) <= 0)
525 #define strGT(s1,s2) (strcmp(s1,s2) > 0)
526 #define strGE(s1,s2) (strcmp(s1,s2) >= 0)
528 #define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
529 #define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
531 #define memNE(s1,s2,l) (memcmp(s1,s2,l) != 0)
532 #define memEQ(s1,s2,l) (memcmp(s1,s2,l) == 0)
534 /* memEQ and memNE where second comparand is a string constant */
535 #define memEQs(s1, l, s2) \
536 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
537 #define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
539 /* Keep these private until we decide it was a good idea */
540 #if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
542 #define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
544 #define memBEGINs(s1, l, s2) \
545 ( (l) >= sizeof(s2) - 1 \
546 && memEQ(s1, "" s2 "", sizeof(s2)-1))
547 #define memBEGINPs(s1, l, s2) \
548 ( (l) > sizeof(s2) - 1 \
549 && memEQ(s1, "" s2 "", sizeof(s2)-1))
550 #define memENDs(s1, l, s2) \
551 ( (l) >= sizeof(s2) - 1 \
552 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
553 #define memENDPs(s1, l, s2) \
555 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
556 #endif /* End of making macros private */
558 #define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
559 #define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
560 #define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
561 #define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
566 * Unfortunately, the introduction of locales means that we
567 * can't trust isupper(), etc. to tell the truth. And when
568 * it comes to /\w+/ with tainting enabled, we *must* be able
569 * to trust our character classes.
571 * Therefore, the default tests in the text of Perl will be
572 * independent of locale. Any code that wants to depend on
573 * the current locale will use the tests that begin with "lc".
576 #ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
584 =head1 Character classification
585 This section is about functions (really macros) that classify characters
586 into types, such as punctuation versus alphabetic, etc. Most of these are
587 analogous to regular expression character classes. (See
588 L<perlrecharclass/POSIX Character Classes>.) There are several variants for
589 each class. (Not all macros have all variants; each item below lists the
590 ones valid for it.) None are affected by C<use bytes>, and only the ones
591 with C<LC> in the name are affected by the current locale.
593 The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a
594 C<U8>) as input and returns a boolean as to whether or not the character
595 represented by that octet is (or on non-ASCII platforms, corresponds to) an
596 ASCII character in the named class based on platform, Unicode, and Perl rules.
597 If the input is a number that doesn't fit in an octet, FALSE is returned.
599 Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
600 with no suffix C<"_A">. This variant is used to emphasize by its name that
601 only ASCII-range characters can return TRUE.
603 Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
604 onto the platform. That is, the code points that are ASCII are unaffected,
605 since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
606 as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
607 true when called with the code point 0xDF, which is a word character in both
608 ASCII and EBCDIC (though it represents different characters in each).
610 Variant C<isI<FOO>_uvchr> is like the C<isI<FOO>_L1> variant, but accepts any UV code
611 point as input. If the code point is larger than 255, Unicode rules are used
612 to determine if it is in the character class. For example,
613 C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
614 WITH MACRON in Unicode, and is a word character.
616 Variant C<isI<FOO>_utf8_safe> is like C<isI<FOO>_uvchr>, but is used for UTF-8
617 encoded strings. Each call classifies one character, even if the string
618 contains many. This variant takes two parameters. The first, C<p>, is a
619 pointer to the first byte of the character to be classified. (Recall that it
620 may take more than one byte to represent a character in UTF-8 strings.) The
621 second parameter, C<e>, points to anywhere in the string beyond the first
622 character, up to one byte past the end of the entire string. The suffix
623 C<_safe> in the function's name indicates that it will not attempt to read
624 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
625 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
626 character is malformed in some way, the program may croak, or the function may
627 return FALSE, at the discretion of the implementation, and subject to change in
630 Variant C<isI<FOO>_utf8> is like C<isI<FOO>_utf8_safe>, but takes just a single
631 parameter, C<p>, which has the same meaning as the corresponding parameter does
632 in C<isI<FOO>_utf8_safe>. The function therefore can't check if it is reading
633 beyond the end of the string. Starting in Perl v5.30, it will take a second
634 parameter, becoming a synonym for C<isI<FOO>_utf8_safe>. At that time every
635 program that uses it will have to be changed to successfully compile. In the
636 meantime, the first runtime call to C<isI<FOO>_utf8> from each call point in the
637 program will raise a deprecation warning, enabled by default. You can convert
638 your program now to use C<isI<FOO>_utf8_safe>, and avoid the warnings, and get an
639 extra measure of protection, or you can wait until v5.30, when you'll be forced
640 to add the C<e> parameter.
642 Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants, but the
643 result is based on the current locale, which is what C<LC> in the name stands
644 for. If Perl can determine that the current locale is a UTF-8 locale, it uses
645 the published Unicode rules; otherwise, it uses the C library function that
646 gives the named classification. For example, C<isDIGIT_LC()> when not in a
647 UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
648 returned if the input won't fit into an octet. On some platforms where the C
649 library function is known to be defective, Perl changes its result to follow
650 the POSIX standard's rules.
652 Variant C<isI<FOO>_LC_uvchr> is like C<isI<FOO>_LC>, but is defined on any UV. It
653 returns the same as C<isI<FOO>_LC> for input code points less than 256, and
654 returns the hard-coded, not-affected-by-locale, Unicode results for larger ones.
656 Variant C<isI<FOO>_LC_utf8_safe> is like C<isI<FOO>_LC_uvchr>, but is used for UTF-8
657 encoded strings. Each call classifies one character, even if the string
658 contains many. This variant takes two parameters. The first, C<p>, is a
659 pointer to the first byte of the character to be classified. (Recall that it
660 may take more than one byte to represent a character in UTF-8 strings.) The
661 second parameter, C<e>, points to anywhere in the string beyond the first
662 character, up to one byte past the end of the entire string. The suffix
663 C<_safe> in the function's name indicates that it will not attempt to read
664 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
665 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
666 character is malformed in some way, the program may croak, or the function may
667 return FALSE, at the discretion of the implementation, and subject to change in
670 Variant C<isI<FOO>_LC_utf8> is like C<isI<FOO>_LC_utf8_safe>, but takes just a single
671 parameter, C<p>, which has the same meaning as the corresponding parameter does
672 in C<isI<FOO>_LC_utf8_safe>. The function therefore can't check if it is reading
673 beyond the end of the string. Starting in Perl v5.30, it will take a second
674 parameter, becoming a synonym for C<isI<FOO>_LC_utf8_safe>. At that time every
675 program that uses it will have to be changed to successfully compile. In the
676 meantime, the first runtime call to C<isI<FOO>_LC_utf8> from each call point in
677 the program will raise a deprecation warning, enabled by default. You can
678 convert your program now to use C<isI<FOO>_LC_utf8_safe>, and avoid the warnings,
679 and get an extra measure of protection, or you can wait until v5.30, when
680 you'll be forced to add the C<e> parameter.
682 =for apidoc Am|bool|isALPHA|char ch
683 Returns a boolean indicating whether the specified character is an
684 alphabetic character, analogous to C<m/[[:alpha:]]/>.
685 See the L<top of this section|/Character classification> for an explanation of
687 C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8_safe>,
688 C<isALPHA_LC>, C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8_safe>.
690 =for apidoc Am|bool|isALPHANUMERIC|char ch
691 Returns a boolean indicating whether the specified character is a either an
692 alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>.
693 See the L<top of this section|/Character classification> for an explanation of
695 C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
696 C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>,
697 and C<isALPHANUMERIC_LC_utf8_safe>.
699 =for apidoc Am|bool|isASCII|char ch
700 Returns a boolean indicating whether the specified character is one of the 128
701 characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
702 On non-ASCII platforms, it returns TRUE iff this
703 character corresponds to an ASCII character. Variants C<isASCII_A()> and
704 C<isASCII_L1()> are identical to C<isASCII()>.
705 See the L<top of this section|/Character classification> for an explanation of
707 C<isASCII_uvchr>, C<isASCII_utf8_safe>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and
708 C<isASCII_LC_utf8_safe>. Note, however, that some platforms do not have the C
709 library routine C<isascii()>. In these cases, the variants whose names contain
710 C<LC> are the same as the corresponding ones without.
712 Also note, that because all ASCII characters are UTF-8 invariant (meaning they
713 have the exact same representation (always a single byte) whether encoded in
714 UTF-8 or not), C<isASCII> will give the correct results when called with any
715 byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8_safe>
716 will work properly on any string encoded or not in UTF-8.
718 =for apidoc Am|bool|isBLANK|char ch
719 Returns a boolean indicating whether the specified character is a
720 character considered to be a blank, analogous to C<m/[[:blank:]]/>.
721 See the L<top of this section|/Character classification> for an explanation of
723 C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8_safe>,
724 C<isBLANK_LC>, C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8_safe>. Note,
725 however, that some platforms do not have the C library routine
726 C<isblank()>. In these cases, the variants whose names contain C<LC> are
727 the same as the corresponding ones without.
