Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* handy.h |
a687059c | 2 | * |
1129b882 | 3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000, |
da5d8dbb | 4 | * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others |
a687059c | 5 | * |
6e21c824 LW |
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. | |
8d063cd8 | 8 | * |
8d063cd8 LW |
9 | */ |
10 | ||
4650c663 KW |
11 | /* IMPORTANT NOTE: Everything whose name begins with an underscore is for |
12 | * internal core Perl use only. */ | |
13 | ||
6a5bc5ac KW |
14 | #ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */ |
15 | #define PERL_HANDY_H_ | |
9d745869 | 16 | |
24792b8d NC |
17 | #ifndef PERL_CORE |
18 | # define Null(type) ((type)NULL) | |
954c1994 GS |
19 | |
20 | /* | |
ccfc67b7 | 21 | =head1 Handy Values |
954c1994 | 22 | |
78342678 | 23 | =for apidoc AmnU||Nullch |
72d33970 FC |
24 | Null character pointer. (No longer available when C<PERL_CORE> is |
25 | defined.) | |
2307c6d0 | 26 | |
78342678 | 27 | =for apidoc AmnU||Nullsv |
72d33970 | 28 | Null SV pointer. (No longer available when C<PERL_CORE> is defined.) |
954c1994 GS |
29 | |
30 | =cut | |
31 | */ | |
32 | ||
24792b8d NC |
33 | # define Nullch Null(char*) |
34 | # define Nullfp Null(PerlIO*) | |
35 | # define Nullsv Null(SV*) | |
36 | #endif | |
8d063cd8 | 37 | |
641d3f0b | 38 | #ifdef TRUE |
39 | #undef TRUE | |
40 | #endif | |
41 | #ifdef FALSE | |
42 | #undef FALSE | |
43 | #endif | |
44 | #define TRUE (1) | |
45 | #define FALSE (0) | |
46 | ||
cf3f0ffb DM |
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; | |
49 | * e.g.: | |
50 | * | |
51 | * const SV *sv = ...; | |
52 | * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away | |
53 | * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn | |
54 | */ | |
55 | ||
b1bc3f34 | 56 | #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN) |
f065df04 | 57 | # define MUTABLE_PTR(p) ({ void *_p = (p); _p; }) |
b1bc3f34 NC |
58 | #else |
59 | # define MUTABLE_PTR(p) ((void *) (p)) | |
60 | #endif | |
61 | ||
a062e10d | 62 | #define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p)) |
ea726b52 | 63 | #define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p)) |
159b6efe | 64 | #define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p)) |
dbebbdb4 | 65 | #define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p)) |
a45c7426 | 66 | #define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p)) |
b1bc3f34 | 67 | #define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p)) |
27d4fb96 | 68 | |
f789f6a4 | 69 | #if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR) |
bd31be4b NC |
70 | # include <stdbool.h> |
71 | # ifndef HAS_BOOL | |
72 | # define HAS_BOOL 1 | |
73 | # endif | |
74 | #endif | |
75 | ||
8e84507e | 76 | /* bool is built-in for g++-2.6.3 and later, which might be used |
c1d22f6b GS |
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 | |
8e84507e | 79 | also defines _G_HAVE_BOOL for both gcc and g++, but only g++ |
c1d22f6b GS |
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 | |
5d94fbed | 83 | */ |
3609ea0d | 84 | #ifdef __GNUG__ /* GNU g++ has bool built-in */ |
f789f6a4 | 85 | # ifndef PERL_BOOL_AS_CHAR |
5d94fbed | 86 | # ifndef HAS_BOOL |
c1d22f6b | 87 | # define HAS_BOOL 1 |
5d94fbed | 88 | # endif |
f789f6a4 | 89 | # endif |
5d94fbed AD |
90 | #endif |
91 | ||
92 | #ifndef HAS_BOOL | |
f789f6a4 FC |
93 | # ifdef bool |
94 | # undef bool | |
95 | # endif | |
70d5cb32 | 96 | # define bool char |
c1d22f6b | 97 | # define HAS_BOOL 1 |
a687059c | 98 | #endif |
0d3e774c | 99 | |
236415e2 KW |
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 | |
103 | * AIX */ | |
18f5643b | 104 | #define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0) |
f2338a2e | 105 | |
46c6c7e2 | 106 | /* Try to figure out __func__ or __FUNCTION__ equivalent, if any. |
e352bcff JH |
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. */ | |
46c6c7e2 JH |
110 | #if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */ |
111 | # define FUNCTION__ __func__ | |
07798b17 AC |
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__ "" | |
46c6c7e2 | 115 | #else |
07798b17 | 116 | # define FUNCTION__ __FUNCTION__ /* Common extension. */ |
46c6c7e2 JH |
117 | #endif |
118 | ||
27d4fb96 | 119 | /* XXX A note on the perl source internal type system. The |
120 | original intent was that I32 be *exactly* 32 bits. | |
121 | ||
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 | |
24fef2a7 | 128 | |
dc45a647 MB |
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. | |
693762b4 | 132 | |
dc45a647 MB |
133 | Similarly, there is no guarantee that I16 and U16 have exactly 16 |
134 | bits. | |
693762b4 | 135 | |
8e84507e NIS |
136 | For dealing with issues that may arise from various 32/64-bit |
137 | systems, we will ask Configure to check out | |
8175356b | 138 | |
3609ea0d JH |
139 | SHORTSIZE == sizeof(short) |
140 | INTSIZE == sizeof(int) | |
141 | LONGSIZE == sizeof(long) | |
dc45a647 | 142 | LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG) |
3609ea0d | 143 | PTRSIZE == sizeof(void *) |
dc45a647 MB |
144 | DOUBLESIZE == sizeof(double) |
145 | LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE). | |
8175356b | 146 | |
27d4fb96 | 147 | */ |
148 | ||
69512466 JH |
149 | #ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */ |
150 | # include <inttypes.h> | |
dd0eed91 JH |
151 | # ifdef INT32_MIN_BROKEN |
152 | # undef INT32_MIN | |
153 | # define INT32_MIN (-2147483647-1) | |
154 | # endif | |
155 | # ifdef INT64_MIN_BROKEN | |
156 | # undef INT64_MIN | |
157 | # define INT64_MIN (-9223372036854775807LL-1) | |
158 | # endif | |
69512466 JH |
159 | #endif |
160 | ||
8175356b JH |
161 | typedef I8TYPE I8; |
162 | typedef U8TYPE U8; | |
163 | typedef I16TYPE I16; | |
164 | typedef U16TYPE U16; | |
165 | typedef I32TYPE I32; | |
166 | typedef U32TYPE U32; | |
16d89be8 | 167 | |
74b807c7 | 168 | #ifdef QUADKIND |
8175356b JH |
169 | typedef I64TYPE I64; |
170 | typedef U64TYPE U64; | |
16d89be8 | 171 | #endif |
8175356b | 172 | |
d8668976 | 173 | #if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX) |
5ff3f7a4 | 174 | |
5ff3f7a4 GS |
175 | /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type. |
176 | Please search CHAR_MAX in perl.h for further details. */ | |
177 | #define U8_MAX UINT8_MAX | |
178 | #define U8_MIN UINT8_MIN | |
179 | ||
5ff3f7a4 GS |
180 | #define I16_MAX INT16_MAX |
181 | #define I16_MIN INT16_MIN | |
182 | #define U16_MAX UINT16_MAX | |
183 | #define U16_MIN UINT16_MIN | |
184 | ||
5ff3f7a4 GS |
185 | #define I32_MAX INT32_MAX |
186 | #define I32_MIN INT32_MIN | |
0e983133 GS |
187 | #ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */ |
188 | # define U32_MAX UINT32_MAX | |
189 | #else | |
190 | # define U32_MAX 4294967295U | |
191 | #endif | |
5ff3f7a4 GS |
192 | #define U32_MIN UINT32_MIN |
193 | ||
194 | #else | |
195 | ||
5c9fa16e KA |
196 | /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type. |
197 | Please search CHAR_MAX in perl.h for further details. */ | |
27d4fb96 | 198 | #define U8_MAX PERL_UCHAR_MAX |
199 | #define U8_MIN PERL_UCHAR_MIN | |
79072805 | 200 | |
27d4fb96 | 201 | #define I16_MAX PERL_SHORT_MAX |
202 | #define I16_MIN PERL_SHORT_MIN | |
203 | #define U16_MAX PERL_USHORT_MAX | |
204 | #define U16_MIN PERL_USHORT_MIN | |
79072805 | 205 | |
c4f23d77 | 206 | #if LONGSIZE > 4 |
27d4fb96 | 207 | # define I32_MAX PERL_INT_MAX |
208 | # define I32_MIN PERL_INT_MIN | |
209 | # define U32_MAX PERL_UINT_MAX | |
210 | # define U32_MIN PERL_UINT_MIN | |
79072805 | 211 | #else |
27d4fb96 | 212 | # define I32_MAX PERL_LONG_MAX |
213 | # define I32_MIN PERL_LONG_MIN | |
214 | # define U32_MAX PERL_ULONG_MAX | |
215 | # define U32_MIN PERL_ULONG_MIN | |
79072805 LW |
216 | #endif |
217 | ||
5ff3f7a4 GS |
218 | #endif |
219 | ||
247cee9f KW |
220 | /* These C99 typedefs are useful sometimes for, say, loop variables whose |
221 | * maximum values are small, but for which speed trumps size. If we have a C99 | |
222 | * compiler, use that. Otherwise, a plain 'int' should be good enough. | |
223 | * | |
224 | * Restrict these to core for now until we are more certain this is a good | |
225 | * idea. */ | |
226 | #if defined(PERL_CORE) || defined(PERL_EXT) | |
227 | # ifdef I_STDINT | |
228 | typedef int_fast8_t PERL_INT_FAST8_T; | |
229 | typedef uint_fast8_t PERL_UINT_FAST8_T; | |
230 | typedef int_fast16_t PERL_INT_FAST16_T; | |
231 | typedef uint_fast16_t PERL_UINT_FAST16_T; | |
232 | # else | |
233 | typedef int PERL_INT_FAST8_T; | |
234 | typedef unsigned int PERL_UINT_FAST8_T; | |
235 | typedef int PERL_INT_FAST16_T; | |
236 | typedef unsigned int PERL_UINT_FAST16_T; | |
237 | # endif | |
238 | #endif | |
239 | ||
464decb6 KW |
240 | /* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case |
241 | * anyone is grepping for it */ | |
242 | #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */ | |
fc36a67e | 243 | #define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8) |
244 | #define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */ | |
245 | ||
88794300 | 246 | /* Unused by core; should be deprecated */ |
ff68c719 | 247 | #define Ctl(ch) ((ch) & 037) |
8d063cd8 | 248 | |
98fce2a4 KW |
249 | #if defined(PERL_CORE) || defined(PERL_EXT) |
250 | # ifndef MIN | |
251 | # define MIN(a,b) ((a) < (b) ? (a) : (b)) | |
252 | # endif | |
253 | # ifndef MAX | |
254 | # define MAX(a,b) ((a) > (b) ? (a) : (b)) | |
255 | # endif | |
256 | #endif | |
257 | ||
84ff4fa9 KW |
258 | /* Returns a boolean as to whether the input unsigned number is a power of 2 |
259 | * (2**0, 2**1, etc). In other words if it has just a single bit set. | |
260 | * If not, subtracting 1 would leave the uppermost bit set, so the & would | |
261 | * yield non-zero */ | |
262 | #if defined(PERL_CORE) || defined(PERL_EXT) | |
011b1419 | 263 | # define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0) |
84ff4fa9 KW |
264 | #endif |
265 | ||
4650c663 KW |
266 | /* This is a helper macro to avoid preprocessor issues, replaced by nothing |
267 | * unless under DEBUGGING, where it expands to an assert of its argument, | |
268 | * followed by a comma (hence the comma operator). If we just used a straight | |
3e94db23 JH |
269 | * assert(), we would get a comma with nothing before it when not DEBUGGING. |
270 | * | |
271 | * We also use empty definition under Coverity since the __ASSERT__ | |
272 | * checks often check for things that Really Cannot Happen, and Coverity | |
273 | * detects that and gets all excited. */ | |
274 | ||
275 | #if defined(DEBUGGING) && !defined(__COVERITY__) | |
0f092d08 KW |
276 | # define __ASSERT_(statement) assert(statement), |
277 | #else | |
278 | # define __ASSERT_(statement) | |
279 | #endif | |
280 | ||
3fe05580 | 281 | /* |
a4ee4fb5 | 282 | =head1 SV Manipulation Functions |
3fe05580 | 283 | |
1568d13a Z |
284 | =for apidoc Ama|SV*|newSVpvs|"literal string" s |
285 | Like C<newSVpvn>, but takes a literal string instead of a | |
30a15352 | 286 | string/length pair. |
3fe05580 | 287 | |
1568d13a Z |
288 | =for apidoc Ama|SV*|newSVpvs_flags|"literal string" s|U32 flags |
289 | Like C<newSVpvn_flags>, but takes a literal string instead of | |
30a15352 | 290 | a string/length pair. |
84bafc02 | 291 | |
1568d13a Z |
292 | =for apidoc Ama|SV*|newSVpvs_share|"literal string" s |
293 | Like C<newSVpvn_share>, but takes a literal string instead of | |
30a15352 | 294 | a string/length pair and omits the hash parameter. |
3fe05580 | 295 | |
1568d13a Z |
296 | =for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string" s|I32 flags |
297 | Like C<sv_catpvn_flags>, but takes a literal string instead | |
30a15352 | 298 | of a string/length pair. |
9dcc53ea | 299 | |
1568d13a Z |
300 | =for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string" s |
301 | Like C<sv_catpvn_nomg>, but takes a literal string instead of | |
0c395ea5 | 302 | a string/length pair. |
9dcc53ea | 303 | |
1568d13a Z |
304 | =for apidoc Am|void|sv_catpvs|SV* sv|"literal string" s |
305 | Like C<sv_catpvn>, but takes a literal string instead of a | |
0c395ea5 | 306 | string/length pair. |
3fe05580 | 307 | |
1568d13a Z |
308 | =for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string" s |
309 | Like C<sv_catpvn_mg>, but takes a literal string instead of a | |
9dcc53ea Z |
310 | string/length pair. |
311 | ||
1568d13a Z |
312 | =for apidoc Am|void|sv_setpvs|SV* sv|"literal string" s |
313 | Like C<sv_setpvn>, but takes a literal string instead of a | |
0c395ea5 | 314 | string/length pair. |
3fe05580 | 315 | |
1568d13a Z |
316 | =for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string" s |
317 | Like C<sv_setpvn_mg>, but takes a literal string instead of a | |
9dcc53ea Z |
318 | string/length pair. |
319 | ||
78342678 | 320 | =for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string" s |
1568d13a | 321 | Like C<sv_setref_pvn>, but takes a literal string instead of |
0c395ea5 | 322 | a string/length pair. |
9dcc53ea | 323 | |
3fe05580 MHM |
324 | =head1 Memory Management |
325 | ||
1568d13a Z |
326 | =for apidoc Ama|char*|savepvs|"literal string" s |
327 | Like C<savepvn>, but takes a literal string instead of a | |
30a15352 | 328 | string/length pair. |
3fe05580 | 329 | |
1568d13a | 330 | =for apidoc Ama|char*|savesharedpvs|"literal string" s |
9dcc53ea Z |
331 | A version of C<savepvs()> which allocates the duplicate string in memory |
332 | which is shared between threads. | |
333 | ||
3fe05580 MHM |
334 | =head1 GV Functions |
335 | ||
1568d13a Z |
336 | =for apidoc Am|HV*|gv_stashpvs|"literal string" name|I32 create |
337 | Like C<gv_stashpvn>, but takes a literal string instead of a | |
0c395ea5 | 338 | string/length pair. |
3fe05580 MHM |
339 | |
340 | =head1 Hash Manipulation Functions | |
341 | ||
1568d13a Z |
342 | =for apidoc Am|SV**|hv_fetchs|HV* tb|"literal string" key|I32 lval |
343 | Like C<hv_fetch>, but takes a literal string instead of a | |
0c395ea5 | 344 | string/length pair. |
3fe05580 | 345 | |
d106cb46 | 346 | =for apidoc Am|SV**|hv_stores|HV* tb|"literal string" key|SV* val |
1568d13a | 347 | Like C<hv_store>, but takes a literal string instead of a |
0c395ea5 | 348 | string/length pair |
3fe05580 MHM |
349 | and omits the hash parameter. |
350 | ||
510966aa Z |
351 | =head1 Lexer interface |
352 | ||
1568d13a | 353 | =for apidoc Amx|void|lex_stuff_pvs|"literal string" pv|U32 flags |
510966aa | 354 | |
1568d13a | 355 | Like L</lex_stuff_pvn>, but takes a literal string instead of |
0c395ea5 | 356 | a string/length pair. |
510966aa | 357 | |
3fe05580 MHM |
358 | =cut |
359 | */ | |
360 | ||
926bb54c KW |
361 | /* concatenating with "" ensures that only literal strings are accepted as |
362 | * argument */ | |
26bfd924 | 363 | #define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1) |
2efa8cc7 | 364 | |
926bb54c KW |
365 | /* note that STR_WITH_LEN() can't be used as argument to macros or functions |
366 | * that under some configurations might be macros, which means that it requires | |
367 | * the full Perl_xxx(aTHX_ ...) form for any API calls where it's used. | |
ba3a79e7 GA |
368 | */ |
369 | ||
370 | /* STR_WITH_LEN() shortcuts */ | |
371 | #define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str)) | |
84bafc02 NC |
372 | #define newSVpvs_flags(str,flags) \ |
373 | Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags) | |
ba3a79e7 | 374 | #define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0) |
9dcc53ea Z |
375 | #define sv_catpvs_flags(sv, str, flags) \ |
376 | Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags) | |
377 | #define sv_catpvs_nomg(sv, str) \ | |
378 | Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0) | |
379 | #define sv_catpvs(sv, str) \ | |
380 | Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC) | |
381 | #define sv_catpvs_mg(sv, str) \ | |
382 | Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC) | |
3fe05580 | 383 | #define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str)) |
9dcc53ea Z |
384 | #define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str)) |
385 | #define sv_setref_pvs(rv, classname, str) \ | |
386 | Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str)) | |
ba3a79e7 | 387 | #define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str)) |
9dcc53ea Z |
388 | #define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str)) |
389 | #define gv_stashpvs(str, create) \ | |
390 | Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create) | |
391 | #define gv_fetchpvs(namebeg, add, sv_type) \ | |
392 | Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type) | |
393 | #define gv_fetchpvn(namebeg, len, add, sv_type) \ | |
394 | Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type) | |
395 | #define sv_catxmlpvs(dsv, str, utf8) \ | |
396 | Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8) | |
4ac46235 | 397 | |
ba3a79e7 | 398 | |
510966aa Z |
399 | #define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags) |
400 | ||
b96d8cd9 NC |
401 | #define get_cvs(str, flags) \ |
402 | Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags)) | |
5c1737d1 | 403 | |
954c1994 | 404 | /* |
ccfc67b7 JH |
405 | =head1 Miscellaneous Functions |
406 | ||
954c1994 | 407 | =for apidoc Am|bool|strNE|char* s1|char* s2 |
dc6b0978 KW |
408 | Test two C<NUL>-terminated strings to see if they are different. Returns true |
409 | or false. | |
954c1994 GS |
410 | |
411 | =for apidoc Am|bool|strEQ|char* s1|char* s2 | |
dc6b0978 KW |
412 | Test two C<NUL>-terminated strings to see if they are equal. Returns true or |
413 | false. | |
954c1994 GS |
414 | |
415 | =for apidoc Am|bool|strLT|char* s1|char* s2 | |
dc6b0978 KW |
416 | Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the |
417 | second, C<s2>. Returns true or false. | |
954c1994 GS |
418 | |
419 | =for apidoc Am|bool|strLE|char* s1|char* s2 | |
dc6b0978 KW |
420 | Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or |
421 | equal to the second, C<s2>. Returns true or false. | |
954c1994 GS |
422 | |
423 | =for apidoc Am|bool|strGT|char* s1|char* s2 | |
dc6b0978 KW |
424 | Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than |
425 | the second, C<s2>. Returns true or false. | |
954c1994 GS |
426 | |
427 | =for apidoc Am|bool|strGE|char* s1|char* s2 | |
dc6b0978 KW |
428 | Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than |
429 | or equal to the second, C<s2>. Returns true or false. | |
954c1994 GS |
430 | |
431 | =for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len | |
dc6b0978 KW |
432 | Test two C<NUL>-terminated strings to see if they are different. The C<len> |
433 | parameter indicates the number of bytes to compare. Returns true or false. (A | |
954c1994 GS |
434 | wrapper for C<strncmp>). |
435 | ||
436 | =for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len | |
dc6b0978 KW |
437 | Test two C<NUL>-terminated strings to see if they are equal. The C<len> |
438 | parameter indicates the number of bytes to compare. Returns true or false. (A | |
439 | wrapper for C<strncmp>). | |
954c1994 | 440 | |
bd18bd40 KW |
441 | =for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len |
442 | Test two buffers (which may contain embedded C<NUL> characters, to see if they | |
443 | are equal. The C<len> parameter indicates the number of bytes to compare. | |
444 | Returns zero if equal, or non-zero if non-equal. | |
445 | ||
446 | =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len | |
447 | Test two buffers (which may contain embedded C<NUL> characters, to see if they | |
448 | are not equal. The C<len> parameter indicates the number of bytes to compare. | |
449 | Returns zero if non-equal, or non-zero if equal. | |
450 | ||
954c1994 | 451 | =cut |
fc169e00 KW |
452 | |
453 | New macros should use the following conventions for their names (which are | |
454 | based on the underlying C library functions): | |
455 | ||
456 | (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s? | |
457 | ||
458 | Each has two main parameters, string-like operands that are compared | |
459 | against each other, as specified by the macro name. Some macros may | |
460 | additionally have one or potentially even two length parameters. If a length | |
461 | parameter applies to both string parameters, it will be positioned third; | |
462 | otherwise any length parameter immediately follows the string parameter it | |
463 | applies to. | |
464 | ||
465 | If the prefix to the name is 'str', the string parameter is a pointer to a C | |
466 | language string. Such a string does not contain embedded NUL bytes; its | |
467 | length may be unknown, but can be calculated by C<strlen()>, since it is | |
468 | terminated by a NUL, which isn't included in its length. | |
469 | ||
470 | The optional 'n' following 'str' means that that there is a third parameter, | |
471 | giving the maximum number of bytes to look at in each string. Even if both | |
472 | strings are longer than the length parameter, those extra bytes will be | |
473 | unexamined. | |
474 | ||
475 | The 's' suffix means that the 2nd byte string parameter is a literal C | |
476 | double-quoted string. Its length will automatically be calculated by the | |
477 | macro, so no length parameter will ever be needed for it. | |
478 | ||
479 | If the prefix is 'mem', the string parameters don't have to be C strings; | |
480 | they may contain embedded NUL bytes, do not necessarily have a terminating | |
481 | NUL, and their lengths can be known only through other means, which in | |
482 | practice are additional parameter(s) passed to the function. All 'mem' | |
483 | functions have at least one length parameter. Barring any 'l' or 's' suffix, | |
484 | there is a single length parameter, in position 3, which applies to both | |
485 | string parameters. The 's' suffix means, as described above, that the 2nd | |
486 | string is a literal double-quoted C string (hence its length is calculated by | |
487 | the macro, and the length parameter to the function applies just to the first | |
488 | string parameter, and hence is positioned just after it). An 'l' suffix | |
489 | means that the 2nd string parameter has its own length parameter, and the | |
490 | signature will look like memFOOl(s1, l1, s2, l2). | |
491 | ||
492 | BEGIN (and END) are for testing if the 2nd string is an initial (or final) | |
493 | substring of the 1st string. 'P' if present indicates that the substring | |
494 | must be a "proper" one in tha mathematical sense that the first one must be | |
495 | strictly larger than the 2nd. | |
496 | ||
954c1994 GS |
497 | */ |
498 | ||
62946e08 | 499 | |
75400963 KW |
500 | #define strNE(s1,s2) (strcmp(s1,s2) != 0) |
501 | #define strEQ(s1,s2) (strcmp(s1,s2) == 0) | |
8d063cd8 LW |
502 | #define strLT(s1,s2) (strcmp(s1,s2) < 0) |
503 | #define strLE(s1,s2) (strcmp(s1,s2) <= 0) | |
504 | #define strGT(s1,s2) (strcmp(s1,s2) > 0) | |
505 | #define strGE(s1,s2) (strcmp(s1,s2) >= 0) | |
62946e08 | 506 | |
75400963 KW |
507 | #define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0) |
508 | #define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0) | |
378cc40b | 509 | |
9d3980bc KW |
510 | #define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0) |
511 | #define memNE(s1,s2,l) (! memEQ(s1,s2,l)) | |
36477c24 | 512 | |
085b7534 | 513 | /* memEQ and memNE where second comparand is a string constant */ |
568a785a | 514 | #define memEQs(s1, l, s2) \ |
777fa2cb | 515 | (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1))) |
5f50c6c9 | 516 | #define memNEs(s1, l, s2) (! memEQs(s1, l, s2)) |
568a785a | 517 | |
fdbb9a7c KW |
518 | /* Keep these private until we decide it was a good idea */ |
519 | #if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX) | |
520 | ||
521 | #define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0) | |
522 | ||
bdb7e3f0 | 523 | #define memBEGINs(s1, l, s2) \ |
30a6480c | 524 | ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \ |
bdb7e3f0 | 525 | && memEQ(s1, "" s2 "", sizeof(s2)-1)) |
de627158 | 526 | #define memBEGINPs(s1, l, s2) \ |
30a6480c | 527 | ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \ |
de627158 | 528 | && memEQ(s1, "" s2 "", sizeof(s2)-1)) |
bdb7e3f0 | 529 | #define memENDs(s1, l, s2) \ |
30a6480c | 530 | ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \ |
bdb7e3f0 | 531 | && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1)) |
b80f8424 | 532 | #define memENDPs(s1, l, s2) \ |
30a6480c | 533 | ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \ |
b80f8424 | 534 | && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1)) |
fdbb9a7c | 535 | #endif /* End of making macros private */ |
bdb7e3f0 | 536 | |
062b6850 KW |
537 | #define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0) |
538 | #define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0) | |
539 | #define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0) | |
540 | #define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0) | |
541 | ||
bbce6d69 | 542 | /* |
543 | * Character classes. | |
544 | * | |
545 | * Unfortunately, the introduction of locales means that we | |
546 | * can't trust isupper(), etc. to tell the truth. And when | |
547 | * it comes to /\w+/ with tainting enabled, we *must* be able | |
548 | * to trust our character classes. | |
549 | * | |
550 | * Therefore, the default tests in the text of Perl will be | |
551 | * independent of locale. Any code that wants to depend on | |
552 | * the current locale will use the tests that begin with "lc". | |
553 | */ | |
554 | ||
2304df62 AD |
555 | #ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */ |
556 | # ifndef CTYPE256 | |
557 | # define CTYPE256 | |
558 | # endif | |
559 | #endif | |
560 | ||
954c1994 | 561 | /* |
ccfc67b7 | 562 | |
dcccc8ff | 563 | =head1 Character classification |
243effed KW |
564 | This section is about functions (really macros) that classify characters |
565 | into types, such as punctuation versus alphabetic, etc. Most of these are | |
566 | analogous to regular expression character classes. (See | |
567 | L<perlrecharclass/POSIX Character Classes>.) There are several variants for | |
568 | each class. (Not all macros have all variants; each item below lists the | |
569 | ones valid for it.) None are affected by C<use bytes>, and only the ones | |
570 | with C<LC> in the name are affected by the current locale. | |
571 | ||
572 | The base function, e.g., C<isALPHA()>, takes an octet (either a C<char> or a | |
573 | C<U8>) as input and returns a boolean as to whether or not the character | |
6aff1f14 KW |
574 | represented by that octet is (or on non-ASCII platforms, corresponds to) an |
575 | ASCII character in the named class based on platform, Unicode, and Perl rules. | |
576 | If the input is a number that doesn't fit in an octet, FALSE is returned. | |
243effed | 577 | |
c98722a4 | 578 | Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function |
550da823 KW |
579 | with no suffix C<"_A">. This variant is used to emphasize by its name that |
580 | only ASCII-range characters can return TRUE. | |
4b9734bf | 581 | |
d60679e1 | 582 | Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set |
4b9734bf KW |
583 | onto the platform. That is, the code points that are ASCII are unaffected, |
584 | since ASCII is a subset of Latin-1. But the non-ASCII code points are treated | |
585 | as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return | |
586 | true when called with the code point 0xDF, which is a word character in both | |
4650c663 | 587 | ASCII and EBCDIC (though it represents different characters in each). |
243effed | 588 | |
c98722a4 | 589 | Variant C<isI<FOO>_uvchr> is like the C<isI<FOO>_L1> variant, but accepts any UV code |
243effed KW |
590 | point as input. If the code point is larger than 255, Unicode rules are used |
591 | to determine if it is in the character class. For example, | |
d0da05db | 592 | C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A |
6aff1f14 | 593 | WITH MACRON in Unicode, and is a word character. |
243effed | 594 | |
c98722a4 | 595 | Variant C<isI<FOO>_utf8_safe> is like C<isI<FOO>_uvchr>, but is used for UTF-8 |
da8c1a98 KW |
596 | encoded strings. Each call classifies one character, even if the string |
597 | contains many. This variant takes two parameters. The first, C<p>, is a | |
598 | pointer to the first byte of the character to be classified. (Recall that it | |
599 | may take more than one byte to represent a character in UTF-8 strings.) The | |
600 | second parameter, C<e>, points to anywhere in the string beyond the first | |
601 | character, up to one byte past the end of the entire string. The suffix | |
602 | C<_safe> in the function's name indicates that it will not attempt to read | |
603 | beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this | |
604 | is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input | |
605 | character is malformed in some way, the program may croak, or the function may | |
606 | return FALSE, at the discretion of the implementation, and subject to change in | |
607 | future releases. | |
608 | ||
c98722a4 | 609 | Variant C<isI<FOO>_utf8> is like C<isI<FOO>_utf8_safe>, but takes just a single |
da8c1a98 | 610 | parameter, C<p>, which has the same meaning as the corresponding parameter does |
c98722a4 | 611 | in C<isI<FOO>_utf8_safe>. The function therefore can't check if it is reading |
34aeb2e9 | 612 | beyond the end of the string. Starting in Perl v5.30, it will take a second |
c98722a4 | 613 | parameter, becoming a synonym for C<isI<FOO>_utf8_safe>. At that time every |
34aeb2e9 | 614 | program that uses it will have to be changed to successfully compile. In the |
c98722a4 | 615 | meantime, the first runtime call to C<isI<FOO>_utf8> from each call point in the |
34aeb2e9 | 616 | program will raise a deprecation warning, enabled by default. You can convert |
c98722a4 | 617 | your program now to use C<isI<FOO>_utf8_safe>, and avoid the warnings, and get an |
34aeb2e9 KW |
618 | extra measure of protection, or you can wait until v5.30, when you'll be forced |
619 | to add the C<e> parameter. | |
243effed | 620 | |
c98722a4 | 621 | Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants, but the |
31f05a37 KW |
622 | result is based on the current locale, which is what C<LC> in the name stands |
623 | for. If Perl can determine that the current locale is a UTF-8 locale, it uses | |
624 | the published Unicode rules; otherwise, it uses the C library function that | |
625 | gives the named classification. For example, C<isDIGIT_LC()> when not in a | |
626 | UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always | |
1a83413c KW |
627 | returned if the input won't fit into an octet. On some platforms where the C |
628 | library function is known to be defective, Perl changes its result to follow | |
629 | the POSIX standard's rules. | |
243effed | 630 | |
c98722a4 KW |
631 | Variant C<isI<FOO>_LC_uvchr> is like C<isI<FOO>_LC>, but is defined on any UV. It |
632 | returns the same as C<isI<FOO>_LC> for input code points less than 256, and | |
243effed KW |
633 | returns the hard-coded, not-affected-by-locale, Unicode results for larger ones. |
634 | ||
c98722a4 | 635 | Variant C<isI<FOO>_LC_utf8_safe> is like C<isI<FOO>_LC_uvchr>, but is used for UTF-8 |
da8c1a98 KW |
636 | encoded strings. Each call classifies one character, even if the string |
637 | contains many. This variant takes two parameters. The first, C<p>, is a | |
638 | pointer to the first byte of the character to be classified. (Recall that it | |
639 | may take more than one byte to represent a character in UTF-8 strings.) The | |
640 | second parameter, C<e>, points to anywhere in the string beyond the first | |
641 | character, up to one byte past the end of the entire string. The suffix | |
642 | C<_safe> in the function's name indicates that it will not attempt to read | |
643 | beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this | |
644 | is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input | |
645 | character is malformed in some way, the program may croak, or the function may | |
646 | return FALSE, at the discretion of the implementation, and subject to change in | |
647 | future releases. | |
648 | ||
c98722a4 | 649 | Variant C<isI<FOO>_LC_utf8> is like C<isI<FOO>_LC_utf8_safe>, but takes just a single |
da8c1a98 | 650 | parameter, C<p>, which has the same meaning as the corresponding parameter does |
c98722a4 | 651 | in C<isI<FOO>_LC_utf8_safe>. The function therefore can't check if it is reading |
34aeb2e9 | 652 | beyond the end of the string. Starting in Perl v5.30, it will take a second |
c98722a4 | 653 | parameter, becoming a synonym for C<isI<FOO>_LC_utf8_safe>. At that time every |
34aeb2e9 | 654 | program that uses it will have to be changed to successfully compile. In the |
c98722a4 | 655 | meantime, the first runtime call to C<isI<FOO>_LC_utf8> from each call point in |
34aeb2e9 | 656 | the program will raise a deprecation warning, enabled by default. You can |
c98722a4 | 657 | convert your program now to use C<isI<FOO>_LC_utf8_safe>, and avoid the warnings, |
34aeb2e9 KW |
658 | and get an extra measure of protection, or you can wait until v5.30, when |
659 | you'll be forced to add the C<e> parameter. | |
ccfc67b7 | 660 | |
954c1994 | 661 | =for apidoc Am|bool|isALPHA|char ch |
2787a470 | 662 | Returns a boolean indicating whether the specified character is an |
6aff1f14 | 663 | alphabetic character, analogous to C<m/[[:alpha:]]/>. |
dcccc8ff KW |
664 | See the L<top of this section|/Character classification> for an explanation of |
665 | variants | |
da8c1a98 KW |
666 | C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8_safe>, |
667 | C<isALPHA_LC>, C<isALPHA_LC_uvchr>, and C<isALPHA_LC_utf8_safe>. | |
8a58bdcf | 668 | |
15861f94 KW |
669 | =for apidoc Am|bool|isALPHANUMERIC|char ch |
670 | Returns a boolean indicating whether the specified character is a either an | |
6aff1f14 | 671 | alphabetic character or decimal digit, analogous to C<m/[[:alnum:]]/>. |
dcccc8ff KW |
672 | See the L<top of this section|/Character classification> for an explanation of |
673 | variants | |
d0da05db | 674 | C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>, |
da8c1a98 KW |
675 | C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>, C<isALPHANUMERIC_LC_uvchr>, |
676 | and C<isALPHANUMERIC_LC_utf8_safe>. | |
15861f94 | 677 | |
8a58bdcf KW |
678 | =for apidoc Am|bool|isASCII|char ch |