729 =for apidoc Am|bool|isCNTRL|char ch
730 Returns a boolean indicating whether the specified character is a
731 control character, analogous to C<m/[[:cntrl:]]/>.
732 See the L<top of this section|/Character classification> for an explanation of
734 C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8_safe>,
735 C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8_safe> On EBCDIC
736 platforms, you almost always want to use the C<isCNTRL_L1> variant.
738 =for apidoc Am|bool|isDIGIT|char ch
739 Returns a boolean indicating whether the specified character is a
740 digit, analogous to C<m/[[:digit:]]/>.
741 Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
742 See the L<top of this section|/Character classification> for an explanation of
744 C<isDIGIT_uvchr>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and
745 C<isDIGIT_LC_utf8_safe>.
747 =for apidoc Am|bool|isGRAPH|char ch
748 Returns a boolean indicating whether the specified character is a
749 graphic character, analogous to C<m/[[:graph:]]/>.
750 See the L<top of this section|/Character classification> for an explanation of
751 variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8_safe>,
752 C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8_safe>.
754 =for apidoc Am|bool|isLOWER|char ch
755 Returns a boolean indicating whether the specified character is a
756 lowercase character, analogous to C<m/[[:lower:]]/>.
757 See the L<top of this section|/Character classification> for an explanation of
759 C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8_safe>,
760 C<isLOWER_LC>, C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8_safe>.
762 =for apidoc Am|bool|isOCTAL|char ch
763 Returns a boolean indicating whether the specified character is an
765 The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
768 =for apidoc Am|bool|isPUNCT|char ch
769 Returns a boolean indicating whether the specified character is a
770 punctuation character, analogous to C<m/[[:punct:]]/>.
771 Note that the definition of what is punctuation isn't as
772 straightforward as one might desire. See L<perlrecharclass/POSIX Character
773 Classes> for details.
774 See the L<top of this section|/Character classification> for an explanation of
775 variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8_safe>,
776 C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8_safe>.
778 =for apidoc Am|bool|isSPACE|char ch
779 Returns a boolean indicating whether the specified character is a
780 whitespace character. This is analogous
781 to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
782 this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
783 locale forms of this macro (the ones with C<LC> in their names) matched
784 precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
785 in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
786 (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
787 See the L<top of this section|/Character classification> for an explanation of
789 C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8_safe>,
790 C<isSPACE_LC>, C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8_safe>.
792 =for apidoc Am|bool|isPSXSPC|char ch
793 (short for Posix Space)
794 Starting in 5.18, this is identical in all its forms to the
795 corresponding C<isSPACE()> macros.
796 The locale forms of this macro are identical to their corresponding
797 C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
798 non-locale forms differ from their C<isSPACE()> forms only in that the
799 C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
800 Otherwise they are identical. Thus this macro is analogous to what
801 C<m/[[:space:]]/> matches in a regular expression.
802 See the L<top of this section|/Character classification> for an explanation of
803 variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8_safe>,
804 C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8_safe>.
806 =for apidoc Am|bool|isUPPER|char ch
807 Returns a boolean indicating whether the specified character is an
808 uppercase character, analogous to C<m/[[:upper:]]/>.
809 See the L<top of this section|/Character classification> for an explanation of
810 variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8_safe>,
811 C<isUPPER_LC>, C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8_safe>.
813 =for apidoc Am|bool|isPRINT|char ch
814 Returns a boolean indicating whether the specified character is a
815 printable character, analogous to C<m/[[:print:]]/>.
816 See the L<top of this section|/Character classification> for an explanation of
818 C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8_safe>,
819 C<isPRINT_LC>, C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8_safe>.
821 =for apidoc Am|bool|isWORDCHAR|char ch
822 Returns a boolean indicating whether the specified character is a character
823 that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
824 in a regular expression. A word character is an alphabetic character, a
825 decimal digit, a connecting punctuation character (such as an underscore), or
826 a "mark" character that attaches to one of those (like some sort of accent).
827 C<isALNUM()> is a synonym provided for backward compatibility, even though a
828 word character includes more than the standard C language meaning of
830 See the L<top of this section|/Character classification> for an explanation of
831 variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and
832 C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and
833 C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
834 include the platform's native underscore.
836 =for apidoc Am|bool|isXDIGIT|char ch
837 Returns a boolean indicating whether the specified character is a hexadecimal
838 digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
839 and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
840 See the L<top of this section|/Character classification> for an explanation of
842 C<isXDIGIT_uvchr>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>,
843 and C<isXDIGIT_LC_utf8_safe>.
845 =for apidoc Am|bool|isIDFIRST|char ch
846 Returns a boolean indicating whether the specified character can be the first
847 character of an identifier. This is very close to, but not quite the same as
848 the official Unicode property C<XID_Start>. The difference is that this
849 returns true only if the input character also matches L</isWORDCHAR>.
850 See the L<top of this section|/Character classification> for an explanation of
852 C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8_safe>,
853 C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8_safe>.
855 =for apidoc Am|bool|isIDCONT|char ch
856 Returns a boolean indicating whether the specified character can be the
857 second or succeeding character of an identifier. This is very close to, but
858 not quite the same as the official Unicode property C<XID_Continue>. The
859 difference is that this returns true only if the input character also matches
860 L</isWORDCHAR>. See the L<top of this section|/Character classification> for
862 explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
863 C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and
864 C<isIDCONT_LC_utf8_safe>.
866 =head1 Miscellaneous Functions
868 =for apidoc Am|U8|READ_XDIGIT|char str*
869 Returns the value of an ASCII-range hex digit and advances the string pointer.
870 Behaviour is only well defined when isXDIGIT(*str) is true.
872 =head1 Character case changing
873 Perl uses "full" Unicode case mappings. This means that converting a single
874 character to another case may result in a sequence of more than one character.
875 For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
876 character sequence C<SS>. This presents some complications The lowercase of
877 all characters in the range 0..255 is a single character, and thus
878 C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
879 return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
880 an API that does allow every possible legal result to be returned.) Likewise
881 no other function that is crippled by not being able to give the correct
882 results for the full range of possible inputs has been implemented here.
884 =for apidoc Am|U8|toUPPER|U8 ch
885 Converts the specified character to uppercase. If the input is anything but an
886 ASCII lowercase character, that input character itself is returned. Variant
887 C<toUPPER_A> is equivalent.
889 =for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
890 Converts the code point C<cp> to its uppercase version, and
891 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
892 point is interpreted as native if less than 256; otherwise as Unicode. Note
893 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
894 bytes since the uppercase version may be longer than the original character.
896 The first code point of the uppercased version is returned
897 (but note, as explained at L<the top of this section|/Character case
898 changing>, that there may be more.)
900 =for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
901 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
902 extending no further than S<C<e - 1>> to its uppercase version, and
903 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
904 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
905 bytes since the uppercase version may be longer than the original character.
907 The first code point of the uppercased version is returned
908 (but note, as explained at L<the top of this section|/Character case
909 changing>, that there may be more).
911 The suffix C<_safe> in the function's name indicates that it will not attempt
912 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
913 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
914 input character is malformed in some way, the program may croak, or the
915 function may return the REPLACEMENT CHARACTER, at the discretion of the
916 implementation, and subject to change in future releases.
918 =for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp
919 This is like C<L</toUPPER_utf8_safe>>, but doesn't have the C<e>
920 parameter The function therefore can't check if it is reading
921 beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
922 parameter, becoming a synonym for C<toUPPER_utf8_safe>. At that time every
923 program that uses it will have to be changed to successfully compile. In the
924 meantime, the first runtime call to C<toUPPER_utf8> from each call point in the
925 program will raise a deprecation warning, enabled by default. You can convert
926 your program now to use C<toUPPER_utf8_safe>, and avoid the warnings, and get an
927 extra measure of protection, or you can wait until v5.30, when you'll be forced
928 to add the C<e> parameter.
930 =for apidoc Am|U8|toFOLD|U8 ch
931 Converts the specified character to foldcase. If the input is anything but an
932 ASCII uppercase character, that input character itself is returned. Variant
933 C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
934 Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
936 =for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
937 Converts the code point C<cp> to its foldcase version, and
938 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
939 point is interpreted as native if less than 256; otherwise as Unicode. Note
940 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
941 bytes since the foldcase version may be longer than the original character.
943 The first code point of the foldcased version is returned
944 (but note, as explained at L<the top of this section|/Character case
945 changing>, that there may be more).
947 =for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
948 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
949 extending no further than S<C<e - 1>> to its foldcase version, and
950 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
951 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
952 bytes since the foldcase version may be longer than the original character.
954 The first code point of the foldcased version is returned
955 (but note, as explained at L<the top of this section|/Character case
956 changing>, that there may be more).
958 The suffix C<_safe> in the function's name indicates that it will not attempt
959 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
960 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
961 input character is malformed in some way, the program may croak, or the
962 function may return the REPLACEMENT CHARACTER, at the discretion of the
963 implementation, and subject to change in future releases.
965 =for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp
966 This is like C<L</toFOLD_utf8_safe>>, but doesn't have the C<e>
967 parameter The function therefore can't check if it is reading
968 beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
969 parameter, becoming a synonym for C<toFOLD_utf8_safe>. At that time every
970 program that uses it will have to be changed to successfully compile. In the
971 meantime, the first runtime call to C<toFOLD_utf8> from each call point in the
972 program will raise a deprecation warning, enabled by default. You can convert
973 your program now to use C<toFOLD_utf8_safe>, and avoid the warnings, and get an
974 extra measure of protection, or you can wait until v5.30, when you'll be forced
975 to add the C<e> parameter.
977 =for apidoc Am|U8|toLOWER|U8 ch
978 Converts the specified character to lowercase. If the input is anything but an
979 ASCII uppercase character, that input character itself is returned. Variant
980 C<toLOWER_A> is equivalent.
982 =for apidoc Am|U8|toLOWER_L1|U8 ch
983 Converts the specified Latin1 character to lowercase. The results are
984 undefined if the input doesn't fit in a byte.
986 =for apidoc Am|U8|toLOWER_LC|U8 ch
987 Converts the specified character to lowercase using the current locale's rules,
988 if possible; otherwise returns the input character itself.
990 =for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
991 Converts the code point C<cp> to its lowercase version, and
992 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
993 point is interpreted as native if less than 256; otherwise as Unicode. Note
994 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
995 bytes since the lowercase version may be longer than the original character.
997 The first code point of the lowercased version is returned
998 (but note, as explained at L<the top of this section|/Character case
999 changing>, that there may be more).
1002 =for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1003 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1004 extending no further than S<C<e - 1>> to its lowercase version, and
1005 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1006 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1007 bytes since the lowercase version may be longer than the original character.
1009 The first code point of the lowercased version is returned
1010 (but note, as explained at L<the top of this section|/Character case
1011 changing>, that there may be more).
1013 The suffix C<_safe> in the function's name indicates that it will not attempt
1014 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1015 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1016 input character is malformed in some way, the program may croak, or the
1017 function may return the REPLACEMENT CHARACTER, at the discretion of the
1018 implementation, and subject to change in future releases.
1020 =for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp
1021 This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1022 parameter The function therefore can't check if it is reading
1023 beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
1024 parameter, becoming a synonym for C<toLOWER_utf8_safe>. At that time every
1025 program that uses it will have to be changed to successfully compile. In the
1026 meantime, the first runtime call to C<toLOWER_utf8> from each call point in the
1027 program will raise a deprecation warning, enabled by default. You can convert
1028 your program now to use C<toLOWER_utf8_safe>, and avoid the warnings, and get an
1029 extra measure of protection, or you can wait until v5.30, when you'll be forced
1030 to add the C<e> parameter.
1032 =for apidoc Am|U8|toTITLE|U8 ch
1033 Converts the specified character to titlecase. If the input is anything but an
1034 ASCII lowercase character, that input character itself is returned. Variant
1035 C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
1036 range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
1037 not a concept used in locale handling, so there is no functionality for that.)
1039 =for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1040 Converts the code point C<cp> to its titlecase version, and
1041 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1042 point is interpreted as native if less than 256; otherwise as Unicode. Note
1043 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1044 bytes since the titlecase version may be longer than the original character.
1046 The first code point of the titlecased version is returned
1047 (but note, as explained at L<the top of this section|/Character case
1048 changing>, that there may be more).
1050 =for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1051 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1052 extending no further than S<C<e - 1>> to its titlecase version, and
1053 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1054 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1055 bytes since the titlecase version may be longer than the original character.
1057 The first code point of the titlecased version is returned
1058 (but note, as explained at L<the top of this section|/Character case
1059 changing>, that there may be more).
1061 The suffix C<_safe> in the function's name indicates that it will not attempt
1062 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1063 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1064 input character is malformed in some way, the program may croak, or the
1065 function may return the REPLACEMENT CHARACTER, at the discretion of the
1066 implementation, and subject to change in future releases.
1068 =for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp
1069 This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e>
1070 parameter The function therefore can't check if it is reading
1071 beyond the end of the string. Starting in Perl v5.30, it will take the C<e>
1072 parameter, becoming a synonym for C<toTITLE_utf8_safe>. At that time every
1073 program that uses it will have to be changed to successfully compile. In the
1074 meantime, the first runtime call to C<toTITLE_utf8> from each call point in the
1075 program will raise a deprecation warning, enabled by default. You can convert
1076 your program now to use C<toTITLE_utf8_safe>, and avoid the warnings, and get an
1077 extra measure of protection, or you can wait until v5.30, when you'll be forced
1078 to add the C<e> parameter.
1082 XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1083 really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1084 and aren't general purpose as they don't work on U+DF, and assert against that.
1086 Note that these macros are repeated in Devel::PPPort, so should also be
1087 patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1091 /* Specify the widest unsigned type on the platform. */
1093 # define WIDEST_UTYPE U64
1095 # define WIDEST_UTYPE U32
1098 /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1099 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1100 * bits of information are lost even on a 64-bit machine, but to get the
1101 * compiler to optimize it out if possible. This is because Configure makes
1102 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1103 * sizeof() guarantees that this evaluates to a constant true at compile time.
1105 * For Coverity, be always true, because otherwise Coverity thinks
1106 * it finds several expressions that are always true, independent
1107 * of operands. Well, they are, but that is kind of the point.
1109 #ifndef __COVERITY__
1110 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1112 #define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
1113 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
1115 #define FITS_IN_8_BITS(c) (1)
1119 # ifndef _ALL_SOURCE
1120 /* The native libc isascii() et.al. functions return the wrong results
1121 * on at least z/OS unless this is defined. */
1122 # error _ALL_SOURCE should probably be defined
1125 /* There is a simple definition of ASCII for ASCII platforms. But the
1126 * EBCDIC one isn't so simple, so is defined using table look-up like the
1127 * other macros below.
1129 * The cast here is used instead of '(c) >= 0', because some compilers emit
1130 * a warning that that test is always true when the parameter is an
1131 * unsigned type. khw supposes that it could be written as
1132 * && ((c) == '\0' || (c) > 0)
1133 * to avoid the message, but the cast will likely avoid extra branches even
1134 * with stupid compilers.
1136 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1138 # define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
1141 /* Take the eight possible bit patterns of the lower 3 bits and you get the
1142 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1143 * can be ignored. If the rest match '0', we have an octal */
1144 #define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
1146 #ifdef H_PERL /* If have access to perl.h, lookup in its table */
1148 /* Character class numbers. For internal core Perl use only. The ones less
1149 * than 32 are used in PL_charclass[] and the ones up through the one that
1150 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1151 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1152 * their actual definitions are here. If that file has a name not used here,
1155 * The first group of these is ordered in what I (khw) estimate to be the
1156 * frequency of their use. This gives a slight edge to exiting a loop earlier
1157 * (in reginclass() in regexec.c) */
1158 # define _CC_WORDCHAR 0 /* \w and [:word:] */
1159 # define _CC_DIGIT 1 /* \d and [:digit:] */
1160 # define _CC_ALPHA 2 /* [:alpha:] */
1161 # define _CC_LOWER 3 /* [:lower:] */
1162 # define _CC_UPPER 4 /* [:upper:] */
1163 # define _CC_PUNCT 5 /* [:punct:] */
1164 # define _CC_PRINT 6 /* [:print:] */
1165 # define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1166 # define _CC_GRAPH 8 /* [:graph:] */
1167 # define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1169 #define _FIRST_NON_SWASH_CC 10
1170 /* The character classes above are implemented with swashes. The second group
1171 * (just below) contains the ones implemented without. These are also sorted
1172 * in rough order of the frequency of their use, except that \v should be last,
1173 * as it isn't a real Posix character class, and some (small) inefficiencies in
1174 * regular expression handling would be introduced by putting it in the middle
1175 * of those that are. Also, cntrl and ascii come after the others as it may be
1176 * useful to group these which have no members that match above Latin1, (or
1177 * above ASCII in the latter case) */
1179 # define _CC_SPACE 10 /* \s, [:space:] */
1180 # define _CC_PSXSPC _CC_SPACE /* XXX Temporary, can be removed
1181 when the deprecated isFOO_utf8()
1182 functions are removed */
1183 # define _CC_BLANK 11 /* [:blank:] */
1184 # define _CC_XDIGIT 12 /* [:xdigit:] */
1185 # define _CC_CNTRL 13 /* [:cntrl:] */
1186 # define _CC_ASCII 14 /* [:ascii:] */
1187 # define _CC_VERTSPACE 15 /* \v */
1189 # define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1191 /* The members of the third group below do not need to be coordinated with data
1192 * structures in regcomp.[ch] and regexec.c. */
1193 # define _CC_IDFIRST 16
1194 # define _CC_CHARNAME_CONT 17
1195 # define _CC_NONLATIN1_FOLD 18
1196 # define _CC_NONLATIN1_SIMPLE_FOLD 19
1197 # define _CC_QUOTEMETA 20
1198 # define _CC_NON_FINAL_FOLD 21
1199 # define _CC_IS_IN_SOME_FOLD 22
1200 # define _CC_MNEMONIC_CNTRL 23
1202 # define _CC_IDCONT 24 /* XXX Temporary, can be removed when the deprecated
1203 isFOO_utf8() functions are removed */
1205 /* This next group is only used on EBCDIC platforms, so theoretically could be
1206 * shared with something entirely different that's only on ASCII platforms */
1207 # define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28
1208 # define _CC_UTF8_IS_START 29
1209 # define _CC_UTF8_IS_DOWNGRADEABLE_START 30
1210 # define _CC_UTF8_IS_CONTINUATION 31
1212 * If more bits are needed, one could add a second word for non-64bit
1213 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
1214 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1215 * is used only for optimization (as of this writing), and differs in the
1216 * Latin1 range from the ALPHA bit only in two relatively unimportant
1217 * characters: the masculine and feminine ordinal indicators, so removing it
1218 * would just cause /i regexes which match them to run less efficiently.