679 | Returns a boolean indicating whether the specified character is one of the 128 | |
243effed | 680 | characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>. |
e5ad6aba | 681 | On non-ASCII platforms, it returns TRUE iff this |
8a58bdcf KW |
682 | character corresponds to an ASCII character. Variants C<isASCII_A()> and |
683 | C<isASCII_L1()> are identical to C<isASCII()>. | |
dcccc8ff KW |
684 | See the L<top of this section|/Character classification> for an explanation of |
685 | variants | |
da8c1a98 KW |
686 | C<isASCII_uvchr>, C<isASCII_utf8_safe>, C<isASCII_LC>, C<isASCII_LC_uvchr>, and |
687 | C<isASCII_LC_utf8_safe>. Note, however, that some platforms do not have the C | |
243effed KW |
688 | library routine C<isascii()>. In these cases, the variants whose names contain |
689 | C<LC> are the same as the corresponding ones without. | |
690 | ||
d98532ea KW |
691 | Also note, that because all ASCII characters are UTF-8 invariant (meaning they |
692 | have the exact same representation (always a single byte) whether encoded in | |
693 | UTF-8 or not), C<isASCII> will give the correct results when called with any | |
da8c1a98 KW |
694 | byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8_safe> |
695 | will work properly on any string encoded or not in UTF-8. | |
d98532ea | 696 | |
243effed KW |
697 | =for apidoc Am|bool|isBLANK|char ch |
698 | Returns a boolean indicating whether the specified character is a | |
6aff1f14 | 699 | character considered to be a blank, analogous to C<m/[[:blank:]]/>. |
dcccc8ff KW |
700 | See the L<top of this section|/Character classification> for an explanation of |
701 | variants | |
da8c1a98 KW |
702 | C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8_safe>, |
703 | C<isBLANK_LC>, C<isBLANK_LC_uvchr>, and C<isBLANK_LC_utf8_safe>. Note, | |
704 | however, that some platforms do not have the C library routine | |
705 | C<isblank()>. In these cases, the variants whose names contain C<LC> are | |
706 | the same as the corresponding ones without. | |
243effed KW |
707 | |
708 | =for apidoc Am|bool|isCNTRL|char ch | |
709 | Returns a boolean indicating whether the specified character is a | |
6aff1f14 | 710 | control character, analogous to C<m/[[:cntrl:]]/>. |
dcccc8ff KW |
711 | See the L<top of this section|/Character classification> for an explanation of |
712 | variants | |
da8c1a98 KW |
713 | C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8_safe>, |
714 | C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, and C<isCNTRL_LC_utf8_safe> On EBCDIC | |
715 | platforms, you almost always want to use the C<isCNTRL_L1> variant. | |
954c1994 | 716 | |
954c1994 | 717 | =for apidoc Am|bool|isDIGIT|char ch |
2787a470 | 718 | Returns a boolean indicating whether the specified character is a |
6aff1f14 | 719 | digit, analogous to C<m/[[:digit:]]/>. |
8a58bdcf | 720 | Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>. |
dcccc8ff KW |
721 | See the L<top of this section|/Character classification> for an explanation of |
722 | variants | |
da8c1a98 KW |
723 | C<isDIGIT_uvchr>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>, C<isDIGIT_LC_uvchr>, and |
724 | C<isDIGIT_LC_utf8_safe>. | |
243effed KW |
725 | |
726 | =for apidoc Am|bool|isGRAPH|char ch | |
727 | Returns a boolean indicating whether the specified character is a | |
6aff1f14 | 728 | graphic character, analogous to C<m/[[:graph:]]/>. |
dcccc8ff | 729 | See the L<top of this section|/Character classification> for an explanation of |
da8c1a98 KW |
730 | variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8_safe>, |
731 | C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>, and C<isGRAPH_LC_utf8_safe>. | |
954c1994 | 732 | |
0c82b6df | 733 | =for apidoc Am|bool|isLOWER|char ch |
2787a470 | 734 | Returns a boolean indicating whether the specified character is a |
6aff1f14 | 735 | lowercase character, analogous to C<m/[[:lower:]]/>. |
dcccc8ff KW |
736 | See the L<top of this section|/Character classification> for an explanation of |
737 | variants | |
da8c1a98 KW |
738 | C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8_safe>, |
739 | C<isLOWER_LC>, C<isLOWER_LC_uvchr>, and C<isLOWER_LC_utf8_safe>. | |
0c82b6df | 740 | |
c99e91e9 | 741 | =for apidoc Am|bool|isOCTAL|char ch |
2787a470 | 742 | Returns a boolean indicating whether the specified character is an |
6aff1f14 | 743 | octal digit, [0-7]. |
243effed KW |
744 | The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to |
745 | C<isOCTAL>. | |
746 | ||
747 | =for apidoc Am|bool|isPUNCT|char ch | |
748 | Returns a boolean indicating whether the specified character is a | |
6aff1f14 KW |
749 | punctuation character, analogous to C<m/[[:punct:]]/>. |
750 | Note that the definition of what is punctuation isn't as | |
243effed KW |
751 | straightforward as one might desire. See L<perlrecharclass/POSIX Character |
752 | Classes> for details. | |
dcccc8ff | 753 | See the L<top of this section|/Character classification> for an explanation of |
da8c1a98 KW |
754 | variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8_safe>, |
755 | C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, and C<isPUNCT_LC_utf8_safe>. | |
c99e91e9 | 756 | |
0c82b6df | 757 | =for apidoc Am|bool|isSPACE|char ch |
2787a470 | 758 | Returns a boolean indicating whether the specified character is a |
6aff1f14 | 759 | whitespace character. This is analogous |
398d098a | 760 | to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18 |
779cf272 | 761 | this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the |
398d098a KW |
762 | locale forms of this macro (the ones with C<LC> in their names) matched |
763 | precisely what C<m/[[:space:]]/> does. In those releases, the only difference, | |
764 | in the non-locale variants, was that C<isSPACE()> did not match a vertical tab. | |
765 | (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.) | |
dcccc8ff KW |
766 | See the L<top of this section|/Character classification> for an explanation of |
767 | variants | |
da8c1a98 KW |
768 | C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8_safe>, |
769 | C<isSPACE_LC>, C<isSPACE_LC_uvchr>, and C<isSPACE_LC_utf8_safe>. | |
0c82b6df | 770 | |
398d098a KW |
771 | =for apidoc Am|bool|isPSXSPC|char ch |
772 | (short for Posix Space) | |
779cf272 KW |
773 | Starting in 5.18, this is identical in all its forms to the |
774 | corresponding C<isSPACE()> macros. | |
398d098a KW |
775 | The locale forms of this macro are identical to their corresponding |
776 | C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the | |
777 | non-locale forms differ from their C<isSPACE()> forms only in that the | |
778 | C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do. | |
779 | Otherwise they are identical. Thus this macro is analogous to what | |
780 | C<m/[[:space:]]/> matches in a regular expression. | |
dcccc8ff | 781 | See the L<top of this section|/Character classification> for an explanation of |
da8c1a98 KW |
782 | variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8_safe>, |
783 | C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>, and C<isPSXSPC_LC_utf8_safe>. | |
398d098a | 784 | |
954c1994 | 785 | =for apidoc Am|bool|isUPPER|char ch |
2787a470 | 786 | Returns a boolean indicating whether the specified character is an |
6aff1f14 | 787 | uppercase character, analogous to C<m/[[:upper:]]/>. |
dcccc8ff | 788 | See the L<top of this section|/Character classification> for an explanation of |
da8c1a98 KW |
789 | variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8_safe>, |
790 | C<isUPPER_LC>, C<isUPPER_LC_uvchr>, and C<isUPPER_LC_utf8_safe>. | |
954c1994 | 791 | |
243effed | 792 | =for apidoc Am|bool|isPRINT|char ch |
8eea39dd | 793 | Returns a boolean indicating whether the specified character is a |
6aff1f14 | 794 | printable character, analogous to C<m/[[:print:]]/>. |
dcccc8ff KW |
795 | See the L<top of this section|/Character classification> for an explanation of |
796 | variants | |
da8c1a98 KW |
797 | C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8_safe>, |
798 | C<isPRINT_LC>, C<isPRINT_LC_uvchr>, and C<isPRINT_LC_utf8_safe>. | |
243effed KW |
799 | |
800 | =for apidoc Am|bool|isWORDCHAR|char ch | |
801 | Returns a boolean indicating whether the specified character is a character | |
802 | that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match | |
803 | in a regular expression. A word character is an alphabetic character, a | |
804 | decimal digit, a connecting punctuation character (such as an underscore), or | |
805 | a "mark" character that attaches to one of those (like some sort of accent). | |
806 | C<isALNUM()> is a synonym provided for backward compatibility, even though a | |
807 | word character includes more than the standard C language meaning of | |
808 | alphanumeric. | |
dcccc8ff | 809 | See the L<top of this section|/Character classification> for an explanation of |
da8c1a98 KW |
810 | variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, and |
811 | C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, and | |
812 | C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally | |
813 | include the platform's native underscore. | |
8a58bdcf KW |
814 | |
815 | =for apidoc Am|bool|isXDIGIT|char ch | |
816 | Returns a boolean indicating whether the specified character is a hexadecimal | |
243effed KW |
817 | digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()> |
818 | and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>. | |
dcccc8ff KW |
819 | See the L<top of this section|/Character classification> for an explanation of |
820 | variants | |
da8c1a98 KW |
821 | C<isXDIGIT_uvchr>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>, C<isXDIGIT_LC_uvchr>, |
822 | and C<isXDIGIT_LC_utf8_safe>. | |
243effed | 823 | |
3c3ecf18 KW |
824 | =for apidoc Am|bool|isIDFIRST|char ch |
825 | Returns a boolean indicating whether the specified character can be the first | |
826 | character of an identifier. This is very close to, but not quite the same as | |
827 | the official Unicode property C<XID_Start>. The difference is that this | |
828 | returns true only if the input character also matches L</isWORDCHAR>. | |
dcccc8ff KW |
829 | See the L<top of this section|/Character classification> for an explanation of |
830 | variants | |
da8c1a98 KW |
831 | C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8_safe>, |
832 | C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>, and C<isIDFIRST_LC_utf8_safe>. | |
3c3ecf18 KW |
833 | |
834 | =for apidoc Am|bool|isIDCONT|char ch | |
835 | Returns a boolean indicating whether the specified character can be the | |
836 | second or succeeding character of an identifier. This is very close to, but | |
837 | not quite the same as the official Unicode property C<XID_Continue>. The | |
838 | difference is that this returns true only if the input character also matches | |
dcccc8ff KW |
839 | L</isWORDCHAR>. See the L<top of this section|/Character classification> for |
840 | an | |
d0da05db | 841 | explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>, |
da8c1a98 KW |
842 | C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>, and |
843 | C<isIDCONT_LC_utf8_safe>. | |
3c3ecf18 | 844 | |
243effed | 845 | =head1 Miscellaneous Functions |
8eea39dd | 846 | |
95a59cab | 847 | =for apidoc Am|U8|READ_XDIGIT|char str* |
243effed | 848 | Returns the value of an ASCII-range hex digit and advances the string pointer. |
95a59cab YO |
849 | Behaviour is only well defined when isXDIGIT(*str) is true. |
850 | ||
e7c1e6c1 | 851 | =head1 Character case changing |
21da7284 KW |
852 | Perl uses "full" Unicode case mappings. This means that converting a single |
853 | character to another case may result in a sequence of more than one character. | |
854 | For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two | |
855 | character sequence C<SS>. This presents some complications The lowercase of | |
856 | all characters in the range 0..255 is a single character, and thus | |
857 | C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't | |
858 | return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has | |
859 | an API that does allow every possible legal result to be returned.) Likewise | |
860 | no other function that is crippled by not being able to give the correct | |
861 | results for the full range of possible inputs has been implemented here. | |
e7c1e6c1 | 862 | |
1f607577 KW |
863 | =for apidoc Am|U8|toUPPER|U8 ch |
864 | Converts the specified character to uppercase. If the input is anything but an | |
865 | ASCII lowercase character, that input character itself is returned. Variant | |
c753c8d3 | 866 | C<toUPPER_A> is equivalent. |
954c1994 | 867 | |
d0da05db KW |
868 | =for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp |
869 | Converts the code point C<cp> to its uppercase version, and | |
870 | stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code | |
871 | point is interpreted as native if less than 256; otherwise as Unicode. Note | |
1f607577 KW |
872 | that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1> |
873 | bytes since the uppercase version may be longer than the original character. | |
874 | ||
875 | The first code point of the uppercased version is returned | |
21da7284 KW |
876 | (but note, as explained at L<the top of this section|/Character case |
877 | changing>, that there may be more.) | |
1f607577 | 878 | |
a239b1e2 KW |
879 | =for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp |
880 | Converts the first UTF-8 encoded character in the sequence starting at C<p> and | |
881 | extending no further than S<C<e - 1>> to its uppercase version, and | |
1f607577 KW |
882 | stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note |
883 | that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1> | |
884 | bytes since the uppercase version may be longer than the original character. | |
885 | ||
886 | The first code point of the uppercased version is returned | |
21da7284 KW |
887 | (but note, as explained at L<the top of this section|/Character case |
888 | changing>, that there may be more). | |
1f607577 | 889 | |
a239b1e2 KW |
890 | The suffix C<_safe> in the function's name indicates that it will not attempt |
891 | to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is | |
892 | true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the | |
893 | input character is malformed in some way, the program may croak, or the | |
894 | function may return the REPLACEMENT CHARACTER, at the discretion of the | |
895 | implementation, and subject to change in future releases. | |
896 | ||
897 | =for apidoc Am|UV|toUPPER_utf8|U8* p|U8* s|STRLEN* lenp | |
898 | This is like C<L</toUPPER_utf8_safe>>, but doesn't have the C<e> | |
899 | parameter The function therefore can't check if it is reading | |
607313a1 KW |
900 | beyond the end of the string. Starting in Perl v5.30, it will take the C<e> |
901 | parameter, becoming a synonym for C<toUPPER_utf8_safe>. At that time every | |
902 | program that uses it will have to be changed to successfully compile. In the | |
903 | meantime, the first runtime call to C<toUPPER_utf8> from each call point in the | |
904 | program will raise a deprecation warning, enabled by default. You can convert | |
905 | your program now to use C<toUPPER_utf8_safe>, and avoid the warnings, and get an | |
906 | extra measure of protection, or you can wait until v5.30, when you'll be forced | |
907 | to add the C<e> parameter. | |
1f607577 | 908 | |
25200305 KW |
909 | =for apidoc Am|U8|toFOLD|U8 ch |
910 | Converts the specified character to foldcase. If the input is anything but an | |
911 | ASCII uppercase character, that input character itself is returned. Variant | |
912 | C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full | |
d0da05db | 913 | Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.) |
25200305 | 914 | |
d0da05db KW |
915 | =for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp |
916 | Converts the code point C<cp> to its foldcase version, and | |
917 | stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code | |
918 | point is interpreted as native if less than 256; otherwise as Unicode. Note | |
1f607577 KW |
919 | that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1> |
920 | bytes since the foldcase version may be longer than the original character. | |
921 | ||
922 | The first code point of the foldcased version is returned | |
21da7284 KW |
923 | (but note, as explained at L<the top of this section|/Character case |
924 | changing>, that there may be more). | |
1f607577 | 925 | |
a239b1e2 KW |
926 | =for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp |
927 | Converts the first UTF-8 encoded character in the sequence starting at C<p> and | |
928 | extending no further than S<C<e - 1>> to its foldcase version, and | |
1f607577 KW |
929 | stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note |
930 | that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1> | |
931 | bytes since the foldcase version may be longer than the original character. | |
932 | ||
933 | The first code point of the foldcased version is returned | |
21da7284 KW |
934 | (but note, as explained at L<the top of this section|/Character case |