1219 * Similarly the EBCDIC-only bits are used just for speed, and could be
1220 * replaced by other means */
1222 #if defined(PERL_CORE) || defined(PERL_EXT)
1223 /* An enum version of the character class numbers, to help compilers
1226 _CC_ENUM_ALPHA = _CC_ALPHA,
1227 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1228 _CC_ENUM_ASCII = _CC_ASCII,
1229 _CC_ENUM_BLANK = _CC_BLANK,
1230 _CC_ENUM_CASED = _CC_CASED,
1231 _CC_ENUM_CNTRL = _CC_CNTRL,
1232 _CC_ENUM_DIGIT = _CC_DIGIT,
1233 _CC_ENUM_GRAPH = _CC_GRAPH,
1234 _CC_ENUM_LOWER = _CC_LOWER,
1235 _CC_ENUM_PRINT = _CC_PRINT,
1236 _CC_ENUM_PUNCT = _CC_PUNCT,
1237 _CC_ENUM_SPACE = _CC_SPACE,
1238 _CC_ENUM_UPPER = _CC_UPPER,
1239 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
1240 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
1241 _CC_ENUM_XDIGIT = _CC_XDIGIT
1242 } _char_class_number;
1245 #define POSIX_SWASH_COUNT _FIRST_NON_SWASH_CC
1246 #define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
1248 #if defined(PERL_IN_UTF8_C) \
1249 || defined(PERL_IN_REGCOMP_C) \
1250 || defined(PERL_IN_REGEXEC_C)
1251 # if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
1252 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
1253 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9
1254 #error Need to adjust order of swash_property_names[]
1257 /* This is declared static in each of the few files that this is #defined for
1258 * to keep them from being publicly accessible. Hence there is a small amount
1259 * of wasted space */
1261 static const char* const swash_property_names[] = {
1277 EXTCONST U32 PL_charclass[] = {
1278 # include "l1_char_class_tab.h"
1281 # else /* ! DOINIT */
1282 EXTCONST U32 PL_charclass[];
1286 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
1287 # define _CC_mask(classnum) (1U << (classnum))
1289 /* For internal core Perl use only: the base macro for defining macros like
1291 # define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
1292 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
1294 /* The mask for the _A versions of the macros; it just adds in the bit for
1296 # define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1298 /* For internal core Perl use only: the base macro for defining macros like
1299 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1300 * the ASCII bit are present */
1301 # define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1302 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1303 == _CC_mask_A(classnum)))
1305 # define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1306 # define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
1307 # define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1308 # define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
1309 # define isDIGIT_A(c) _generic_isCC(c, _CC_DIGIT) /* No non-ASCII digits */
1310 # define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1311 # define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1312 # define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1313 # define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1314 # define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1315 # define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1316 # define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
1317 # define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1319 # define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
1320 # define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1321 # define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1322 # define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1324 /* continuation character for legal NAME in \N{NAME} */
1325 # define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1327 # define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1328 # define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1329 # define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1330 # define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1331 # define isPSXSPC_L1(c) isSPACE_L1(c)
1332 # define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1333 # define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1334 # define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1335 # define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1336 # define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
1339 # define isASCII(c) _generic_isCC(c, _CC_ASCII)
1342 /* Participates in a single-character fold with a character above 255 */
1343 # 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)))
1345 /* Like the above, but also can be part of a multi-char fold */
1346 # 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)))
1348 # define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
1349 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1350 _generic_isCC(c, _CC_NON_FINAL_FOLD)
1351 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1352 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1353 # define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1354 _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1355 #else /* else we don't have perl.h H_PERL */
1357 /* If we don't have perl.h, we are compiling a utility program. Below we
1358 * hard-code various macro definitions that wouldn't otherwise be available
1359 * to it. Most are coded based on first principles. These are written to
1360 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
1361 # define isDIGIT_A(c) ((c) <= '9' && (c) >= '0')
1362 # define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1363 # define isSPACE_A(c) (isBLANK_A(c) \
1368 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1369 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1370 * performance (if optimization isn't on), and allow the same code to be
1371 * used for both platform types */
1372 # define isLOWER_A(c) ((c) >= 'a' && (c) <= 'z' \
1374 || ((c) >= 'j' && (c) <= 'r') \
1376 # define isUPPER_A(c) ((c) >= 'A' && (c) <= 'Z' \
1378 || ((c) >= 'J' && (c) <= 'R') \
1380 # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1381 # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
1382 # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
1383 # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1384 # define isXDIGIT_A(c) (isDIGIT_A(c) \
1385 || ((c) >= 'a' && (c) <= 'f') \
1386 || ((c) <= 'F' && (c) >= 'A'))
1387 # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1388 || (c) == '#' || (c) == '$' || (c) == '%' \
1389 || (c) == '&' || (c) == '\'' || (c) == '(' \
1390 || (c) == ')' || (c) == '*' || (c) == '+' \
1391 || (c) == ',' || (c) == '.' || (c) == '/' \
1392 || (c) == ':' || (c) == ';' || (c) == '<' \
1393 || (c) == '=' || (c) == '>' || (c) == '?' \
1394 || (c) == '@' || (c) == '[' || (c) == '\\' \
1395 || (c) == ']' || (c) == '^' || (c) == '_' \
1396 || (c) == '`' || (c) == '{' || (c) == '|' \
1397 || (c) == '}' || (c) == '~')
1398 # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1399 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1402 /* The below is accurate for the 3 EBCDIC code pages traditionally
1403 * supported by perl. The only difference between them in the controls
1404 * is the position of \n, and that is represented symbolically below */
1405 # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1406 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1407 || (c) == '\t' || (c) == '\v' \
1408 || ((c) <= 3 && (c) >= 1) /* SOH, STX, ETX */ \
1409 || (c) == 7 /* U+7F DEL */ \
1410 || ((c) <= 0x13 && (c) >= 0x0E) /* SO, SI */ \
1411 /* DLE, DC[1-3] */ \
1412 || (c) == 0x18 /* U+18 CAN */ \
1413 || (c) == 0x19 /* U+19 EOM */ \
1414 || ((c) <= 0x1F && (c) >= 0x1C) /* [FGRU]S */ \
1415 || (c) == 0x26 /* U+17 ETB */ \
1416 || (c) == 0x27 /* U+1B ESC */ \
1417 || (c) == 0x2D /* U+05 ENQ */ \
1418 || (c) == 0x2E /* U+06 ACK */ \
1419 || (c) == 0x32 /* U+16 SYN */ \
1420 || (c) == 0x37 /* U+04 EOT */ \
1421 || (c) == 0x3C /* U+14 DC4 */ \
1422 || (c) == 0x3D /* U+15 NAK */ \
1423 || (c) == 0x3F)/* U+1A SUB */
1424 # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1425 # else /* isASCII is already defined for ASCII platforms, so can use that to
1427 # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
1430 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
1431 * during debugging, and it seems best to keep them. We may be called
1432 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1433 * do anything anyway, so make it not a problem */
1434 # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1435 # define NATIVE_TO_LATIN1(ch) (ch)
1437 # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1438 # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1439 # define isBLANK_L1(c) (isBLANK_A(c) \
1440 || (FITS_IN_8_BITS(c) \
1441 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1442 # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1443 # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1444 # define isLOWER_L1(c) (isLOWER_A(c) \
1445 || (FITS_IN_8_BITS(c) \
1446 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1447 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1448 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1449 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1450 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1451 # define isPRINT_L1(c) (isPRINT_A(c) \
1452 || (FITS_IN_8_BITS(c) \
1453 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1454 # define isPUNCT_L1(c) (isPUNCT_A(c) \
1455 || (FITS_IN_8_BITS(c) \
1456 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1457 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1458 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1459 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1460 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1461 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1462 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1463 # define isSPACE_L1(c) (isSPACE_A(c) \
1464 || (FITS_IN_8_BITS(c) \
1465 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
1466 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1467 # define isUPPER_L1(c) (isUPPER_A(c) \
1468 || (FITS_IN_8_BITS(c) \
1469 && ( NATIVE_TO_LATIN1((U8) c) >= 0xC0 \
1470 && NATIVE_TO_LATIN1((U8) c) <= 0xDE \
1471 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1472 # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1473 # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1474 # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1479 /* The following are not fully accurate in the above-ASCII range. I (khw)
1480 * don't think it's necessary to be so for the purposes where this gets
1482 # define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1483 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1485 /* And these aren't accurate at all. They are useful only for above
1486 * Latin1, which utilities and bootstrapping don't deal with */
1487 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
1488 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1489 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1491 /* Many of the macros later in this file are defined in terms of these. By
1492 * implementing them with a function, which converts the class number into
1493 * a call to the desired macro, all of the later ones work. However, that
1494 * function won't be actually defined when building a utility program (no
1495 * perl.h), and so a compiler error will be generated if one is attempted
1496 * to be used. And the above-Latin1 code points require Unicode tables to
1497 * be present, something unlikely to be the case when bootstrapping */
1498 # define _generic_isCC(c, classnum) \
1499 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1500 # define _generic_isCC_A(c, classnum) \
1501 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
1502 #endif /* End of no perl.h H_PERL */
1504 #define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1505 #define isALPHA(c) isALPHA_A(c)
1506 #define isASCII_A(c) isASCII(c)
1507 #define isASCII_L1(c) isASCII(c)
1508 #define isBLANK(c) isBLANK_A(c)
1509 #define isCNTRL(c) isCNTRL_A(c)
1510 #define isDIGIT(c) isDIGIT_A(c)
1511 #define isGRAPH(c) isGRAPH_A(c)
1512 #define isIDFIRST(c) isIDFIRST_A(c)
1513 #define isLOWER(c) isLOWER_A(c)
1514 #define isPRINT(c) isPRINT_A(c)
1515 #define isPSXSPC_A(c) isSPACE_A(c)
1516 #define isPSXSPC(c) isPSXSPC_A(c)
1517 #define isPSXSPC_L1(c) isSPACE_L1(c)
1518 #define isPUNCT(c) isPUNCT_A(c)
1519 #define isSPACE(c) isSPACE_A(c)
1520 #define isUPPER(c) isUPPER_A(c)
1521 #define isWORDCHAR(c) isWORDCHAR_A(c)
1522 #define isXDIGIT(c) isXDIGIT_A(c)
1524 /* ASCII casing. These could also be written as
1525 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1526 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1527 which uses table lookup and mask instead of subtraction. (This would
1528 work because the _MOD does not apply in the ASCII range) */
1529 #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1530 #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
1532 /* In the ASCII range, these are equivalent to what they're here defined to be.