935 | changing>, that there may be more). | |
1f607577 | 936 | |
a239b1e2 KW |
937 | The suffix C<_safe> in the function's name indicates that it will not attempt |
938 | to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is | |
939 | true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the | |
940 | input character is malformed in some way, the program may croak, or the | |
941 | function may return the REPLACEMENT CHARACTER, at the discretion of the | |
942 | implementation, and subject to change in future releases. | |
943 | ||
944 | =for apidoc Am|UV|toFOLD_utf8|U8* p|U8* s|STRLEN* lenp | |
945 | This is like C<L</toFOLD_utf8_safe>>, but doesn't have the C<e> | |
946 | parameter The function therefore can't check if it is reading | |
607313a1 KW |
947 | beyond the end of the string. Starting in Perl v5.30, it will take the C<e> |
948 | parameter, becoming a synonym for C<toFOLD_utf8_safe>. At that time every | |
949 | program that uses it will have to be changed to successfully compile. In the | |
950 | meantime, the first runtime call to C<toFOLD_utf8> from each call point in the | |
951 | program will raise a deprecation warning, enabled by default. You can convert | |
952 | your program now to use C<toFOLD_utf8_safe>, and avoid the warnings, and get an | |
953 | extra measure of protection, or you can wait until v5.30, when you'll be forced | |
954 | to add the C<e> parameter. | |
1f607577 KW |
955 | |
956 | =for apidoc Am|U8|toLOWER|U8 ch | |
957 | Converts the specified character to lowercase. If the input is anything but an | |
958 | ASCII uppercase character, that input character itself is returned. Variant | |
c753c8d3 | 959 | C<toLOWER_A> is equivalent. |
954c1994 | 960 | |
1f607577 | 961 | =for apidoc Am|U8|toLOWER_L1|U8 ch |
b7d90381 KW |
962 | Converts the specified Latin1 character to lowercase. The results are |
963 | undefined if the input doesn't fit in a byte. | |
1f607577 KW |
964 | |
965 | =for apidoc Am|U8|toLOWER_LC|U8 ch | |
966 | Converts the specified character to lowercase using the current locale's rules, | |
967 | if possible; otherwise returns the input character itself. | |
968 | ||
d0da05db KW |
969 | =for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp |
970 | Converts the code point C<cp> to its lowercase version, and | |
971 | stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code | |
972 | point is interpreted as native if less than 256; otherwise as Unicode. Note | |
1f607577 KW |
973 | that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1> |
974 | bytes since the lowercase version may be longer than the original character. | |
975 | ||
976 | The first code point of the lowercased version is returned | |
21da7284 KW |
977 | (but note, as explained at L<the top of this section|/Character case |
978 | changing>, that there may be more). | |
1f607577 | 979 | |
a239b1e2 KW |
980 | |
981 | =for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp | |
982 | Converts the first UTF-8 encoded character in the sequence starting at C<p> and | |
983 | extending no further than S<C<e - 1>> to its lowercase version, and | |
1f607577 KW |
984 | stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note |
985 | that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1> | |
986 | bytes since the lowercase version may be longer than the original character. | |
987 | ||
988 | The first code point of the lowercased version is returned | |
21da7284 KW |
989 | (but note, as explained at L<the top of this section|/Character case |
990 | changing>, that there may be more). | |
1f607577 | 991 | |
a239b1e2 KW |
992 | The suffix C<_safe> in the function's name indicates that it will not attempt |
993 | to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is | |
994 | true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the | |
995 | input character is malformed in some way, the program may croak, or the | |
996 | function may return the REPLACEMENT CHARACTER, at the discretion of the | |
997 | implementation, and subject to change in future releases. | |
998 | ||
999 | =for apidoc Am|UV|toLOWER_utf8|U8* p|U8* s|STRLEN* lenp | |
1000 | This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e> | |
1001 | parameter The function therefore can't check if it is reading | |
607313a1 KW |
1002 | beyond the end of the string. Starting in Perl v5.30, it will take the C<e> |
1003 | parameter, becoming a synonym for C<toLOWER_utf8_safe>. At that time every | |
1004 | program that uses it will have to be changed to successfully compile. In the | |
1005 | meantime, the first runtime call to C<toLOWER_utf8> from each call point in the | |
1006 | program will raise a deprecation warning, enabled by default. You can convert | |
1007 | your program now to use C<toLOWER_utf8_safe>, and avoid the warnings, and get an | |
1008 | extra measure of protection, or you can wait until v5.30, when you'll be forced | |
1009 | to add the C<e> parameter. | |
1f607577 | 1010 | |
25200305 KW |
1011 | =for apidoc Am|U8|toTITLE|U8 ch |
1012 | Converts the specified character to titlecase. If the input is anything but an | |
1013 | ASCII lowercase character, that input character itself is returned. Variant | |
b7d90381 | 1014 | C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1 |
d0da05db | 1015 | range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is |
b7d90381 | 1016 | not a concept used in locale handling, so there is no functionality for that.) |
25200305 | 1017 | |
d0da05db KW |
1018 | =for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp |
1019 | Converts the code point C<cp> to its titlecase version, and | |
1020 | stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code | |
1021 | point is interpreted as native if less than 256; otherwise as Unicode. Note | |
1f607577 KW |
1022 | that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1> |
1023 | bytes since the titlecase version may be longer than the original character. | |
1024 | ||
1025 | The first code point of the titlecased version is returned | |
21da7284 KW |
1026 | (but note, as explained at L<the top of this section|/Character case |
1027 | changing>, that there may be more). | |
1f607577 | 1028 | |
a239b1e2 KW |
1029 | =for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp |
1030 | Converts the first UTF-8 encoded character in the sequence starting at C<p> and | |
1031 | extending no further than S<C<e - 1>> to its titlecase version, and | |
1f607577 KW |
1032 | stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note |
1033 | that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1> | |
1034 | bytes since the titlecase version may be longer than the original character. | |
1035 | ||
1036 | The first code point of the titlecased version is returned | |
21da7284 KW |
1037 | (but note, as explained at L<the top of this section|/Character case |
1038 | changing>, that there may be more). | |
1f607577 | 1039 | |
a239b1e2 KW |
1040 | The suffix C<_safe> in the function's name indicates that it will not attempt |
1041 | to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is | |
1042 | true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the | |
1043 | input character is malformed in some way, the program may croak, or the | |
1044 | function may return the REPLACEMENT CHARACTER, at the discretion of the | |
1045 | implementation, and subject to change in future releases. | |
1046 | ||
1047 | =for apidoc Am|UV|toTITLE_utf8|U8* p|U8* s|STRLEN* lenp | |
1048 | This is like C<L</toLOWER_utf8_safe>>, but doesn't have the C<e> | |
1049 | parameter The function therefore can't check if it is reading | |
607313a1 KW |
1050 | beyond the end of the string. Starting in Perl v5.30, it will take the C<e> |
1051 | parameter, becoming a synonym for C<toTITLE_utf8_safe>. At that time every | |
1052 | program that uses it will have to be changed to successfully compile. In the | |
1053 | meantime, the first runtime call to C<toTITLE_utf8> from each call point in the | |
1054 | program will raise a deprecation warning, enabled by default. You can convert | |
1055 | your program now to use C<toTITLE_utf8_safe>, and avoid the warnings, and get an | |
1056 | extra measure of protection, or you can wait until v5.30, when you'll be forced | |
1057 | to add the C<e> parameter. | |
1f607577 | 1058 | |
954c1994 | 1059 | =cut |
353c9b6f | 1060 | |
d0da05db | 1061 | XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names |
1e222e4f KW |
1062 | really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change, |
1063 | and aren't general purpose as they don't work on U+DF, and assert against that. | |
243effed | 1064 | |
8a58bdcf | 1065 | Note that these macros are repeated in Devel::PPPort, so should also be |
62fa66b6 KW |
1066 | patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc |
1067 | ||
954c1994 GS |
1068 | */ |
1069 | ||
de9e2639 KW |
1070 | /* Specify the widest unsigned type on the platform. */ |
1071 | #ifdef QUADKIND | |
1072 | # define WIDEST_UTYPE U64 | |
7c062697 KW |
1073 | #else |
1074 | # define WIDEST_UTYPE U32 | |
1075 | #endif | |
1076 | ||
3912bc88 KW |
1077 | /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in |
1078 | * the lower 8. It is designed to be hopefully bomb-proof, making sure that no | |
1079 | * bits of information are lost even on a 64-bit machine, but to get the | |
1080 | * compiler to optimize it out if possible. This is because Configure makes | |
1081 | * sure that the machine has an 8-bit byte, so if c is stored in a byte, the | |
1082 | * sizeof() guarantees that this evaluates to a constant true at compile time. | |
7e75d1a1 JH |
1083 | * |
1084 | * For Coverity, be always true, because otherwise Coverity thinks | |
1085 | * it finds several expressions that are always true, independent | |
1086 | * of operands. Well, they are, but that is kind of the point. | |
220c71bf | 1087 | */ |
7e75d1a1 | 1088 | #ifndef __COVERITY__ |
6c5b02ac KW |
1089 | /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a |
1090 | * pointer) */ | |
1091 | #define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \ | |
ace3ad0f | 1092 | || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF)) |
7e75d1a1 JH |
1093 | #else |
1094 | #define FITS_IN_8_BITS(c) (1) | |
1095 | #endif | |
cf301eb7 | 1096 | |
94250c4f KW |
1097 | /* Returns true if c is in the range l..u, where 'l' is non-negative |
1098 | * Written this way so that after optimization, only one conditional test is | |
1099 | * needed. | |
1100 | * | |
1101 | * This isn't fully general, except for the special cased 'signed char' (which | |
1102 | * should be resolved at compile time): It won't work if 'c' is negative, and | |
1103 | * 'l' is larger than the max for that signed type. Thus if 'c' is a negative | |
1104 | * int, and 'l' is larger than INT_MAX, it will fail. To protect agains this | |
1105 | * happening, there is an assert that will generate a warning if c is larger | |
1106 | * than e.g. INT_MAX if it is an 'unsigned int'. This could be a false | |
1107 | * positive, but khw couldn't figure out a way to make it better. It's good | |
1108 | * enough so far */ | |
1109 | #define inRANGE(c, l, u) (__ASSERT_((l) >= 0) __ASSERT_((u) >= (l)) \ | |
1110 | ((sizeof(c) == 1) \ | |
1111 | ? (((WIDEST_UTYPE) ((((U8) (c))|0) - (l))) <= ((WIDEST_UTYPE) ((u) - (l)))) \ | |
1112 | : (__ASSERT_( (((WIDEST_UTYPE) 1) << (CHARBITS * sizeof(c) - 1) & (c)) \ | |
1113 | /* sign bit of c is 0 */ == 0 \ | |
1114 | || (((~ ((WIDEST_UTYPE) 1) << ((CHARBITS * sizeof(c) - 1) - 1))\ | |
1115 | /* l not larger than largest value in c's signed type */ \ | |
1116 | & ~ ((WIDEST_UTYPE) 0)) & (l)) == 0) \ | |
1117 | ((WIDEST_UTYPE) (((c) - (l)) | 0) <= ((WIDEST_UTYPE) ((u) - (l))))))) | |
305fe86e | 1118 | |
41f43cc2 | 1119 | #ifdef EBCDIC |
b6340bd0 | 1120 | # ifndef _ALL_SOURCE |
0852beac KW |
1121 | /* The native libc isascii() et.al. functions return the wrong results |
1122 | * on at least z/OS unless this is defined. */ | |
b6340bd0 KW |
1123 | # error _ALL_SOURCE should probably be defined |
1124 | # endif | |
41f43cc2 | 1125 | #else |
0852beac KW |
1126 | /* There is a simple definition of ASCII for ASCII platforms. But the |
1127 | * EBCDIC one isn't so simple, so is defined using table look-up like the | |
9c903d59 | 1128 | * other macros below. |
3f3c579d KW |
1129 | * |
1130 | * The cast here is used instead of '(c) >= 0', because some compilers emit | |
1131 | * a warning that that test is always true when the parameter is an | |
1132 | * unsigned type. khw supposes that it could be written as | |
1133 | * && ((c) == '\0' || (c) > 0) | |
1134 | * to avoid the message, but the cast will likely avoid extra branches even | |
1135 | * with stupid compilers. | |
1136 | * | |
1137 | * The '| 0' part ensures a compiler error if c is not integer (like e.g., | |
1138 | * a pointer) */ | |
9c903d59 | 1139 | # define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128) |
41f43cc2 KW |
1140 | #endif |
1141 | ||
38694112 KW |
1142 | /* Take the eight possible bit patterns of the lower 3 bits and you get the |
1143 | * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits | |
1144 | * can be ignored. If the rest match '0', we have an octal */ | |
1145 | #define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0') | |
c2da0b36 | 1146 | |
9fb1bf9d | 1147 | #ifdef H_PERL /* If have access to perl.h, lookup in its table */ |
f4cdb42c | 1148 | |
a500dc72 KW |
1149 | /* Character class numbers. For internal core Perl use only. The ones less |
1150 | * than 32 are used in PL_charclass[] and the ones up through the one that | |
1151 | * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and | |
1152 | * related files. PL_charclass ones use names used in l1_char_class_tab.h but | |
1153 | * their actual definitions are here. If that file has a name not used here, | |
1154 | * it won't compile. | |
1709d539 KW |
1155 | * |
1156 | * The first group of these is ordered in what I (khw) estimate to be the | |
31c7f561 | 1157 | * frequency of their use. This gives a slight edge to exiting a loop earlier |
58a3ba2c KW |
1158 | * (in reginclass() in regexec.c). Except \v should be last, as it isn't a |
1159 | * real Posix character class, and some (small) inefficiencies in regular | |
1160 | * expression handling would be introduced by putting it in the middle of those | |
1161 | * that are. Also, cntrl and ascii come after the others as it may be useful | |
1162 | * to group these which have no members that match above Latin1, (or above | |
1163 | * ASCII in the latter case) */ | |
1164 | ||
1709d539 KW |
1165 | # define _CC_WORDCHAR 0 /* \w and [:word:] */ |
1166 | # define _CC_DIGIT 1 /* \d and [:digit:] */ | |
1167 | # define _CC_ALPHA 2 /* [:alpha:] */ | |
1168 | # define _CC_LOWER 3 /* [:lower:] */ | |
1169 | # define _CC_UPPER 4 /* [:upper:] */ | |
1170 | # define _CC_PUNCT 5 /* [:punct:] */ | |
1171 | # define _CC_PRINT 6 /* [:print:] */ | |
15861f94 | 1172 | # define _CC_ALPHANUMERIC 7 /* [:alnum:] */ |
1709d539 | 1173 | # define _CC_GRAPH 8 /* [:graph:] */ |
359b005e | 1174 | # define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */ |
779cf272 | 1175 | # define _CC_SPACE 10 /* \s, [:space:] */ |
34aeb2e9 KW |
1176 | # define _CC_PSXSPC _CC_SPACE /* XXX Temporary, can be removed |
1177 | when the deprecated isFOO_utf8() | |
1178 | functions are removed */ | |
b0d691b2 KW |
1179 | # define _CC_BLANK 11 /* [:blank:] */ |
1180 | # define _CC_XDIGIT 12 /* [:xdigit:] */ | |
779cf272 KW |
1181 | # define _CC_CNTRL 13 /* [:cntrl:] */ |
1182 | # define _CC_ASCII 14 /* [:ascii:] */ | |
1183 | # define _CC_VERTSPACE 15 /* \v */ | |
1709d539 | 1184 | |
a0947d7b KW |
1185 | # define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE |
1186 | ||
1709d539 | 1187 | /* The members of the third group below do not need to be coordinated with data |
3ffc8c70 | 1188 | * structures in regcomp.[ch] and regexec.c. */ |
779cf272 KW |
1189 | # define _CC_IDFIRST 16 |
1190 | # define _CC_CHARNAME_CONT 17 | |
1191 | # define _CC_NONLATIN1_FOLD 18 | |
1192 | # define _CC_NONLATIN1_SIMPLE_FOLD 19 | |
1193 | # define _CC_QUOTEMETA 20 | |
1194 | # define _CC_NON_FINAL_FOLD 21 | |
1195 | # define _CC_IS_IN_SOME_FOLD 22 | |
1196 | # define _CC_MNEMONIC_CNTRL 23 | |
073c22b3 | 1197 | |
34aeb2e9 KW |
1198 | # define _CC_IDCONT 24 /* XXX Temporary, can be removed when the deprecated |
1199 | isFOO_utf8() functions are removed */ | |
1200 | ||
073c22b3 KW |
1201 | /* This next group is only used on EBCDIC platforms, so theoretically could be |
1202 | * shared with something entirely different that's only on ASCII platforms */ | |
5d5376e2 KW |
1203 | # define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 28 |
1204 | # define _CC_UTF8_IS_START 29 | |
1205 | # define _CC_UTF8_IS_DOWNGRADEABLE_START 30 | |
1206 | # define _CC_UTF8_IS_CONTINUATION 31 | |
1207 | /* Unused: 24-27 | |
f4cdb42c KW |
1208 | * If more bits are needed, one could add a second word for non-64bit |
1209 | * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd | |
37ede926 KW |
1210 | * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it |
1211 | * is used only for optimization (as of this writing), and differs in the | |
1212 | * Latin1 range from the ALPHA bit only in two relatively unimportant | |
a500dc72 | 1213 | * characters: the masculine and feminine ordinal indicators, so removing it |
073c22b3 KW |
1214 | * would just cause /i regexes which match them to run less efficiently. |
1215 | * Similarly the EBCDIC-only bits are used just for speed, and could be | |
1216 | * replaced by other means */ | |
96ac0975 | 1217 | |
3a371f2f KW |
1218 | #if defined(PERL_CORE) || defined(PERL_EXT) |
1219 | /* An enum version of the character class numbers, to help compilers | |
1220 | * optimize */ | |
1221 | typedef enum { | |
3a371f2f | 1222 | _CC_ENUM_ALPHA = _CC_ALPHA, |
e8d596e0 KW |
1223 | _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC, |
1224 | _CC_ENUM_ASCII = _CC_ASCII, | |
1225 | _CC_ENUM_BLANK = _CC_BLANK, | |
b0d691b2 | 1226 | _CC_ENUM_CASED = _CC_CASED, |
e8d596e0 | 1227 | _CC_ENUM_CNTRL = _CC_CNTRL, |
3a371f2f KW |
1228 | _CC_ENUM_DIGIT = _CC_DIGIT, |
1229 | _CC_ENUM_GRAPH = _CC_GRAPH, | |
1230 | _CC_ENUM_LOWER = _CC_LOWER, | |
1231 | _CC_ENUM_PRINT = _CC_PRINT, | |
1232 | _CC_ENUM_PUNCT = _CC_PUNCT, | |
e8d596e0 | 1233 | _CC_ENUM_SPACE = _CC_SPACE, |
3a371f2f | 1234 | _CC_ENUM_UPPER = _CC_UPPER, |
e8d596e0 | 1235 | _CC_ENUM_VERTSPACE = _CC_VERTSPACE, |
3a371f2f | 1236 | _CC_ENUM_WORDCHAR = _CC_WORDCHAR, |
e8d596e0 | 1237 | _CC_ENUM_XDIGIT = _CC_XDIGIT |
3a371f2f KW |
1238 | } _char_class_number; |
1239 | #endif | |
1240 | ||
86f72d56 | 1241 | #define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1) |
63c61c3f | 1242 | |
6635f04f | 1243 | START_EXTERN_C |
96ac0975 NC |
1244 | # ifdef DOINIT |
1245 | EXTCONST U32 PL_charclass[] = { | |
1246 | # include "l1_char_class_tab.h" | |
1247 | }; | |
1248 | ||
1249 | # else /* ! DOINIT */ | |
1250 | EXTCONST U32 PL_charclass[]; | |
1251 | # endif | |
6635f04f | 1252 | END_EXTERN_C |
96ac0975 | 1253 | |
265c1f46 | 1254 | /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */ |
430b7c70 | 1255 | # define _CC_mask(classnum) (1U << (classnum)) |
4650c663 KW |
1256 | |
1257 | /* For internal core Perl use only: the base macro for defining macros like | |
1258 | * isALPHA */ | |
ff7ecfc3 | 1259 | # define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \ |
f4cd282c | 1260 | && (PL_charclass[(U8) (c)] & _CC_mask(classnum))) |
4eeeb416 | 1261 | |
f4cdb42c KW |
1262 | /* The mask for the _A versions of the macros; it just adds in the bit for |
1263 | * ASCII. */ | |
1264 | # define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII)) | |
1265 | ||
4650c663 KW |
1266 | /* For internal core Perl use only: the base macro for defining macros like |
1267 | * isALPHA_A. The foo_A version makes sure that both the desired bit and | |
1268 | * the ASCII bit are present */ | |
b7d90381 KW |
1269 | # define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \ |
1270 | && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \ | |
1271 | == _CC_mask_A(classnum))) | |
f4cdb42c | 1272 | |
26c1d9d8 KW |
1273 | /* On ASCII platforms certain classes form a single range. It's faster to |
1274 | * special case these. isDIGIT is a single range on all platforms */ | |
b877c1ff KW |
1275 | # ifdef EBCDIC |
1276 | # define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA) | |
1277 | # define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH) | |
1278 | # define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER) | |
1279 | # define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT) | |
1280 | # define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER) | |
1281 | # else | |
26c1d9d8 | 1282 | /* By folding the upper and lowercase, we can use a single range */ |
b877c1ff | 1283 | # define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z') |
26c1d9d8 | 1284 | # define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e) |
b877c1ff KW |
1285 | # define isLOWER_A(c) inRANGE(c, 'a', 'z') |
1286 | # define isPRINT_A(c) inRANGE(c, ' ', 0x7e) | |
1287 | # define isUPPER_A(c) inRANGE(c, 'A', 'Z') | |
1288 | # endif | |
15861f94 | 1289 | # define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC) |
f4cdb42c KW |
1290 | # define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK) |
1291 | # define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL) | |
b877c1ff | 1292 | # define isDIGIT_A(c) inRANGE(c, '0', '9') |
f4cdb42c KW |
1293 | # define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT) |
1294 | # define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE) | |
f4cdb42c | 1295 | # define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR) |
b7d90381 KW |
1296 | # define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits |
1297 | */ | |
d95f8b6a | 1298 | # define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST) |
3ded5eb0 KW |
1299 | # define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA) |
1300 | # define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC) | |
1301 | # define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK) | |
1302 | ||
1303 | /* continuation character for legal NAME in \N{NAME} */ | |
1304 | # define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT) | |
1305 | ||
1306 | # define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL) | |
1307 | # define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH) | |
1308 | # define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER) | |
1309 | # define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT) | |
b7d90381 | 1310 | # define isPSXSPC_L1(c) isSPACE_L1(c) |
3ded5eb0 KW |
1311 | # define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT) |
1312 | # define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE) | |
1313 | # define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER) | |
1314 | # define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR) | |
1315 | # define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST) | |
f4cdb42c | 1316 | |
0852beac KW |
1317 | # ifdef EBCDIC |
1318 | # define isASCII(c) _generic_isCC(c, _CC_ASCII) | |
1319 | # endif | |
1320 | ||
f12c0118 KW |
1321 | /* Participates in a single-character fold with a character above 255 */ |
1322 | # 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))) | |
1323 | ||
1324 | /* Like the above, but also can be part of a multi-char fold */ | |
f4cd282c | 1325 | # 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))) |
430b7c70 | 1326 | |
4eeeb416 | 1327 | # define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA) |
26faadbd | 1328 | # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \ |
b7d90381 | 1329 | _generic_isCC(c, _CC_NON_FINAL_FOLD) |
37ede926 | 1330 | # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \ |
b7d90381 | 1331 | _generic_isCC(c, _CC_IS_IN_SOME_FOLD) |
8e35b056 KW |
1332 | # define _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \ |
1333 | _generic_isCC(c, _CC_MNEMONIC_CNTRL) | |
687c8d01 | 1334 | #else /* else we don't have perl.h H_PERL */ |
3ded5eb0 KW |
1335 | |
1336 | /* If we don't have perl.h, we are compiling a utility program. Below we | |
1337 | * hard-code various macro definitions that wouldn't otherwise be available | |
fc273927 | 1338 | * to it. Most are coded based on first principles. These are written to |
74665a89 | 1339 | * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */ |
182c4ace | 1340 | # define isDIGIT_A(c) inRANGE(c, '0', '9') |
0852beac | 1341 | # define isBLANK_A(c) ((c) == ' ' || (c) == '\t') |
74665a89 KW |
1342 | # define isSPACE_A(c) (isBLANK_A(c) \ |
1343 | || (c) == '\n' \ | |
1344 | || (c) == '\r' \ | |
1345 | || (c) == '\v' \ | |
0852beac | 1346 | || (c) == '\f') |
74665a89 KW |
1347 | /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for |
1348 | * uppercase. The tests for those aren't necessary on ASCII, but hurt only | |
1349 | * performance (if optimization isn't on), and allow the same code to be | |
1350 | * used for both platform types */ | |
182c4ace KW |
1351 | # define isLOWER_A(c) inRANGE((c), 'a', 'i') \ |
1352 | || inRANGE((c), 'j', 'r') \ | |
1353 | || inRANGE((c), 's', 'z') | |
1354 | # define isUPPER_A(c) inRANGE((c), 'A', 'I') \ | |
1355 | || inRANGE((c), 'J', 'R') \ | |
1356 | || inRANGE((c), 'S', 'Z') | |
a4d7a999 KW |
1357 | # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c)) |
1358 | # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c)) | |
3ded5eb0 | 1359 | # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_') |
0852beac | 1360 | # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_') |
182c4ace KW |
1361 | # define isXDIGIT_A(c) ( isDIGIT_A(c) \ |
1362 | || inRANGE((c), 'a', 'f') \ | |
1363 | || inRANGE((c), 'A', 'F') | |
74665a89 KW |
1364 | # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \ |
1365 | || (c) == '#' || (c) == '$' || (c) == '%' \ | |
1366 | || (c) == '&' || (c) == '\'' || (c) == '(' \ | |
1367 | || (c) == ')' || (c) == '*' || (c) == '+' \ | |
1368 | || (c) == ',' || (c) == '.' || (c) == '/' \ | |
1369 | || (c) == ':' || (c) == ';' || (c) == '<' \ | |
1370 | || (c) == '=' || (c) == '>' || (c) == '?' \ | |
1371 | || (c) == '@' || (c) == '[' || (c) == '\\' \ | |
1372 | || (c) == ']' || (c) == '^' || (c) == '_' \ | |
1373 | || (c) == '`' || (c) == '{' || (c) == '|' \ | |
1374 | || (c) == '}' || (c) == '~') | |
1375 | # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c)) | |
1376 | # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ') | |
3ded5eb0 | 1377 | |
0852beac | 1378 | # ifdef EBCDIC |
74665a89 KW |
1379 | /* The below is accurate for the 3 EBCDIC code pages traditionally |
1380 | * supported by perl. The only difference between them in the controls | |
1381 | * is the position of \n, and that is represented symbolically below */ | |
1382 | # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \ | |
1383 | || (c) == '\f' || (c) == '\n' || (c) == '\r' \ | |
1384 | || (c) == '\t' || (c) == '\v' \ | |
182c4ace | 1385 | || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \ |
74665a89 | 1386 | || (c) == 7 /* U+7F DEL */ \ |
182c4ace KW |
1387 | || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \ |
1388 | DC[1-3] */ \ | |
74665a89 KW |
1389 | || (c) == 0x18 /* U+18 CAN */ \ |
1390 | || (c) == 0x19 /* U+19 EOM */ \ | |
182c4ace | 1391 | || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \ |
74665a89 KW |
1392 | || (c) == 0x26 /* U+17 ETB */ \ |
1393 | || (c) == 0x27 /* U+1B ESC */ \ | |
1394 | || (c) == 0x2D /* U+05 ENQ */ \ | |
1395 | || (c) == 0x2E /* U+06 ACK */ \ | |
1396 | || (c) == 0x32 /* U+16 SYN */ \ | |
1397 | || (c) == 0x37 /* U+04 EOT */ \ | |
1398 | || (c) == 0x3C /* U+14 DC4 */ \ | |
1399 | || (c) == 0x3D /* U+15 NAK */ \ | |
1400 | || (c) == 0x3F)/* U+1A SUB */ | |
0852beac | 1401 | # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c)) |
74665a89 KW |
1402 | # else /* isASCII is already defined for ASCII platforms, so can use that to |
1403 | define isCNTRL */ | |
1404 | # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c)) | |
0852beac KW |
1405 | # endif |
1406 | ||
3ffc8c70 | 1407 | /* The _L1 macros may be unnecessary for the utilities; I (khw) added them |
caa94d35 KW |
1408 | * during debugging, and it seems best to keep them. We may be called |
1409 | * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't | |
1410 | * do anything anyway, so make it not a problem */ | |
1411 | # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1) | |
1412 | # define NATIVE_TO_LATIN1(ch) (ch) | |
1413 | # endif | |
3ded5eb0 KW |
1414 | # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c)) |
1415 | # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c)) | |
1416 | # define isBLANK_L1(c) (isBLANK_A(c) \ | |
1417 | || (FITS_IN_8_BITS(c) \ | |
1418 | && NATIVE_TO_LATIN1((U8) c) == 0xA0)) | |
1419 | # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c))) | |
1420 | # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c))) | |
1421 | # define isLOWER_L1(c) (isLOWER_A(c) \ | |
1422 | || (FITS_IN_8_BITS(c) \ | |
ae683a5f | 1423 | && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \ |
3ded5eb0 KW |
1424 | && NATIVE_TO_LATIN1((U8) c) != 0xF7) \ |
1425 | || NATIVE_TO_LATIN1((U8) c) == 0xAA \ | |
1426 | || NATIVE_TO_LATIN1((U8) c) == 0xBA \ | |
1427 | || NATIVE_TO_LATIN1((U8) c) == 0xB5))) | |
1428 | # define isPRINT_L1(c) (isPRINT_A(c) \ | |
1429 | || (FITS_IN_8_BITS(c) \ | |
1430 | && NATIVE_TO_LATIN1((U8) c) >= 0xA0)) | |
3ded5eb0 KW |
1431 | # define isPUNCT_L1(c) (isPUNCT_A(c) \ |
1432 | || (FITS_IN_8_BITS(c) \ | |
ae683a5f | 1433 | && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \ |
3ded5eb0 KW |
1434 | || NATIVE_TO_LATIN1((U8) c) == 0xA7 \ |
1435 | || NATIVE_TO_LATIN1((U8) c) == 0xAB \ | |
1436 | || NATIVE_TO_LATIN1((U8) c) == 0xB6 \ | |
1437 | || NATIVE_TO_LATIN1((U8) c) == 0xB7 \ | |
1438 | || NATIVE_TO_LATIN1((U8) c) == 0xBB \ | |
1439 | || NATIVE_TO_LATIN1((U8) c) == 0xBF))) | |
1440 | # define isSPACE_L1(c) (isSPACE_A(c) \ | |
1441 | || (FITS_IN_8_BITS(c) \ | |
ae683a5f | 1442 | && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \ |
3ded5eb0 KW |
1443 | || NATIVE_TO_LATIN1((U8) c) == 0xA0))) |
1444 | # define isUPPER_L1(c) (isUPPER_A(c) \ | |
1445 | || (FITS_IN_8_BITS(c) \ | |
182c4ace KW |
1446 | && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \ |
1447 | 0xC0, 0xDE) \ | |
3ded5eb0 KW |
1448 | && NATIVE_TO_LATIN1((U8) c) != 0xD7))) |
1449 | # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c)) | |
1450 | # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_') | |
1451 | # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \ | |
1452 | || isBLANK_L1(c) \ | |
1453 | || (c) == '-' \ | |
1454 | || (c) == '(' \ | |
1455 | || (c) == ')') | |
1456 | /* The following are not fully accurate in the above-ASCII range. I (khw) | |
1457 | * don't think it's necessary to be so for the purposes where this gets | |
1458 | * compiled */ | |
1459 | # define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c)) | |
1460 | # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c) | |
1461 | ||
1462 | /* And these aren't accurate at all. They are useful only for above | |
1463 | * Latin1, which utilities and bootstrapping don't deal with */ | |
1464 | # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0 | |
6838b41e | 1465 | # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0 |
3ded5eb0 KW |
1466 | # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0 |
1467 | ||
1468 | /* Many of the macros later in this file are defined in terms of these. By | |
1469 | * implementing them with a function, which converts the class number into | |
1470 | * a call to the desired macro, all of the later ones work. However, that | |
1471 | * function won't be actually defined when building a utility program (no | |
1472 | * perl.h), and so a compiler error will be generated if one is attempted | |
1473 | * to be used. And the above-Latin1 code points require Unicode tables to | |
1474 | * be present, something unlikely to be the case when bootstrapping */ | |
1475 | # define _generic_isCC(c, classnum) \ | |
1476 | (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE)) | |
1477 | # define _generic_isCC_A(c, classnum) \ | |
1478 | (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE)) | |
687c8d01 | 1479 | #endif /* End of no perl.h H_PERL */ |
8a58bdcf | 1480 | |
e66b99e9 KW |
1481 | #define isALPHANUMERIC(c) isALPHANUMERIC_A(c) |
1482 | #define isALPHA(c) isALPHA_A(c) | |
0852beac KW |
1483 | #define isASCII_A(c) isASCII(c) |
1484 | #define isASCII_L1(c) isASCII(c) | |
e66b99e9 KW |
1485 | #define isBLANK(c) isBLANK_A(c) |
1486 | #define isCNTRL(c) isCNTRL_A(c) | |
1487 | #define isDIGIT(c) isDIGIT_A(c) | |
1488 | #define isGRAPH(c) isGRAPH_A(c) | |
1489 | #define isIDFIRST(c) isIDFIRST_A(c) | |
1490 | #define isLOWER(c) isLOWER_A(c) | |
1491 | #define isPRINT(c) isPRINT_A(c) | |
779cf272 | 1492 | #define isPSXSPC_A(c) isSPACE_A(c) |
e66b99e9 | 1493 | #define isPSXSPC(c) isPSXSPC_A(c) |
779cf272 | 1494 | #define isPSXSPC_L1(c) isSPACE_L1(c) |
e66b99e9 KW |
1495 | #define isPUNCT(c) isPUNCT_A(c) |
1496 | #define isSPACE(c) isSPACE_A(c) | |
1497 | #define isUPPER(c) isUPPER_A(c) | |
1498 | #define isWORDCHAR(c) isWORDCHAR_A(c) | |
1499 | #define isXDIGIT(c) isXDIGIT_A(c) | |
1500 | ||
1501 | /* ASCII casing. These could also be written as | |
1502 | #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c)) | |
1503 | #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c)) | |
1504 | which uses table lookup and mask instead of subtraction. (This would | |
c5e9991e KW |
1505 | work because the _MOD does not apply in the ASCII range). |
1506 | ||
1507 | These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII | |
1508 | UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are | |
1509 | there non-ASCII invariants, and all of them are controls.) */ | |
68067e4e DM |
1510 | #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c)) |
1511 | #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c)) | |
bbce6d69 | 1512 | |
25200305 KW |
1513 | /* In the ASCII range, these are equivalent to what they're here defined to be. |
1514 | * But by creating these definitions, other code doesn't have to be aware of | |
c5e9991e KW |
1515 | * this detail. Actually this works for all UTF-8 invariants, not just the |