1533 * But by creating these definitions, other code doesn't have to be aware of
1535 #define toFOLD(c) toLOWER(c)
1536 #define toTITLE(c) toUPPER(c)
1538 #define toLOWER_A(c) toLOWER(c)
1539 #define toUPPER_A(c) toUPPER(c)
1540 #define toFOLD_A(c) toFOLD(c)
1541 #define toTITLE_A(c) toTITLE(c)
1543 /* Use table lookup for speed; returns the input itself if is out-of-range */
1544 #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
1546 : PL_latin1_lc[ (U8) (c) ])
1547 #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1549 /* Modified uc. Is correct uc except for three non-ascii chars which are
1550 * all mapped to one of them, and these need special handling; returns the
1551 * input itself if is out-of-range */
1552 #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
1554 : PL_mod_latin1_uc[ (U8) (c) ])
1555 #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
1557 /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1559 /* For internal core Perl use only: the base macro for defining macros like
1560 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
1561 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1562 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1563 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1564 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1565 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1566 * code points under locale rules, use the macros based on _generic_LC_uvchr
1567 * instead of this. */
1568 #define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1569 (! FITS_IN_8_BITS(c) \
1571 : IN_UTF8_CTYPE_LOCALE \
1572 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
1575 /* For internal core Perl use only: a helper macro for defining macros like
1576 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1577 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1578 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1579 * _generic_LC_base for more info */
1580 #define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1581 _generic_LC_base(c,utf8_locale_classnum, \
1582 non_utf8_func( (_LC_CAST) (c)))
1584 /* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1585 * 'c' is the platform's native underscore character */
1586 #define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1587 _generic_LC_base(c, utf8_locale_classnum, \
1588 (non_utf8_func( (_LC_CAST) (c)) \
1589 || (char)(c) == '_'))
1591 /* These next three are also for internal core Perl use only: case-change
1593 #define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1595 : (IN_UTF8_CTYPE_LOCALE) \
1596 ? PL_latin1_lc[ (U8) (c) ] \
1597 : (cast)function((cast)(c)))
1599 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1600 * returns a single value, so can't adequately return the upper case of LATIN
1601 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1602 * values "SS"); instead it asserts against that under DEBUGGING, and
1603 * otherwise returns its input */
1604 #define _generic_toUPPER_LC(c, function, cast) \
1605 (! FITS_IN_8_BITS(c) \
1607 : ((! IN_UTF8_CTYPE_LOCALE) \
1608 ? (cast)function((cast)(c)) \
1609 : ((((U8)(c)) == MICRO_SIGN) \
1610 ? GREEK_CAPITAL_LETTER_MU \
1611 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1612 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1613 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1614 ? (__ASSERT_(0) (c)) \
1615 : PL_mod_latin1_uc[ (U8) (c) ])))))
1617 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1618 * returns a single value, so can't adequately return the fold case of LATIN
1619 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1620 * values "ss"); instead it asserts against that under DEBUGGING, and
1621 * otherwise returns its input */
1622 #define _generic_toFOLD_LC(c, function, cast) \
1623 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
1624 ? GREEK_SMALL_LETTER_MU \
1625 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1626 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1627 _generic_toLOWER_LC(c, function, cast)))
1629 /* Use the libc versions for these if available. */
1630 #if defined(HAS_ISASCII)
1631 # define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1633 # define isASCII_LC(c) isASCII(c)
1636 #if defined(HAS_ISBLANK)
1637 # define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
1638 #else /* Unlike isASCII, varies if in a UTF-8 locale */
1639 # define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
1645 /* The Windows functions don't bother to follow the POSIX standard, which
1646 * for example says that something can't both be a printable and a control.
1647 * But Windows treats the \t control as a printable, and does such things
1648 * as making superscripts into both digits and punctuation. This tames
1649 * these flaws by assuming that the definitions of both controls and space
1650 * are correct, and then making sure that other definitions don't have
1651 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1652 * Not all possible weirdnesses are checked for, just the ones that were
1653 * detected on actual Microsoft code pages */
1655 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1656 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1658 # define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1659 && isALPHANUMERIC_LC(c))
1660 # define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1662 # define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1663 isALPHANUMERIC_LC(c))
1664 # define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1665 # define isIDFIRST_LC(c) (((c) == '_') \
1666 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1667 # define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1668 # define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1669 # define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1670 # define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1671 # define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1672 # define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1673 && isALPHANUMERIC_LC(c))
1675 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1676 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1677 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1679 #elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
1680 /* For most other platforms */
1682 # define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1683 # define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1684 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1685 # define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1686 # define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1687 # define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1688 # define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1689 # define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1690 # define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1691 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1692 # define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1693 # define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1694 # define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1697 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1698 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1699 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1701 #else /* The final fallback position */
1703 # define isALPHA_LC(c) (isascii(c) && isalpha(c))
1704 # define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1705 # define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1706 # define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1707 # define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1708 # define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1709 # define isLOWER_LC(c) (isascii(c) && islower(c))
1710 # define isPRINT_LC(c) (isascii(c) && isprint(c))
1711 # define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1712 # define isSPACE_LC(c) (isascii(c) && isspace(c))
1713 # define isUPPER_LC(c) (isascii(c) && isupper(c))
1714 # define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1715 # define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1717 # define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1718 # define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1719 # define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
1723 #define isIDCONT(c) isWORDCHAR(c)
1724 #define isIDCONT_A(c) isWORDCHAR_A(c)
1725 #define isIDCONT_L1(c) isWORDCHAR_L1(c)
1726 #define isIDCONT_LC(c) isWORDCHAR_LC(c)
1727 #define isPSXSPC_LC(c) isSPACE_LC(c)
1729 /* For internal core Perl use only: the base macros for defining macros like
1730 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1731 * number defined earlier in this file. _generic_uvchr() is used for POSIX
1732 * classes where there is a macro or function 'above_latin1' that takes the
1733 * single argument 'c' and returns the desired value. These exist for those
1734 * classes which have simple definitions, avoiding the overhead of a hash
1735 * lookup or inversion list binary search. _generic_swash_uvchr() can be used
1736 * for classes where that overhead is faster than a direct lookup.
1737 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
1738 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1739 * have duplicate checks here, so could create versions of the macros that
1740 * don't, but experiments show that gcc optimizes them out anyway. */
1742 /* Note that all ignore 'use bytes' */
1743 #define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1744 ? _generic_isCC(c, classnum) \
1746 #define _generic_swash_uvchr(classnum, c) ((c) < 256 \
1747 ? _generic_isCC(c, classnum) \
1748 : _is_uni_FOO(classnum, c))
1749 #define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c)
1750 #define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c)
1751 #define isASCII_uvchr(c) isASCII(c)
1752 #define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
1753 #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
1754 #define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c)
1755 #define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c)
1756 #define isIDCONT_uvchr(c) \
1757 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
1758 #define isIDFIRST_uvchr(c) \
1759 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
1760 #define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c)
1761 #define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c)
1763 #define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c)
1764 #define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
1765 #define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
1767 #define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c)
1768 #define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
1769 #define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c)
1770 #define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
1772 #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
1773 #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
1774 #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
1775 #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
1777 /* For backwards compatibility, even though '_uni' should mean official Unicode
1778 * code points, in Perl it means native for those below 256 */
1779 #define isALPHA_uni(c) isALPHA_uvchr(c)
1780 #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
1781 #define isASCII_uni(c) isASCII_uvchr(c)
1782 #define isBLANK_uni(c) isBLANK_uvchr(c)
1783 #define isCNTRL_uni(c) isCNTRL_uvchr(c)
1784 #define isDIGIT_uni(c) isDIGIT_uvchr(c)
1785 #define isGRAPH_uni(c) isGRAPH_uvchr(c)
1786 #define isIDCONT_uni(c) isIDCONT_uvchr(c)
1787 #define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
1788 #define isLOWER_uni(c) isLOWER_uvchr(c)
1789 #define isPRINT_uni(c) isPRINT_uvchr(c)
1790 #define isPUNCT_uni(c) isPUNCT_uvchr(c)
1791 #define isSPACE_uni(c) isSPACE_uvchr(c)
1792 #define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
1793 #define isUPPER_uni(c) isUPPER_uvchr(c)
1794 #define isVERTWS_uni(c) isVERTWS_uvchr(c)
1795 #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
1796 #define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
1797 #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
1798 #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
1799 #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
1800 #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
1802 /* For internal core Perl use only: the base macros for defining macros like
1803 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
1804 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
1805 * simple class definitions; the other for more complex. These are like
1806 * _generic_uvchr, so see it for more info. */
1807 #define _generic_LC_uvchr(latin1, above_latin1, c) \
1808 (c < 256 ? latin1(c) : above_latin1(c))
1809 #define _generic_LC_swash_uvchr(latin1, classnum, c) \
1810 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
1812 #define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c)
1813 #define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \
1814 _CC_ALPHANUMERIC, c)
1815 #define isASCII_LC_uvchr(c) isASCII_LC(c)
1816 #define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
1817 is_HORIZWS_cp_high, c)
1818 #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
1819 #define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c)
1820 #define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c)
1821 #define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
1822 _is_uni_perl_idcont, c)
1823 #define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
1824 _is_uni_perl_idstart, c)
1825 #define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c)
1826 #define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c)
1827 #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
1828 #define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c)
1829 #define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
1830 is_XPERLSPACE_cp_high, c)
1831 #define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c)
1832 #define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \
1834 #define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
1835 is_XDIGIT_cp_high, c)
1837 #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
1839 /* For internal core Perl use only: the base macros for defining macros like
1840 * isALPHA_utf8. These are like the earlier defined macros, but take an input
1841 * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use
1842 * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the
1843 * 'utf8' parameter. This relies on the fact that ASCII characters have the
1844 * same representation whether utf8 or not. Note that it assumes that the utf8
1845 * has been validated, and ignores 'use bytes' */
1846 #define _base_generic_utf8(enum_name, name, p, use_locale ) \
1847 _is_utf8_FOO(CAT2(_CC_, enum_name), \
1849 "is" STRINGIFY(name) "_utf8", \
1850 "is" STRINGIFY(name) "_utf8_safe", \
1851 1, use_locale, __FILE__,__LINE__)
1853 #define _generic_utf8(name, p) _base_generic_utf8(name, name, p, 0)
1855 /* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
1856 * they don't otherwise go out of their way to look for malformed UTF-8. If
1857 * they can return accurate results without knowing if the input is otherwise
1858 * malformed, they do so. For example isASCII is accurate in spite of any
1859 * non-length malformations because it looks only at a single byte. Likewise
1860 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
1861 * variant ones return FALSE. But, if the input has to be well-formed in order
1862 * for the results to be accurate, the macros will test and if malformed will
1863 * call a routine to die
1865 * Except for toke.c, the macros do assume that e > p, asserting that on
1866 * DEBUGGING builds. Much code that calls these depends on this being true,
1867 * for other reasons. toke.c is treated specially as using the regular
1868 * assertion breaks it in many ways. All strings that these operate on there
1869 * are supposed to have an extra NUL character at the end, so that *e = \0. A
1870 * bunch of code in toke.c assumes that this is true, so the assertion allows
1872 #ifdef PERL_IN_TOKE_C
1873 # define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
1875 # define _utf8_safe_assert(p,e) ((e) > (p))
1878 #define _generic_utf8_safe(classnum, p, e, above_latin1) \
1879 (__ASSERT_(_utf8_safe_assert(p, e)) \
1880 (UTF8_IS_INVARIANT(*(p))) \
1881 ? _generic_isCC(*(p), classnum) \
1882 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
1883 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
1884 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
1886 : (_force_out_malformed_utf8_message( \
1887 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
1889 /* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
1890 * 'above_latin1' can be a macro */
1891 #define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
1892 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
1893 #define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \
1894 _generic_utf8_safe(classnum, p, e, \
1895 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1896 ? (_force_out_malformed_utf8_message( \
1897 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1899 /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
1900 * 'above_latin1' parameter */
1901 #define _generic_swash_utf8_safe(classnum, p, e) \
1902 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e))
1904 /* Like the above, but should be used only when it is known that there are no
1905 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
1906 * class is TRUE for. Hence it can skip the tests for this range.
1907 * 'above_latin1' should include its arguments */
1908 #define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
1909 (__ASSERT_(_utf8_safe_assert(p, e)) \
1910 (UTF8_IS_INVARIANT(*(p))) \
1911 ? _generic_isCC(*(p), classnum) \
1912 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
1913 ? 0 /* Note that doesn't check validity for latin1 */ \
1916 /* NOTE that some of these macros have very similar ones in regcharclass.h.
1917 * For example, there is (at the time of this writing) an 'is_SPACE_utf8()'
1918 * there, differing in name only by an underscore from the one here
1919 * 'isSPACE_utf8(). The difference is that the ones here are probably more
1920 * efficient and smaller, using an O(1) array lookup for Latin1-range code
1921 * points; the regcharclass.h ones are implemented as a series of
1922 * "if-else-if-else ..." */
1924 #define isALPHA_utf8(p) _generic_utf8(ALPHA, p)
1925 #define isALPHANUMERIC_utf8(p) _generic_utf8(ALPHANUMERIC, p)
1926 #define isASCII_utf8(p) _generic_utf8(ASCII, p)
1927 #define isBLANK_utf8(p) _generic_utf8(BLANK, p)
1928 #define isCNTRL_utf8(p) _generic_utf8(CNTRL, p)
1929 #define isDIGIT_utf8(p) _generic_utf8(DIGIT, p)
1930 #define isGRAPH_utf8(p) _generic_utf8(GRAPH, p)
1931 #define isIDCONT_utf8(p) _generic_utf8(IDCONT, p)
1932 #define isIDFIRST_utf8(p) _generic_utf8(IDFIRST, p)
1933 #define isLOWER_utf8(p) _generic_utf8(LOWER, p)
1934 #define isPRINT_utf8(p) _generic_utf8(PRINT, p)
1935 #define isPSXSPC_utf8(p) _generic_utf8(PSXSPC, p)
1936 #define isPUNCT_utf8(p) _generic_utf8(PUNCT, p)
1937 #define isSPACE_utf8(p) _generic_utf8(SPACE, p)
1938 #define isUPPER_utf8(p) _generic_utf8(UPPER, p)
1939 #define isVERTWS_utf8(p) _generic_utf8(VERTSPACE, p)
1940 #define isWORDCHAR_utf8(p) _generic_utf8(WORDCHAR, p)
1941 #define isXDIGIT_utf8(p) _generic_utf8(XDIGIT, p)
1943 #define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e)
1944 #define isALPHANUMERIC_utf8_safe(p, e) \
1945 _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e)
1946 #define isASCII_utf8_safe(p, e) \
1947 /* Because ASCII is invariant under utf8, the non-utf8 macro \
1949 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
1950 #define isBLANK_utf8_safe(p, e) \
1951 _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
1954 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
1955 * more efficient macro instead of the more general one */
1956 # define isCNTRL_utf8_safe(p, e) \
1957 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
1959 # define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
1962 #define isDIGIT_utf8_safe(p, e) \
1963 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
1964 _is_utf8_FOO_with_len(_CC_DIGIT, p, e))
1965 #define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e)
1966 #define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
1967 _is_utf8_perl_idcont_with_len, p, e)
1969 /* To prevent S_scan_word in toke.c from hanging, we have to make sure that
1970 * IDFIRST is an alnum. See