1516 | * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */ | |
25200305 | 1517 | #define toFOLD(c) toLOWER(c) |
25200305 KW |
1518 | #define toTITLE(c) toUPPER(c) |
1519 | ||
c753c8d3 KW |
1520 | #define toLOWER_A(c) toLOWER(c) |
1521 | #define toUPPER_A(c) toUPPER(c) | |
25200305 KW |
1522 | #define toFOLD_A(c) toFOLD(c) |
1523 | #define toTITLE_A(c) toTITLE(c) | |
1a0901db | 1524 | |
4650c663 | 1525 | /* Use table lookup for speed; returns the input itself if is out-of-range */ |
b2bf251f | 1526 | #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \ |
8e7c6e7d | 1527 | ? (c) \ |
f4cd282c | 1528 | : PL_latin1_lc[ (U8) (c) ]) |
c753c8d3 KW |
1529 | #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */ |
1530 | ||
1a0901db | 1531 | /* Modified uc. Is correct uc except for three non-ascii chars which are |
4650c663 KW |
1532 | * all mapped to one of them, and these need special handling; returns the |
1533 | * input itself if is out-of-range */ | |
b2bf251f | 1534 | #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \ |
8e7c6e7d | 1535 | ? (c) \ |
f4cd282c | 1536 | : PL_mod_latin1_uc[ (U8) (c) ]) |
31f05a37 | 1537 | #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale |
84061b6a | 1538 | |
beab9ebe KW |
1539 | /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */ |
1540 | ||
1541 | /* For internal core Perl use only: the base macro for defining macros like | |
1542 | * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point | |
31f05a37 KW |
1543 | * (0-255) to check. In a UTF-8 locale, the result is the same as calling |
1544 | * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like | |
1545 | * _CC_UPPER, which gives the class number for doing this. For non-UTF-8 | |
1546 | * locales, the code to actually do the test this is passed in 'non_utf8'. If | |
1547 | * 'c' is above 255, 0 is returned. For accessing the full range of possible | |
1548 | * code points under locale rules, use the macros based on _generic_LC_uvchr | |
1549 | * instead of this. */ | |
beab9ebe KW |
1550 | #define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \ |
1551 | (! FITS_IN_8_BITS(c) \ | |
1552 | ? 0 \ | |
31f05a37 KW |
1553 | : IN_UTF8_CTYPE_LOCALE \ |
1554 | ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \ | |
beab9ebe KW |
1555 | : cBOOL(non_utf8)) |
1556 | ||
1557 | /* For internal core Perl use only: a helper macro for defining macros like | |
1558 | * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to | |
1559 | * actually do this test is passed in 'non_utf8_func', which is called on 'c', | |
1560 | * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See | |
1561 | * _generic_LC_base for more info */ | |
1562 | #define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \ | |
1563 | _generic_LC_base(c,utf8_locale_classnum, \ | |
1564 | non_utf8_func( (_LC_CAST) (c))) | |
1565 | ||
1566 | /* For internal core Perl use only: like _generic_LC, but also returns TRUE if | |
1567 | * 'c' is the platform's native underscore character */ | |
1568 | #define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \ | |
1569 | _generic_LC_base(c, utf8_locale_classnum, \ | |
1570 | (non_utf8_func( (_LC_CAST) (c)) \ | |
1571 | || (char)(c) == '_')) | |
1572 | ||
1573 | /* These next three are also for internal core Perl use only: case-change | |
247985d4 KW |
1574 | * helper macros. The reason for using the PL_latin arrays is in case the |
1575 | * system function is defective; it ensures uniform results that conform to the | |
b257a28c | 1576 | * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */ |
beab9ebe KW |
1577 | #define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \ |
1578 | ? (c) \ | |
31f05a37 KW |
1579 | : (IN_UTF8_CTYPE_LOCALE) \ |
1580 | ? PL_latin1_lc[ (U8) (c) ] \ | |
5a10328c | 1581 | : (cast)function((cast)(c))) |
beab9ebe | 1582 | |
31f05a37 KW |
1583 | /* Note that the result can be larger than a byte in a UTF-8 locale. It |
1584 | * returns a single value, so can't adequately return the upper case of LATIN | |
1585 | * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two | |
1586 | * values "SS"); instead it asserts against that under DEBUGGING, and | |
b257a28c KW |
1587 | * otherwise returns its input. It does not handle the anomalies in UTF-8 |
1588 | * Turkic locales. */ | |
beab9ebe KW |
1589 | #define _generic_toUPPER_LC(c, function, cast) \ |
1590 | (! FITS_IN_8_BITS(c) \ | |
1591 | ? (c) \ | |
31f05a37 | 1592 | : ((! IN_UTF8_CTYPE_LOCALE) \ |
b7d90381 | 1593 | ? (cast)function((cast)(c)) \ |
31f05a37 KW |
1594 | : ((((U8)(c)) == MICRO_SIGN) \ |
1595 | ? GREEK_CAPITAL_LETTER_MU \ | |
1596 | : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \ | |
1597 | ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \ | |
1598 | : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \ | |
1599 | ? (__ASSERT_(0) (c)) \ | |
1600 | : PL_mod_latin1_uc[ (U8) (c) ]))))) | |
1601 | ||
1602 | /* Note that the result can be larger than a byte in a UTF-8 locale. It | |
1603 | * returns a single value, so can't adequately return the fold case of LATIN | |
1604 | * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two | |
1605 | * values "ss"); instead it asserts against that under DEBUGGING, and | |
b257a28c KW |
1606 | * otherwise returns its input. It does not handle the anomalies in UTF-8 |
1607 | * Turkic locales */ | |
beab9ebe | 1608 | #define _generic_toFOLD_LC(c, function, cast) \ |
4d7db1b9 | 1609 | ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \ |
31f05a37 | 1610 | ? GREEK_SMALL_LETTER_MU \ |
4d7db1b9 KW |
1611 | : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \ |
1612 | || (c) != LATIN_SMALL_LETTER_SHARP_S) \ | |
1613 | _generic_toLOWER_LC(c, function, cast))) | |
beab9ebe | 1614 | |
84061b6a | 1615 | /* Use the libc versions for these if available. */ |
f05550c0 | 1616 | #if defined(HAS_ISASCII) |
84061b6a KW |
1617 | # define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c))) |
1618 | #else | |
1619 | # define isASCII_LC(c) isASCII(c) | |
1620 | #endif | |
1621 | ||
f05550c0 | 1622 | #if defined(HAS_ISBLANK) |
84061b6a | 1623 | # define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank) |
95f34b6f | 1624 | #else /* Unlike isASCII, varies if in a UTF-8 locale */ |
278b2b56 | 1625 | # define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c)) |
84061b6a KW |
1626 | #endif |
1627 | ||
f05550c0 | 1628 | #define _LC_CAST U8 |
bbce6d69 | 1629 | |
f05550c0 | 1630 | #ifdef WIN32 |
375f5f06 KW |
1631 | /* The Windows functions don't bother to follow the POSIX standard, which |
1632 | * for example says that something can't both be a printable and a control. | |
1633 | * But Windows treats the \t control as a printable, and does such things | |
1634 | * as making superscripts into both digits and punctuation. This tames | |
1635 | * these flaws by assuming that the definitions of both controls and space | |
1636 | * are correct, and then making sure that other definitions don't have | |
1637 | * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc. | |
1638 | * Not all possible weirdnesses are checked for, just the ones that were | |
1639 | * detected on actual Microsoft code pages */ | |
1640 | ||
b7d90381 KW |
1641 | # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl) |
1642 | # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace) | |
1643 | ||
1644 | # define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \ | |
1645 | && isALPHANUMERIC_LC(c)) | |
1646 | # define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \ | |
1647 | ! isPUNCT_LC(c)) | |
1648 | # define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \ | |
1649 | isALPHANUMERIC_LC(c)) | |
1650 | # define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c)) | |
1651 | # define isIDFIRST_LC(c) (((c) == '_') \ | |
1652 | || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c))) | |
1653 | # define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c)) | |
1654 | # define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c)) | |
1655 | # define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c)) | |
1656 | # define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c)) | |
f05550c0 | 1657 | # define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c)) |
b7d90381 KW |
1658 | # define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \ |
1659 | && isALPHANUMERIC_LC(c)) | |
f05550c0 BF |
1660 | |
1661 | # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8) | |
1662 | # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8) | |
1663 | # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8) | |
1664 | ||
1665 | #elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII)) | |
4650c663 | 1666 | /* For most other platforms */ |
beab9ebe | 1667 | |
f05550c0 BF |
1668 | # define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha) |
1669 | # define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum) | |
1670 | # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl) | |
1671 | # define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit) | |
1672 | # define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph) | |
1673 | # define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha) | |
1674 | # define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower) | |
1675 | # define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint) | |
1676 | # define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct) | |
1677 | # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace) | |
1678 | # define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper) | |
1679 | # define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum) | |
1680 | # define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit) | |
1681 | ||
1682 | ||
1683 | # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8) | |
1684 | # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8) | |
1685 | # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8) | |
1686 | ||
1687 | #else /* The final fallback position */ | |
1688 | ||
b7d90381 KW |
1689 | # define isALPHA_LC(c) (isascii(c) && isalpha(c)) |
1690 | # define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c)) | |
1691 | # define isCNTRL_LC(c) (isascii(c) && iscntrl(c)) | |
1692 | # define isDIGIT_LC(c) (isascii(c) && isdigit(c)) | |
1693 | # define isGRAPH_LC(c) (isascii(c) && isgraph(c)) | |
f05550c0 | 1694 | # define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_')) |
b7d90381 KW |
1695 | # define isLOWER_LC(c) (isascii(c) && islower(c)) |
1696 | # define isPRINT_LC(c) (isascii(c) && isprint(c)) | |
1697 | # define isPUNCT_LC(c) (isascii(c) && ispunct(c)) | |
1698 | # define isSPACE_LC(c) (isascii(c) && isspace(c)) | |
1699 | # define isUPPER_LC(c) (isascii(c) && isupper(c)) | |
f05550c0 | 1700 | # define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_')) |
b7d90381 | 1701 | # define isXDIGIT_LC(c) (isascii(c) && isxdigit(c)) |
f05550c0 BF |
1702 | |
1703 | # define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c)) | |
1704 | # define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c)) | |
1705 | # define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c)) | |
bbce6d69 | 1706 | |
f05550c0 | 1707 | #endif |
55204971 | 1708 | |
eba68aa0 KW |
1709 | #define isIDCONT(c) isWORDCHAR(c) |
1710 | #define isIDCONT_A(c) isWORDCHAR_A(c) | |
1711 | #define isIDCONT_L1(c) isWORDCHAR_L1(c) | |
1712 | #define isIDCONT_LC(c) isWORDCHAR_LC(c) | |
13380643 | 1713 | #define isPSXSPC_LC(c) isSPACE_LC(c) |
aaa51d5e | 1714 | |
4650c663 | 1715 | /* For internal core Perl use only: the base macros for defining macros like |
d0da05db KW |
1716 | * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class |
1717 | * number defined earlier in this file. _generic_uvchr() is used for POSIX | |
4650c663 KW |
1718 | * classes where there is a macro or function 'above_latin1' that takes the |
1719 | * single argument 'c' and returns the desired value. These exist for those | |
1720 | * classes which have simple definitions, avoiding the overhead of a hash | |
d0da05db | 1721 | * lookup or inversion list binary search. _generic_swash_uvchr() can be used |
4650c663 | 1722 | * for classes where that overhead is faster than a direct lookup. |
d0da05db | 1723 | * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the |
4650c663 KW |
1724 | * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so |
1725 | * have duplicate checks here, so could create versions of the macros that | |
1726 | * don't, but experiments show that gcc optimizes them out anyway. */ | |
66c17564 KW |
1727 | |
1728 | /* Note that all ignore 'use bytes' */ | |
1e222e4f KW |
1729 | #define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \ |
1730 | ? _generic_isCC(c, classnum) \ | |
cd500f2f | 1731 | : above_latin1(c)) |
1e222e4f KW |
1732 | #define _generic_swash_uvchr(classnum, c) ((c) < 256 \ |
1733 | ? _generic_isCC(c, classnum) \ | |
922e8cb4 | 1734 | : _is_uni_FOO(classnum, c)) |
d0da05db KW |
1735 | #define isALPHA_uvchr(c) _generic_swash_uvchr(_CC_ALPHA, c) |
1736 | #define isALPHANUMERIC_uvchr(c) _generic_swash_uvchr(_CC_ALPHANUMERIC, c) | |
1737 | #define isASCII_uvchr(c) isASCII(c) | |
1738 | #define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c) | |
1739 | #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */ | |
1740 | #define isDIGIT_uvchr(c) _generic_swash_uvchr(_CC_DIGIT, c) | |
1741 | #define isGRAPH_uvchr(c) _generic_swash_uvchr(_CC_GRAPH, c) | |
1e222e4f KW |
1742 | #define isIDCONT_uvchr(c) \ |
1743 | _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c) | |
1744 | #define isIDFIRST_uvchr(c) \ | |
1745 | _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c) | |
d0da05db KW |
1746 | #define isLOWER_uvchr(c) _generic_swash_uvchr(_CC_LOWER, c) |
1747 | #define isPRINT_uvchr(c) _generic_swash_uvchr(_CC_PRINT, c) | |
1748 | ||
1749 | #define isPUNCT_uvchr(c) _generic_swash_uvchr(_CC_PUNCT, c) | |
1750 | #define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c) | |
1751 | #define isPSXSPC_uvchr(c) isSPACE_uvchr(c) | |
1752 | ||
1753 | #define isUPPER_uvchr(c) _generic_swash_uvchr(_CC_UPPER, c) | |
1754 | #define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c) | |
1755 | #define isWORDCHAR_uvchr(c) _generic_swash_uvchr(_CC_WORDCHAR, c) | |
1756 | #define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c) | |
1757 | ||
1758 | #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l) | |
1759 | #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l) | |
1760 | #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l) | |
1761 | #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l) | |
1762 | ||
1763 | /* For backwards compatibility, even though '_uni' should mean official Unicode | |
1764 | * code points, in Perl it means native for those below 256 */ | |
1765 | #define isALPHA_uni(c) isALPHA_uvchr(c) | |
1766 | #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c) | |
1767 | #define isASCII_uni(c) isASCII_uvchr(c) | |
1768 | #define isBLANK_uni(c) isBLANK_uvchr(c) | |
1769 | #define isCNTRL_uni(c) isCNTRL_uvchr(c) | |
1770 | #define isDIGIT_uni(c) isDIGIT_uvchr(c) | |
1771 | #define isGRAPH_uni(c) isGRAPH_uvchr(c) | |
1772 | #define isIDCONT_uni(c) isIDCONT_uvchr(c) | |
1773 | #define isIDFIRST_uni(c) isIDFIRST_uvchr(c) | |
1774 | #define isLOWER_uni(c) isLOWER_uvchr(c) | |
1775 | #define isPRINT_uni(c) isPRINT_uvchr(c) | |
1776 | #define isPUNCT_uni(c) isPUNCT_uvchr(c) | |
1777 | #define isSPACE_uni(c) isSPACE_uvchr(c) | |
1778 | #define isPSXSPC_uni(c) isPSXSPC_uvchr(c) | |
1779 | #define isUPPER_uni(c) isUPPER_uvchr(c) | |
1780 | #define isVERTWS_uni(c) isVERTWS_uvchr(c) | |
1781 | #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c) | |
1782 | #define isXDIGIT_uni(c) isXDIGIT_uvchr(c) | |
1783 | #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l) | |
1784 | #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l) | |
1785 | #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l) | |
1786 | #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l) | |
a0ed51b3 | 1787 | |
4650c663 KW |
1788 | /* For internal core Perl use only: the base macros for defining macros like |
1789 | * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code | |
d0da05db | 1790 | * point, not just 0-255. Like _generic_uvchr, there are two versions, one for |
4650c663 | 1791 | * simple class definitions; the other for more complex. These are like |
d0da05db | 1792 | * _generic_uvchr, so see it for more info. */ |
cd500f2f KW |
1793 | #define _generic_LC_uvchr(latin1, above_latin1, c) \ |
1794 | (c < 256 ? latin1(c) : above_latin1(c)) | |
1795 | #define _generic_LC_swash_uvchr(latin1, classnum, c) \ | |
1796 | (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c)) | |
1797 | ||
1798 | #define isALPHA_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHA_LC, _CC_ALPHA, c) | |
1799 | #define isALPHANUMERIC_LC_uvchr(c) _generic_LC_swash_uvchr(isALPHANUMERIC_LC, \ | |
1800 | _CC_ALPHANUMERIC, c) | |
b7d90381 KW |
1801 | #define isASCII_LC_uvchr(c) isASCII_LC(c) |
1802 | #define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \ | |
1803 | is_HORIZWS_cp_high, c) | |
feeab5a9 | 1804 | #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0) |
cd500f2f KW |
1805 | #define isDIGIT_LC_uvchr(c) _generic_LC_swash_uvchr(isDIGIT_LC, _CC_DIGIT, c) |
1806 | #define isGRAPH_LC_uvchr(c) _generic_LC_swash_uvchr(isGRAPH_LC, _CC_GRAPH, c) | |
b7d90381 | 1807 | #define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \ |
eba68aa0 | 1808 | _is_uni_perl_idcont, c) |
b7d90381 | 1809 | #define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \ |
cd500f2f KW |
1810 | _is_uni_perl_idstart, c) |
1811 | #define isLOWER_LC_uvchr(c) _generic_LC_swash_uvchr(isLOWER_LC, _CC_LOWER, c) | |
1812 | #define isPRINT_LC_uvchr(c) _generic_LC_swash_uvchr(isPRINT_LC, _CC_PRINT, c) | |
b7d90381 | 1813 | #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c) |
cd500f2f | 1814 | #define isPUNCT_LC_uvchr(c) _generic_LC_swash_uvchr(isPUNCT_LC, _CC_PUNCT, c) |
b7d90381 | 1815 | #define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \ |
509fb054 | 1816 | is_XPERLSPACE_cp_high, c) |
cd500f2f | 1817 | #define isUPPER_LC_uvchr(c) _generic_LC_swash_uvchr(isUPPER_LC, _CC_UPPER, c) |
b7d90381 | 1818 | #define isWORDCHAR_LC_uvchr(c) _generic_LC_swash_uvchr(isWORDCHAR_LC, \ |
cd500f2f | 1819 | _CC_WORDCHAR, c) |
b7d90381 KW |
1820 | #define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \ |
1821 | is_XDIGIT_cp_high, c) | |
e712593e | 1822 | |
b7d90381 | 1823 | #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c)) |
aaa51d5e | 1824 | |
4650c663 KW |
1825 | /* For internal core Perl use only: the base macros for defining macros like |
1826 | * isALPHA_utf8. These are like the earlier defined macros, but take an input | |
1827 | * UTF-8 encoded string 'p'. If the input is in the Latin1 range, use | |
1828 | * the Latin1 macro 'classnum' on 'p'. Otherwise use the value given by the | |
1829 | * 'utf8' parameter. This relies on the fact that ASCII characters have the | |
1830 | * same representation whether utf8 or not. Note that it assumes that the utf8 | |
1831 | * has been validated, and ignores 'use bytes' */ | |
34aeb2e9 KW |
1832 | #define _base_generic_utf8(enum_name, name, p, use_locale ) \ |
1833 | _is_utf8_FOO(CAT2(_CC_, enum_name), \ | |
1834 | (const U8 *) p, \ | |
1835 | "is" STRINGIFY(name) "_utf8", \ | |
1836 | "is" STRINGIFY(name) "_utf8_safe", \ | |
1837 | 1, use_locale, __FILE__,__LINE__) | |
1838 | ||
1839 | #define _generic_utf8(name, p) _base_generic_utf8(name, name, p, 0) | |
da8c1a98 KW |
1840 | |
1841 | /* The "_safe" macros make sure that we don't attempt to read beyond 'e', but | |
1842 | * they don't otherwise go out of their way to look for malformed UTF-8. If | |
1843 | * they can return accurate results without knowing if the input is otherwise | |
1844 | * malformed, they do so. For example isASCII is accurate in spite of any | |
1845 | * non-length malformations because it looks only at a single byte. Likewise | |
1846 | * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8 | |
1847 | * variant ones return FALSE. But, if the input has to be well-formed in order | |
1848 | * for the results to be accurate, the macros will test and if malformed will | |
1849 | * call a routine to die | |
1850 | * | |
1851 | * Except for toke.c, the macros do assume that e > p, asserting that on | |
1852 | * DEBUGGING builds. Much code that calls these depends on this being true, | |
1853 | * for other reasons. toke.c is treated specially as using the regular | |
1854 | * assertion breaks it in many ways. All strings that these operate on there | |
1855 | * are supposed to have an extra NUL character at the end, so that *e = \0. A | |
1856 | * bunch of code in toke.c assumes that this is true, so the assertion allows | |
1857 | * for that */ | |
1858 | #ifdef PERL_IN_TOKE_C | |
1859 | # define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0')) | |
1860 | #else | |
1861 | # define _utf8_safe_assert(p,e) ((e) > (p)) | |
1862 | #endif | |
1863 | ||
1864 | #define _generic_utf8_safe(classnum, p, e, above_latin1) \ | |
1865 | (__ASSERT_(_utf8_safe_assert(p, e)) \ | |
1866 | (UTF8_IS_INVARIANT(*(p))) \ | |
1867 | ? _generic_isCC(*(p), classnum) \ | |
1868 | : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \ | |
1869 | ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \ | |
1870 | ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \ | |
1871 | classnum) \ | |
1872 | : (_force_out_malformed_utf8_message( \ | |
1873 | (U8 *) (p), (U8 *) (e), 0, 1), 0)) \ | |
1874 | : above_latin1)) | |
b7d90381 KW |
1875 | /* Like the above, but calls 'above_latin1(p)' to get the utf8 value. |
1876 | * 'above_latin1' can be a macro */ | |
da8c1a98 KW |
1877 | #define _generic_func_utf8_safe(classnum, above_latin1, p, e) \ |
1878 | _generic_utf8_safe(classnum, p, e, above_latin1(p, e)) | |
1879 | #define _generic_non_swash_utf8_safe(classnum, above_latin1, p, e) \ | |
1880 | _generic_utf8_safe(classnum, p, e, \ | |
1881 | (UNLIKELY((e) - (p) < UTF8SKIP(p)) \ | |
1882 | ? (_force_out_malformed_utf8_message( \ | |
1883 | (U8 *) (p), (U8 *) (e), 0, 1), 0) \ | |
1884 | : above_latin1(p))) | |
4650c663 | 1885 | /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an |
cd500f2f | 1886 | * 'above_latin1' parameter */ |
da8c1a98 KW |
1887 | #define _generic_swash_utf8_safe(classnum, p, e) \ |
1888 | _generic_utf8_safe(classnum, p, e, _is_utf8_FOO_with_len(classnum, p, e)) | |
922e8cb4 | 1889 | |
cc8ab7c0 | 1890 | /* Like the above, but should be used only when it is known that there are no |
ff7ecfc3 KW |
1891 | * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the |
1892 | * class is TRUE for. Hence it can skip the tests for this range. | |
1893 | * 'above_latin1' should include its arguments */ | |
da8c1a98 KW |
1894 | #define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \ |
1895 | (__ASSERT_(_utf8_safe_assert(p, e)) \ | |
1896 | (UTF8_IS_INVARIANT(*(p))) \ | |
1897 | ? _generic_isCC(*(p), classnum) \ | |
1898 | : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \ | |
1899 | ? 0 /* Note that doesn't check validity for latin1 */ \ | |
1900 | : above_latin1) | |
1901 | ||
84238efa | 1902 | |
34aeb2e9 KW |
1903 | #define isALPHA_utf8(p) _generic_utf8(ALPHA, p) |
1904 | #define isALPHANUMERIC_utf8(p) _generic_utf8(ALPHANUMERIC, p) | |
1905 | #define isASCII_utf8(p) _generic_utf8(ASCII, p) | |
1906 | #define isBLANK_utf8(p) _generic_utf8(BLANK, p) | |
1907 | #define isCNTRL_utf8(p) _generic_utf8(CNTRL, p) | |
1908 | #define isDIGIT_utf8(p) _generic_utf8(DIGIT, p) | |
1909 | #define isGRAPH_utf8(p) _generic_utf8(GRAPH, p) | |
1910 | #define isIDCONT_utf8(p) _generic_utf8(IDCONT, p) | |
1911 | #define isIDFIRST_utf8(p) _generic_utf8(IDFIRST, p) | |
1912 | #define isLOWER_utf8(p) _generic_utf8(LOWER, p) | |
1913 | #define isPRINT_utf8(p) _generic_utf8(PRINT, p) | |
1914 | #define isPSXSPC_utf8(p) _generic_utf8(PSXSPC, p) | |
1915 | #define isPUNCT_utf8(p) _generic_utf8(PUNCT, p) | |
1916 | #define isSPACE_utf8(p) _generic_utf8(SPACE, p) | |
1917 | #define isUPPER_utf8(p) _generic_utf8(UPPER, p) | |
1918 | #define isVERTWS_utf8(p) _generic_utf8(VERTSPACE, p) | |
1919 | #define isWORDCHAR_utf8(p) _generic_utf8(WORDCHAR, p) | |
1920 | #define isXDIGIT_utf8(p) _generic_utf8(XDIGIT, p) | |
e8fa43e2 | 1921 | |
da8c1a98 KW |
1922 | #define isALPHA_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_ALPHA, p, e) |
1923 | #define isALPHANUMERIC_utf8_safe(p, e) \ | |
1924 | _generic_swash_utf8_safe(_CC_ALPHANUMERIC, p, e) | |
1925 | #define isASCII_utf8_safe(p, e) \ | |
1926 | /* Because ASCII is invariant under utf8, the non-utf8 macro \ | |
1927 | * works */ \ | |
1928 | (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p))) | |
1929 | #define isBLANK_utf8_safe(p, e) \ | |
1930 | _generic_non_swash_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e) | |
1931 | ||
e8fa43e2 KW |
1932 | #ifdef EBCDIC |
1933 | /* Because all controls are UTF-8 invariants in EBCDIC, we can use this | |
1934 | * more efficient macro instead of the more general one */ | |
da8c1a98 | 1935 | # define isCNTRL_utf8_safe(p, e) \ |
56d02b8c | 1936 | (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p))) |
e8fa43e2 | 1937 | #else |
da8c1a98 | 1938 | # define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0) |
e8fa43e2 KW |
1939 | #endif |
1940 | ||
da8c1a98 KW |
1941 | #define isDIGIT_utf8_safe(p, e) \ |
1942 | _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \ | |
1943 | _is_utf8_FOO_with_len(_CC_DIGIT, p, e)) | |
1944 | #define isGRAPH_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_GRAPH, p, e) | |
1945 | #define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \ | |
1946 | _is_utf8_perl_idcont_with_len, p, e) | |
e5dcd934 | 1947 | |
c11ff943 KW |
1948 | /* To prevent S_scan_word in toke.c from hanging, we have to make sure that |
1949 | * IDFIRST is an alnum. See | |
c8362b00 | 1950 | * http://rt.perl.org/rt3/Ticket/Display.html?id=74022 for more detail than you |
f91dcd13 KW |
1951 | * ever wanted to know about. (In the ASCII range, there isn't a difference.) |
1952 | * This used to be not the XID version, but we decided to go with the more | |
1953 | * modern Unicode definition */ | |
da8c1a98 KW |
1954 | #define isIDFIRST_utf8_safe(p, e) \ |
1955 | _generic_func_utf8_safe(_CC_IDFIRST, \ | |
1956 | _is_utf8_perl_idstart_with_len, (U8 *) (p), (U8 *) (e)) | |
1957 | ||
1958 | #define isLOWER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_LOWER, p, e) | |
1959 | #define isPRINT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PRINT, p, e) | |
1960 | #define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e) | |
1961 | #define isPUNCT_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_PUNCT, p, e) | |
1962 | #define isSPACE_utf8_safe(p, e) \ | |
1963 | _generic_non_swash_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e) | |
1964 | #define isUPPER_utf8_safe(p, e) _generic_swash_utf8_safe(_CC_UPPER, p, e) | |
1965 | #define isVERTWS_utf8_safe(p, e) \ | |
1966 | _generic_non_swash_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e) | |
1967 | #define isWORDCHAR_utf8_safe(p, e) \ | |
1968 | _generic_swash_utf8_safe(_CC_WORDCHAR, p, e) | |
1969 | #define isXDIGIT_utf8_safe(p, e) \ | |
1970 | _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \ | |
1971 | (UNLIKELY((e) - (p) < UTF8SKIP(p)) \ | |
1972 | ? (_force_out_malformed_utf8_message( \ | |
1973 | (U8 *) (p), (U8 *) (e), 0, 1), 0) \ | |
1974 | : is_XDIGIT_high(p))) | |
a0ed51b3 | 1975 | |
7d2b61dd | 1976 | #define toFOLD_utf8(p,s,l) to_utf8_fold(p,s,l) |
509fb054 KW |
1977 | #define toLOWER_utf8(p,s,l) to_utf8_lower(p,s,l) |
1978 | #define toTITLE_utf8(p,s,l) to_utf8_title(p,s,l) | |
1979 | #define toUPPER_utf8(p,s,l) to_utf8_upper(p,s,l) | |
2e8adce6 | 1980 | |
567b353c | 1981 | /* For internal core use only, subject to change */ |
607313a1 KW |
1982 | #define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f, "", 0) |
1983 | #define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f, "", 0) | |
1984 | #define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f, "", 0) | |
1985 | #define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f, "", 0) | |
a1a5ec35 KW |
1986 | |
1987 | #define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL) | |
1988 | #define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0) | |
1989 | #define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0) | |
1990 | #define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0) | |
567b353c | 1991 | |
4650c663 KW |
1992 | /* For internal core Perl use only: the base macros for defining macros like |
1993 | * isALPHA_LC_utf8. These are like _generic_utf8, but if the first code point | |
1994 | * in 'p' is within the 0-255 range, it uses locale rules from the passed-in | |
1995 | * 'macro' parameter */ | |
34aeb2e9 KW |
1996 | #define _generic_LC_utf8(name, p) _base_generic_utf8(name, name, p, 1) |
1997 | ||
1998 | #define isALPHA_LC_utf8(p) _generic_LC_utf8(ALPHA, p) | |
1999 | #define isALPHANUMERIC_LC_utf8(p) _generic_LC_utf8(ALPHANUMERIC, p) | |
2000 | #define isASCII_LC_utf8(p) _generic_LC_utf8(ASCII, p) | |
2001 | #define isBLANK_LC_utf8(p) _generic_LC_utf8(BLANK, p) | |
2002 | #define isCNTRL_LC_utf8(p) _generic_LC_utf8(CNTRL, p) | |
2003 | #define isDIGIT_LC_utf8(p) _generic_LC_utf8(DIGIT, p) | |
2004 | #define isGRAPH_LC_utf8(p) _generic_LC_utf8(GRAPH, p) | |
2005 | #define isIDCONT_LC_utf8(p) _generic_LC_utf8(IDCONT, p) | |
2006 | #define isIDFIRST_LC_utf8(p) _generic_LC_utf8(IDFIRST, p) | |
2007 | #define isLOWER_LC_utf8(p) _generic_LC_utf8(LOWER, p) | |
2008 | #define isPRINT_LC_utf8(p) _generic_LC_utf8(PRINT, p) | |
2009 | #define isPSXSPC_LC_utf8(p) _generic_LC_utf8(PSXSPC, p) | |
2010 | #define isPUNCT_LC_utf8(p) _generic_LC_utf8(PUNCT, p) | |
2011 | #define isSPACE_LC_utf8(p) _generic_LC_utf8(SPACE, p) | |
2012 | #define isUPPER_LC_utf8(p) _generic_LC_utf8(UPPER, p) | |
2013 | #define isWORDCHAR_LC_utf8(p) _generic_LC_utf8(WORDCHAR, p) | |
2014 | #define isXDIGIT_LC_utf8(p) _generic_LC_utf8(XDIGIT, p) | |
2015 | ||
da8c1a98 KW |
2016 | /* For internal core Perl use only: the base macros for defining macros like |
2017 | * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code | |
2018 | * point in 'p' is within the 0-255 range, it uses locale rules from the | |
2019 | * passed-in 'macro' parameter */ | |
2020 | #define _generic_LC_utf8_safe(macro, p, e, above_latin1) \ | |
2021 | (__ASSERT_(_utf8_safe_assert(p, e)) \ | |
2022 | (UTF8_IS_INVARIANT(*(p))) \ | |
2023 | ? macro(*(p)) \ | |
2024 | : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \ | |
2025 | ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \ | |
2026 | ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \ | |
2027 | : (_force_out_malformed_utf8_message( \ | |
2028 | (U8 *) (p), (U8 *) (e), 0, 1), 0)) \ | |
2029 | : above_latin1)) | |
2030 | ||
2031 | #define _generic_LC_swash_utf8_safe(macro, classnum, p, e) \ | |
2032 | _generic_LC_utf8_safe(macro, p, e, \ | |
2033 | _is_utf8_FOO_with_len(classnum, p, e)) | |
2034 | ||
2035 | #define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \ | |
2036 | _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e)) | |
2037 | ||
2038 | #define _generic_LC_non_swash_utf8_safe(classnum, above_latin1, p, e) \ | |
2039 | _generic_LC_utf8_safe(classnum, p, e, \ | |
2040 | (UNLIKELY((e) - (p) < UTF8SKIP(p)) \ | |
2041 | ? (_force_out_malformed_utf8_message( \ | |
2042 | (U8 *) (p), (U8 *) (e), 0, 1), 0) \ | |
2043 | : above_latin1(p))) | |
2044 | ||
2045 | #define isALPHANUMERIC_LC_utf8_safe(p, e) \ | |
2046 | _generic_LC_swash_utf8_safe(isALPHANUMERIC_LC, \ | |
2047 | _CC_ALPHANUMERIC, p, e) | |
2048 | #define isALPHA_LC_utf8_safe(p, e) \ | |
2049 | _generic_LC_swash_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e) | |
2050 | #define isASCII_LC_utf8_safe(p, e) \ | |
2051 | (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p))) | |
2052 | #define isBLANK_LC_utf8_safe(p, e) \ | |
2053 | _generic_LC_non_swash_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e) | |
2054 | #define isCNTRL_LC_utf8_safe(p, e) \ | |
2055 | _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0) | |
2056 | #define isDIGIT_LC_utf8_safe(p, e) \ | |
2057 | _generic_LC_swash_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e) | |
2058 | #define isGRAPH_LC_utf8_safe(p, e) \ | |
2059 | _generic_LC_swash_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e) | |
2060 | #define isIDCONT_LC_utf8_safe(p, e) \ | |
2061 | _generic_LC_func_utf8_safe(isIDCONT_LC, \ | |
2062 | _is_utf8_perl_idcont_with_len, p, e) | |
2063 | #define isIDFIRST_LC_utf8_safe(p, e) \ | |
2064 | _generic_LC_func_utf8_safe(isIDFIRST_LC, \ | |
2065 | _is_utf8_perl_idstart_with_len, p, e) | |
2066 | #define isLOWER_LC_utf8_safe(p, e) \ | |
2067 | _generic_LC_swash_utf8_safe(isLOWER_LC, _CC_LOWER, p, e) | |
2068 | #define isPRINT_LC_utf8_safe(p, e) \ | |
2069 | _generic_LC_swash_utf8_safe(isPRINT_LC, _CC_PRINT, p, e) | |
2070 | #define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e) | |
2071 | #define isPUNCT_LC_utf8_safe(p, e) \ | |
2072 | _generic_LC_swash_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e) | |
2073 | #define isSPACE_LC_utf8_safe(p, e) \ | |
2074 | _generic_LC_non_swash_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e) | |
2075 | #define isUPPER_LC_utf8_safe(p, e) \ | |
2076 | _generic_LC_swash_utf8_safe(isUPPER_LC, _CC_UPPER, p, e) | |
2077 | #define isWORDCHAR_LC_utf8_safe(p, e) \ | |
2078 | _generic_LC_swash_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e) | |
2079 | #define isXDIGIT_LC_utf8_safe(p, e) \ | |
2080 | _generic_LC_non_swash_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e) | |
aaa51d5e | 2081 | |
fbc19f27 KW |
2082 | /* Macros for backwards compatibility and for completeness when the ASCII and |
2083 | * Latin1 values are identical */ | |
b7d90381 KW |
2084 | #define isALPHAU(c) isALPHA_L1(c) |
2085 | #define isDIGIT_L1(c) isDIGIT_A(c) | |
2086 | #define isOCTAL(c) isOCTAL_A(c) | |
2087 | #define isOCTAL_L1(c) isOCTAL_A(c) | |
2088 | #define isXDIGIT_L1(c) isXDIGIT_A(c) | |
2089 | #define isALNUM(c) isWORDCHAR(c) | |
a377c856 | 2090 | #define isALNUM_A(c) isALNUM(c) |
b7d90381 KW |
2091 | #define isALNUMU(c) isWORDCHAR_L1(c) |
2092 | #define isALNUM_LC(c) isWORDCHAR_LC(c) | |
2093 | #define isALNUM_uni(c) isWORDCHAR_uni(c) | |