1971 * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you
1972 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
1973 * This used to be not the XID version, but we decided to go with the more
1974 * modern Unicode definition */
1975 #define isIDFIRST_utf8_safe(p, e) \
1976 _generic_func_utf8_safe(_CC_IDFIRST, \
1977 _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e))
1979 #define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e)
1980 #define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e)
1981 #define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
1982 #define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e)
1983 #define isSPACE_utf8_safe(p, e) \
1984 _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
1985 #define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e)
1986 #define isVERTWS_utf8_safe(p, e) \
1987 _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
1988 #define isWORDCHAR_utf8_safe(p, e) \
1989 _generic_swash_utf8_safe(_CC_WORDCHAR, p, e)
1990 #define isXDIGIT_utf8_safe(p, e) \
1991 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
1992 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
1993 ? (_force_out_malformed_utf8_message( \
1994 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
1995 : is_XDIGIT_high(p)))
1997 #define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l)
1998 #define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l)
1999 #define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l)
2000 #define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l)
2002 /* For internal core use only, subject to change */
2003 #define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f, "", 0)
2004 #define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f, "", 0)
2005 #define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f, "", 0)
2006 #define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f, "", 0)
2008 #define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2009 #define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2010 #define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2011 #define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
2013 /* For internal core Perl use only: the base macros for defining macros like
2014 * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point
2015 * in 'p' is within the 0-255 range, it uses locale rules from the passed-in
2016 * 'macro' parameter */
2017 #define _generic_LC_utf8(name, p) _base_generic_utf8(name, name, p, 1)
2019 #define isALPHA_LC_utf8(p) _generic_LC_utf8(ALPHA, p)
2020 #define isALPHANUMERIC_LC_utf8(p) _generic_LC_utf8(ALPHANUMERIC, p)
2021 #define isASCII_LC_utf8(p) _generic_LC_utf8(ASCII, p)
2022 #define isBLANK_LC_utf8(p) _generic_LC_utf8(BLANK, p)
2023 #define isCNTRL_LC_utf8(p) _generic_LC_utf8(CNTRL, p)
2024 #define isDIGIT_LC_utf8(p) _generic_LC_utf8(DIGIT, p)
2025 #define isGRAPH_LC_utf8(p) _generic_LC_utf8(GRAPH, p)
2026 #define isIDCONT_LC_utf8(p) _generic_LC_utf8(IDCONT, p)
2027 #define isIDFIRST_LC_utf8(p) _generic_LC_utf8(IDFIRST, p)
2028 #define isLOWER_LC_utf8(p) _generic_LC_utf8(LOWER, p)
2029 #define isPRINT_LC_utf8(p) _generic_LC_utf8(PRINT, p)
2030 #define isPSXSPC_LC_utf8(p) _generic_LC_utf8(PSXSPC, p)
2031 #define isPUNCT_LC_utf8(p) _generic_LC_utf8(PUNCT, p)
2032 #define isSPACE_LC_utf8(p) _generic_LC_utf8(SPACE, p)
2033 #define isUPPER_LC_utf8(p) _generic_LC_utf8(UPPER, p)
2034 #define isWORDCHAR_LC_utf8(p) _generic_LC_utf8(WORDCHAR, p)
2035 #define isXDIGIT_LC_utf8(p) _generic_LC_utf8(XDIGIT, p)
2037 /* For internal core Perl use only: the base macros for defining macros like
2038 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2039 * point in 'p' is within the 0-255 range, it uses locale rules from the
2040 * passed-in 'macro' parameter */
2041 #define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2042 (__ASSERT_(_utf8_safe_assert(p, e)) \
2043 (UTF8_IS_INVARIANT(*(p))) \
2045 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2046 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2047 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2048 : (_force_out_malformed_utf8_message( \
2049 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2052 #define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \
2053 _generic_LC_utf8_safe(macro, p, e, \
2054 _is_utf8_FOO_with_len(classnum, p, e))
2056 #define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2057 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2059 #define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \
2060 _generic_LC_utf8_safe(classnum, p, e, \
2061 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2062 ? (_force_out_malformed_utf8_message( \
2063 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2066 #define isALPHANUMERIC_LC_utf8_safe(p, e) \
2067 _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \
2068 _CC_ALPHANUMERIC, p, e)
2069 #define isALPHA_LC_utf8_safe(p, e) \
2070 _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2071 #define isASCII_LC_utf8_safe(p, e) \
2072 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2073 #define isBLANK_LC_utf8_safe(p, e) \
2074 _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2075 #define isCNTRL_LC_utf8_safe(p, e) \
2076 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2077 #define isDIGIT_LC_utf8_safe(p, e) \
2078 _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2079 #define isGRAPH_LC_utf8_safe(p, e) \
2080 _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2081 #define isIDCONT_LC_utf8_safe(p, e) \
2082 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2083 _is_utf8_perl_idcont_with_len, p, e)
2084 #define isIDFIRST_LC_utf8_safe(p, e) \
2085 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2086 _is_utf8_perl_idstart_with_len, p, e)
2087 #define isLOWER_LC_utf8_safe(p, e) \
2088 _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2089 #define isPRINT_LC_utf8_safe(p, e) \
2090 _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2091 #define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2092 #define isPUNCT_LC_utf8_safe(p, e) \
2093 _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2094 #define isSPACE_LC_utf8_safe(p, e) \
2095 _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2096 #define isUPPER_LC_utf8_safe(p, e) \
2097 _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2098 #define isWORDCHAR_LC_utf8_safe(p, e) \
2099 _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2100 #define isXDIGIT_LC_utf8_safe(p, e) \
2101 _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
2103 /* Macros for backwards compatibility and for completeness when the ASCII and
2104 * Latin1 values are identical */
2105 #define isALPHAU(c) isALPHA_L1(c)
2106 #define isDIGIT_L1(c) isDIGIT_A(c)
2107 #define isOCTAL(c) isOCTAL_A(c)
2108 #define isOCTAL_L1(c) isOCTAL_A(c)
2109 #define isXDIGIT_L1(c) isXDIGIT_A(c)
2110 #define isALNUM(c) isWORDCHAR(c)
2111 #define isALNUMU(c) isWORDCHAR_L1(c)
2112 #define isALNUM_LC(c) isWORDCHAR_LC(c)
2113 #define isALNUM_uni(c) isWORDCHAR_uni(c)
2114 #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
2115 #define isALNUM_utf8(p) isWORDCHAR_utf8(p)
2116 #define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p)
2117 #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2118 #define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2119 #define isALNUMC(c) isALPHANUMERIC(c)
2120 #define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2121 #define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
2122 #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
2123 #define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p)
2124 #define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p)
2126 /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2127 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2128 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2129 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2130 * the outlier from the block that contains the other controls, just like
2131 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2132 * block. If it weren't special cased, it would yield a non-control.
2133 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2136 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2138 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2140 ? (UNLIKELY((c) == '?') \
2141 ? QUESTION_MARK_CTRL \
2142 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2143 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2145 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2148 /* Line numbers are unsigned, 32 bits. */
2150 #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
2152 /* Helpful alias for version prescan */
2153 #define is_LAX_VERSION(a,b) \
2154 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2156 #define is_STRICT_VERSION(a,b) \
2157 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2159 #define BADVERSION(a,b,c) \
2165 /* Converts a character known to represent a hexadecimal digit (0-9, A-F, or
2166 * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer,
2167 * which is advanced. The input is validated only by an assert() in DEBUGGING
2168 * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and
2169 * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */
2170 #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \
2173 #define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \
2177 /* Converts a character known to represent an octal digit (0-7) to its numeric
2178 * value. The input is validated only by an assert() in DEBUGGING builds. In
2179 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2180 #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2182 /* Efficiently returns a boolean as to if two native characters are equivalent
2183 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2184 * the ALPHA in the name is to remind you of that. This is asserted() in
2185 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2186 * works (on valid input) for both non- and UTF-8-encoded bytes.
2188 * When one of the inputs is a compile-time constant and gets folded by the
2189 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2190 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2191 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
2192 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2193 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2194 * just a single 0, in the bit position where the upper- and lowercase differ.
2196 #define isALPHA_FOLD_EQ(c1, c2) \
2197 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2198 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2199 #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2202 =head1 Memory Management
2204 =for apidoc Am|void|Newx|void* ptr|int nitems|type
2205 The XSUB-writer's interface to the C C<malloc> function.
2207 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2209 In 5.9.3, Newx() and friends replace the older New() API, and drops
2210 the first parameter, I<x>, a debug aid which allowed callers to identify
2211 themselves. This aid has been superseded by a new build option,
2212 PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
2213 there for use in XS modules supporting older perls.
2215 =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
2216 The XSUB-writer's interface to the C C<malloc> function, with
2217 cast. See also C<L</Newx>>.
2219 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2221 =for apidoc Am|void|Newxz|void* ptr|int nitems|type
2222 The XSUB-writer's interface to the C C<malloc> function. The allocated
2223 memory is zeroed with C<memzero>. See also C<L</Newx>>.
2225 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2227 =for apidoc Am|void|Renew|void* ptr|int nitems|type
2228 The XSUB-writer's interface to the C C<realloc> function.
2230 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2232 =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2233 The XSUB-writer's interface to the C C<realloc> function, with
2236 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2238 =for apidoc Am|void|Safefree|void* ptr
2239 The XSUB-writer's interface to the C C<free> function.
2241 This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
2243 =for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2244 The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
2245 source, C<dest> is the destination, C<nitems> is the number of items, and
2246 C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
2248 =for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
2249 Like C<Move> but returns C<dest>. Useful
2250 for encouraging compilers to tail-call
2253 =for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2254 The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
2255 source, C<dest> is the destination, C<nitems> is the number of items, and
2256 C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
2258 =for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2260 Like C<Copy> but returns C<dest>. Useful
2261 for encouraging compilers to tail-call
2264 =for apidoc Am|void|Zero|void* dest|int nitems|type
2266 The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2267 destination, C<nitems> is the number of items, and C<type> is the type.
2269 =for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2271 Like C<Zero> but returns dest. Useful
2272 for encouraging compilers to tail-call
2275 =for apidoc Am|void|StructCopy|type *src|type *dest|type
2276 This is an architecture-independent macro to copy one structure to another.
2278 =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2280 Fill up memory with a byte pattern (a byte repeated over and over
2281 again) that hopefully catches attempts to access uninitialized memory.
2283 =for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2285 PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2287 =for apidoc Am|void|PoisonFree|void* dest|int nitems|type
2289 PoisonWith(0xEF) for catching access to freed memory.
2291 =for apidoc Am|void|Poison|void* dest|int nitems|type
2293 PoisonWith(0xEF) for catching access to freed memory.
2297 /* Maintained for backwards-compatibility only. Use newSV() instead. */
2299 #define NEWSV(x,len) newSV(len)
2302 #define MEM_SIZE_MAX ((MEM_SIZE)-1)
2304 #define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
2306 #ifdef PERL_MALLOC_WRAP
2308 /* This expression will be constant-folded at compile time. It checks
2309 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2310 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2311 * As well as avoiding the need for a run-time check in some cases, it's
2312 * designed to avoid compiler warnings like:
2313 * comparison is always false due to limited range of data type
2314 * It's mathematically equivalent to
2315 * max(n) * sizeof(t) > MEM_SIZE_MAX
2318 # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
2319 ( sizeof(MEM_SIZE) < sizeof(n) \
2320 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
2322 /* This is written in a slightly odd way to avoid various spurious
2323 * compiler warnings. We *want* to write the expression as
2324 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2325 * (for some compile-time constant C), but even when the LHS
2326 * constant-folds to false at compile-time, g++ insists on emitting
2327 * warnings about the RHS (e.g. "comparison is always false"), so instead
2330 * (cond ? n : X) > C
2332 * where X is a constant with X > C always false. Choosing a value for X
2333 * is tricky. If 0, some compilers will complain about 0 > C always being
2334 * false; if 1, Coverity complains when n happens to be the constant value
2335 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2336 * for X and hope that nothing else whines.
2339 # define _MEM_WRAP_WILL_WRAP(n,t) \
2340 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2341 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
2343 # define MEM_WRAP_CHECK(n,t) \
2344 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2345 && (croak_memory_wrap(),0))
2347 # define MEM_WRAP_CHECK_1(n,t,a) \
2348 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2349 && (Perl_croak_nocontext("%s",(a)),0))
2351 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
2353 #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))
2356 #define MEM_WRAP_CHECK(n,t)
2357 #define MEM_WRAP_CHECK_1(n,t,a)
2358 #define MEM_WRAP_CHECK_2(n,t,a,b)
2359 #define MEM_WRAP_CHECK_(n,t)
2361 #define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
2367 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
2368 * go through functions, which are handy for debugging breakpoints, but
2369 * which more importantly get the immediate calling environment (file and
2370 * line number, and C function name if available) passed in. This info can
2371 * then be used for logging the calls, for which one gets a sample
2372 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
2375 * - not all memory allocs get logged, only those
2376 * that go through Newx() and derivatives (while all
2377 * Safefrees do get logged)
2378 * - __FILE__ and __LINE__ do not work everywhere
2379 * - __func__ or __FUNCTION__ even less so
2380 * - I think more goes on after the perlio frees but
2381 * the thing is that STDERR gets closed (as do all
2382 * the file descriptors)
2383 * - no deeper calling stack than the caller of the Newx()
2384 * or the kind, but do I look like a C reflection/introspection
2386 * - the function prototypes for the logging functions
2387 * probably should maybe be somewhere else than handy.h
2388 * - one could consider inlining (macrofying) the logging
2389 * for speed, but I am too lazy
2390 * - one could imagine recording the allocations in a hash,
2391 * (keyed by the allocation address?), and maintain that
2392 * through reallocs and frees, but how to do that without
2393 * any News() happening...?
2394 * - lots of -Ddefines to get useful/controllable output
2395 * - lots of ENV reads
2399 # ifndef PERL_MEM_LOG_NOIMPL
2408 # if defined(PERL_IN_SV_C) /* those are only used in sv.c */
2409 void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2410 void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2417 #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2418 #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
2419 #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
2422 #ifndef MEM_LOG_ALLOC
2423 #define MEM_LOG_ALLOC(n,t,a) (a)
2425 #ifndef MEM_LOG_REALLOC
2426 #define MEM_LOG_REALLOC(n,t,v,a) (a)
2428 #ifndef MEM_LOG_FREE
2429 #define MEM_LOG_FREE(a) (a)
2432 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2433 #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2434 #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
2437 /* pre 5.9.x compatibility */
2438 #define New(x,v,n,t) Newx(v,n,t)
2439 #define Newc(x,v,n,t,c) Newxc(v,n,t,c)
2440 #define Newz(x,v,n,t) Newxz(v,n,t)
2443 #define Renew(v,n,t) \
2444 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2445 #define Renewc(v,n,t,c) \
2446 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2449 #define Safefree(d) \
2450 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
2452 #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
2455 /* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2456 * as it handles cases like constant string arguments without throwing warnings *
2457 * the cast is required, as is the inequality check, to avoid warnings */
2458 #define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
2461 #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)))
2462 #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)))
2463 #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2465 #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)))
2466 #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)))
2467 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
2469 #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2470 #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2471 #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2472 #define Poison(d,n,t) PoisonFree(d,n,t)
2475 # define PERL_POISON_EXPR(x) x
2477 # define PERL_POISON_EXPR(x)
2480 #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2482 /* C_ARRAY_LENGTH is the number of elements in the C array (so you
2483 * want your zero-based indices to be less than but not equal to).
2485 * C_ARRAY_END is one past the last: half-open/half-closed range,
2486 * not last-inclusive range. */
2487 #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
2488 #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
2492 # define Perl_va_copy(s, d) va_copy(d, s)
2493 # elif defined(__va_copy)
2494 # define Perl_va_copy(s, d) __va_copy(d, s)
2496 # define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2500 /* convenience debug macros */
2502 #define pTHX_FORMAT "Perl interpreter: 0x%p"
2503 #define pTHX__FORMAT ", Perl interpreter: 0x%p"
2504 #define pTHX_VALUE_ (void *)my_perl,
2505 #define pTHX_VALUE (void *)my_perl
2506 #define pTHX__VALUE_ ,(void *)my_perl,
2507 #define pTHX__VALUE ,(void *)my_perl
2510 #define pTHX__FORMAT
2513 #define pTHX__VALUE_
2515 #endif /* USE_ITHREADS */
2517 /* Perl_deprecate was not part of the public API, and did not have a deprecate()
2518 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2519 CPAN::Unpack show any users outside the core. */
2521 # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2522 "Use of " s " is deprecated")
2523 # define deprecate_disappears_in(when,message) \
2524 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2525 message ", and will disappear in Perl " when)
2526 # define deprecate_fatal_in(when,message) \
2527 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2528 message ". Its use will be fatal in Perl " when)
2531 /* Internal macros to deal with gids and uids */
2534 # if Uid_t_size > IVSIZE
2535 # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2536 # define SvUID(sv) SvNV(sv)
2537 # elif Uid_t_sign <= 0
2538 # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2539 # define SvUID(sv) SvIV(sv)
2541 # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2542 # define SvUID(sv) SvUV(sv)
2543 # endif /* Uid_t_size */
2545 # if Gid_t_size > IVSIZE
2546 # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2547 # define SvGID(sv) SvNV(sv)
2548 # elif Gid_t_sign <= 0
2549 # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2550 # define SvGID(sv) SvIV(sv)
2552 # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2553 # define SvGID(sv) SvUV(sv)
2554 # endif /* Gid_t_size */
2558 #endif /* PERL_HANDY_H_ */
2561 * ex: set ts=8 sts=4 sw=4 et:
https://perl5.git.perl.org / perl5.git / blob