2e28f0b9 | 2094 | #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c) |
b7d90381 KW |
2095 | #define isALNUM_utf8(p) isWORDCHAR_utf8(p) |
2096 | #define isALNUM_LC_utf8(p) isWORDCHAR_LC_utf8(p) | |
2097 | #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */ | |
2098 | #define isALNUMC_L1(c) isALPHANUMERIC_L1(c) | |
2099 | #define isALNUMC(c) isALPHANUMERIC(c) | |
2100 | #define isALNUMC_LC(c) isALPHANUMERIC_LC(c) | |
2101 | #define isALNUMC_uni(c) isALPHANUMERIC_uni(c) | |
15861f94 | 2102 | #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c) |
b7d90381 | 2103 | #define isALNUMC_utf8(p) isALPHANUMERIC_utf8(p) |
15861f94 | 2104 | #define isALNUMC_LC_utf8(p) isALPHANUMERIC_LC_utf8(p) |
fbc19f27 | 2105 | |
2bd1cbf6 KW |
2106 | /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII, |
2107 | * except that they don't necessarily mean the same characters, e.g. CTRL-D is | |
2108 | * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC. | |
2109 | * '?' is special-cased on EBCDIC to APC, which is the control there that is | |
2110 | * the outlier from the block that contains the other controls, just like | |
2111 | * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0 | |
2112 | * block. If it weren't special cased, it would yield a non-control. | |
88794300 KW |
2113 | * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D, |
2114 | * etc. */ | |
2bd1cbf6 | 2115 | #ifndef EBCDIC |
75763b3a | 2116 | # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64) |
2bd1cbf6 | 2117 | #else |
75763b3a KW |
2118 | # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ |
2119 | ((isPRINT_A(c)) \ | |
2120 | ? (UNLIKELY((c) == '?') \ | |
2121 | ? QUESTION_MARK_CTRL \ | |
2122 | : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \ | |
2123 | : (UNLIKELY((c) == QUESTION_MARK_CTRL) \ | |
2124 | ? '?' \ | |
2125 | : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64))))) | |
2bd1cbf6 | 2126 | #endif |
bbce6d69 | 2127 | |
dea28490 JJ |
2128 | /* Line numbers are unsigned, 32 bits. */ |
2129 | typedef U32 line_t; | |
e5dcd934 | 2130 | #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */ |
378cc40b | 2131 | |
91152fc1 DG |
2132 | /* Helpful alias for version prescan */ |
2133 | #define is_LAX_VERSION(a,b) \ | |
2134 | (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL)) | |
2135 | ||
2136 | #define is_STRICT_VERSION(a,b) \ | |
2137 | (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL)) | |
2138 | ||
2139 | #define BADVERSION(a,b,c) \ | |
2140 | if (b) { \ | |
2141 | *b = c; \ | |
2142 | } \ | |
2143 | return a; | |
8c52afec | 2144 | |
cb27eebd KW |
2145 | /* Converts a character known to represent a hexadecimal digit (0-9, A-F, or |
2146 | * a-f) to its numeric value. READ_XDIGIT's argument is a string pointer, | |
2147 | * which is advanced. The input is validated only by an assert() in DEBUGGING | |
2148 | * builds. In both ASCII and EBCDIC the last 4 bits of the digits are 0-9; and | |
2149 | * the last 4 bits of A-F and a-f are 1-6, so adding 9 yields 10-15 */ | |
2150 | #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) (0xf & (isDIGIT(c) \ | |
2151 | ? (c) \ | |
2152 | : ((c) + 9)))) | |
902008b8 KW |
2153 | #define READ_XDIGIT(s) (__ASSERT_(isXDIGIT(*s)) (0xf & (isDIGIT(*(s)) \ |
2154 | ? (*(s)++) \ | |
2155 | : (*(s)++ + 9)))) | |
95a59cab | 2156 | |
cb27eebd KW |
2157 | /* Converts a character known to represent an octal digit (0-7) to its numeric |
2158 | * value. The input is validated only by an assert() in DEBUGGING builds. In | |
2159 | * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */ | |
2160 | #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c))) | |
2161 | ||
305b8651 KW |
2162 | /* Efficiently returns a boolean as to if two native characters are equivalent |
2163 | * case-insenstively. At least one of the characters must be one of [A-Za-z]; | |
2164 | * the ALPHA in the name is to remind you of that. This is asserted() in | |
2165 | * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro | |
2166 | * works (on valid input) for both non- and UTF-8-encoded bytes. | |
2167 | * | |
2168 | * When one of the inputs is a compile-time constant and gets folded by the | |
2169 | * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII | |
2170 | * machines, 'A' and 'a' differ by a single bit; the same with the upper and | |
2171 | * lower case of all other ASCII-range alphabetics. On ASCII platforms, they | |
96ca48da KW |
2172 | * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an |
2173 | * exclusive 'or' to find that bit and then inverts it to form a mask, with | |
2174 | * just a single 0, in the bit position where the upper- and lowercase differ. | |
2175 | * */ | |
305b8651 KW |
2176 | #define isALPHA_FOLD_EQ(c1, c2) \ |
2177 | (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \ | |
2178 | ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a'))) | |
2179 | #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2))) | |
2180 | ||
8e84507e | 2181 | /* |
ccfc67b7 JH |
2182 | =head1 Memory Management |
2183 | ||
a02a5408 | 2184 | =for apidoc Am|void|Newx|void* ptr|int nitems|type |
954c1994 GS |
2185 | The XSUB-writer's interface to the C C<malloc> function. |
2186 | ||
596f7718 | 2187 | Memory obtained by this should B<ONLY> be freed with L</"Safefree">. |
0d7b2759 | 2188 | |
c5008215 JC |
2189 | In 5.9.3, Newx() and friends replace the older New() API, and drops |
2190 | the first parameter, I<x>, a debug aid which allowed callers to identify | |
37b8b4c9 | 2191 | themselves. This aid has been superseded by a new build option, |
d10b4965 | 2192 | PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still |
c5008215 JC |
2193 | there for use in XS modules supporting older perls. |
2194 | ||
a02a5408 | 2195 | =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast |
954c1994 | 2196 | The XSUB-writer's interface to the C C<malloc> function, with |
fbe13c60 | 2197 | cast. See also C<L</Newx>>. |
954c1994 | 2198 | |
596f7718 | 2199 | Memory obtained by this should B<ONLY> be freed with L</"Safefree">. |
0d7b2759 | 2200 | |
a02a5408 | 2201 | =for apidoc Am|void|Newxz|void* ptr|int nitems|type |
954c1994 | 2202 | The XSUB-writer's interface to the C C<malloc> function. The allocated |
fbe13c60 | 2203 | memory is zeroed with C<memzero>. See also C<L</Newx>>. |
a02a5408 | 2204 | |
596f7718 | 2205 | Memory obtained by this should B<ONLY> be freed with L</"Safefree">. |
0d7b2759 | 2206 | |
954c1994 GS |
2207 | =for apidoc Am|void|Renew|void* ptr|int nitems|type |
2208 | The XSUB-writer's interface to the C C<realloc> function. | |
2209 | ||
596f7718 | 2210 | Memory obtained by this should B<ONLY> be freed with L</"Safefree">. |
0d7b2759 | 2211 | |
954c1994 GS |
2212 | =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast |
2213 | The XSUB-writer's interface to the C C<realloc> function, with | |
2214 | cast. | |
2215 | ||
596f7718 | 2216 | Memory obtained by this should B<ONLY> be freed with L</"Safefree">. |
0d7b2759 | 2217 | |
49b8b560 | 2218 | =for apidoc Am|void|Safefree|void* ptr |
954c1994 GS |
2219 | The XSUB-writer's interface to the C C<free> function. |
2220 | ||
596f7718 | 2221 | This should B<ONLY> be used on memory obtained using L</"Newx"> and friends. |
0d7b2759 | 2222 | |
954c1994 GS |
2223 | =for apidoc Am|void|Move|void* src|void* dest|int nitems|type |
2224 | The XSUB-writer's interface to the C C<memmove> function. The C<src> is the | |
926bb54c | 2225 | source, C<dest> is the destination, C<nitems> is the number of items, and |
fbe13c60 | 2226 | C<type> is the type. Can do overlapping moves. See also C<L</Copy>>. |
954c1994 | 2227 | |
e90e2364 | 2228 | =for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type |
796b6530 | 2229 | Like C<Move> but returns C<dest>. Useful |
72d33970 | 2230 | for encouraging compilers to tail-call |
e90e2364 NC |
2231 | optimise. |
2232 | ||
954c1994 GS |
2233 | =for apidoc Am|void|Copy|void* src|void* dest|int nitems|type |
2234 | The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the | |
926bb54c | 2235 | source, C<dest> is the destination, C<nitems> is the number of items, and |
fbe13c60 | 2236 | C<type> is the type. May fail on overlapping copies. See also C<L</Move>>. |
954c1994 | 2237 | |
e90e2364 NC |
2238 | =for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type |
2239 | ||
796b6530 | 2240 | Like C<Copy> but returns C<dest>. Useful |
72d33970 | 2241 | for encouraging compilers to tail-call |
e90e2364 NC |
2242 | optimise. |
2243 | ||
954c1994 GS |
2244 | =for apidoc Am|void|Zero|void* dest|int nitems|type |
2245 | ||
2246 | The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the | |
2247 | destination, C<nitems> is the number of items, and C<type> is the type. | |
2248 | ||
e90e2364 NC |
2249 | =for apidoc Am|void *|ZeroD|void* dest|int nitems|type |
2250 | ||
72d33970 FC |
2251 | Like C<Zero> but returns dest. Useful |
2252 | for encouraging compilers to tail-call | |
e90e2364 NC |
2253 | optimise. |
2254 | ||
da5d8dbb | 2255 | =for apidoc Am|void|StructCopy|type *src|type *dest|type |
4375e838 | 2256 | This is an architecture-independent macro to copy one structure to another. |
954c1994 | 2257 | |
7e337ee0 JH |
2258 | =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte |
2259 | ||
2260 | Fill up memory with a byte pattern (a byte repeated over and over | |
2261 | again) that hopefully catches attempts to access uninitialized memory. | |
2262 | ||
2263 | =for apidoc Am|void|PoisonNew|void* dest|int nitems|type | |
2264 | ||
2265 | PoisonWith(0xAB) for catching access to allocated but uninitialized memory. | |
2266 | ||
1c12ffb4 | 2267 | =for apidoc Am|void|PoisonFree|void* dest|int nitems|type |
7e337ee0 JH |
2268 | |
2269 | PoisonWith(0xEF) for catching access to freed memory. | |
2270 | ||
9965345d JH |
2271 | =for apidoc Am|void|Poison|void* dest|int nitems|type |
2272 | ||
7e337ee0 | 2273 | PoisonWith(0xEF) for catching access to freed memory. |
9965345d JH |
2274 | |
2275 | =cut */ | |
954c1994 | 2276 | |
561b68a9 SH |
2277 | /* Maintained for backwards-compatibility only. Use newSV() instead. */ |
2278 | #ifndef PERL_CORE | |
ff06c60c | 2279 | #define NEWSV(x,len) newSV(len) |
561b68a9 | 2280 | #endif |
ff06c60c | 2281 | |
b7112dce | 2282 | #define MEM_SIZE_MAX ((MEM_SIZE)-1) |
19a94d75 | 2283 | |
a500027b | 2284 | #define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1)) |
e6bdf523 | 2285 | |
27d5b266 | 2286 | #ifdef PERL_MALLOC_WRAP |
e6bdf523 DM |
2287 | |
2288 | /* This expression will be constant-folded at compile time. It checks | |
2289 | * whether or not the type of the count n is so small (e.g. U8 or U16, or | |
2290 | * U32 on 64-bit systems) that there's no way a wrap-around could occur. | |
2291 | * As well as avoiding the need for a run-time check in some cases, it's | |
2292 | * designed to avoid compiler warnings like: | |
2293 | * comparison is always false due to limited range of data type | |
73e8ff00 DM |
2294 | * It's mathematically equivalent to |
2295 | * max(n) * sizeof(t) > MEM_SIZE_MAX | |
e6bdf523 DM |
2296 | */ |
2297 | ||
2298 | # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \ | |
445198b9 LM |
2299 | ( sizeof(MEM_SIZE) < sizeof(n) \ |
2300 | || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n)))) | |
e6bdf523 | 2301 | |
88f9f128 | 2302 | /* This is written in a slightly odd way to avoid various spurious |
d98e5cde DM |
2303 | * compiler warnings. We *want* to write the expression as |
2304 | * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C) | |
2305 | * (for some compile-time constant C), but even when the LHS | |
2306 | * constant-folds to false at compile-time, g++ insists on emitting | |
2307 | * warnings about the RHS (e.g. "comparison is always false"), so instead | |
2308 | * we write it as | |
e6bdf523 | 2309 | * |
d98e5cde | 2310 | * (cond ? n : X) > C |
88f9f128 | 2311 | * |
d98e5cde DM |
2312 | * where X is a constant with X > C always false. Choosing a value for X |
2313 | * is tricky. If 0, some compilers will complain about 0 > C always being | |
2314 | * false; if 1, Coverity complains when n happens to be the constant value | |
2315 | * '1', that cond ? 1 : 1 has the same value on both branches; so use C | |
2316 | * for X and hope that nothing else whines. | |
e6bdf523 DM |
2317 | */ |
2318 | ||
2319 | # define _MEM_WRAP_WILL_WRAP(n,t) \ | |
88f9f128 DM |
2320 | ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \ |
2321 | MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t)) | |
e6bdf523 DM |
2322 | |
2323 | # define MEM_WRAP_CHECK(n,t) \ | |
2324 | (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \ | |
2325 | && (croak_memory_wrap(),0)) | |
2326 | ||
2327 | # define MEM_WRAP_CHECK_1(n,t,a) \ | |
2328 | (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \ | |
2329 | && (Perl_croak_nocontext("%s",(a)),0)) | |
2330 | ||
814eedc8 DD |
2331 | /* "a" arg must be a string literal */ |
2332 | # define MEM_WRAP_CHECK_s(n,t,a) \ | |
2333 | (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \ | |
2334 | && (Perl_croak_nocontext("" a ""),0)) | |
2335 | ||
8b44ba4c | 2336 | #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t), |
27d5b266 | 2337 | |
a500027b | 2338 | #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)) |
27d5b266 JH |
2339 | #else |
2340 | ||
410319be NC |
2341 | #define MEM_WRAP_CHECK(n,t) |
2342 | #define MEM_WRAP_CHECK_1(n,t,a) | |
814eedc8 | 2343 | #define MEM_WRAP_CHECK_s(n,t,a) |
8b44ba4c NC |
2344 | #define MEM_WRAP_CHECK_(n,t) |
2345 | ||
a500027b | 2346 | #define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n) |
27d5b266 | 2347 | |
1936d2a7 | 2348 | #endif |
8b44ba4c | 2349 | |
fe4f188c | 2350 | #ifdef PERL_MEM_LOG |
46c6c7e2 | 2351 | /* |
9f653bb5 | 2352 | * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s |
46c6c7e2 JH |
2353 | * go through functions, which are handy for debugging breakpoints, but |
2354 | * which more importantly get the immediate calling environment (file and | |
e352bcff JH |
2355 | * line number, and C function name if available) passed in. This info can |
2356 | * then be used for logging the calls, for which one gets a sample | |
73d1d973 | 2357 | * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined. |
3609ea0d | 2358 | * |
46c6c7e2 | 2359 | * Known problems: |
94e892a6 | 2360 | * - not all memory allocs get logged, only those |
46c6c7e2 | 2361 | * that go through Newx() and derivatives (while all |
94e892a6 | 2362 | * Safefrees do get logged) |
46c6c7e2 JH |
2363 | * - __FILE__ and __LINE__ do not work everywhere |
2364 | * - __func__ or __FUNCTION__ even less so | |
2365 | * - I think more goes on after the perlio frees but | |
2366 | * the thing is that STDERR gets closed (as do all | |
2367 | * the file descriptors) | |
2368 | * - no deeper calling stack than the caller of the Newx() | |
2369 | * or the kind, but do I look like a C reflection/introspection | |
2370 | * utility to you? | |
2371 | * - the function prototypes for the logging functions | |
2372 | * probably should maybe be somewhere else than handy.h | |
2373 | * - one could consider inlining (macrofying) the logging | |
2374 | * for speed, but I am too lazy | |
2375 | * - one could imagine recording the allocations in a hash, | |
2376 | * (keyed by the allocation address?), and maintain that | |
2377 | * through reallocs and frees, but how to do that without | |
2378 | * any News() happening...? | |
73d1d973 | 2379 | * - lots of -Ddefines to get useful/controllable output |
b953482e | 2380 | * - lots of ENV reads |
46c6c7e2 JH |
2381 | */ |
2382 | ||
0b0ab801 | 2383 | # ifdef PERL_CORE |
73d1d973 | 2384 | # ifndef PERL_MEM_LOG_NOIMPL |
0b0ab801 MHM |
2385 | enum mem_log_type { |
2386 | MLT_ALLOC, | |
2387 | MLT_REALLOC, | |
d7a2c63c MHM |
2388 | MLT_FREE, |
2389 | MLT_NEW_SV, | |
2390 | MLT_DEL_SV | |
0b0ab801 MHM |
2391 | }; |
2392 | # endif | |
12754f92 | 2393 | # if defined(PERL_IN_SV_C) /* those are only used in sv.c */ |
d7a2c63c MHM |
2394 | void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname); |
2395 | void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname); | |
12754f92 | 2396 | # endif |
0b0ab801 MHM |
2397 | # endif |
2398 | ||
fe4f188c JH |
2399 | #endif |
2400 | ||
2401 | #ifdef PERL_MEM_LOG | |
d1401ee9 MHM |
2402 | #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__) |
2403 | #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__) | |
46c6c7e2 | 2404 | #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__) |
fe4f188c JH |
2405 | #endif |
2406 | ||
2407 | #ifndef MEM_LOG_ALLOC | |
2408 | #define MEM_LOG_ALLOC(n,t,a) (a) | |
2409 | #endif | |
2410 | #ifndef MEM_LOG_REALLOC | |
2411 | #define MEM_LOG_REALLOC(n,t,v,a) (a) | |
2412 | #endif | |
2413 | #ifndef MEM_LOG_FREE | |
2414 | #define MEM_LOG_FREE(a) (a) | |
2415 | #endif | |
2416 | ||
d1401ee9 MHM |
2417 | #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t)))))) |
2418 | #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t)))))) | |
2419 | #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t))))) | |
a6f6820f NC |
2420 | |
2421 | #ifndef PERL_CORE | |
a02a5408 JC |
2422 | /* pre 5.9.x compatibility */ |
2423 | #define New(x,v,n,t) Newx(v,n,t) | |
2424 | #define Newc(x,v,n,t,c) Newxc(v,n,t,c) | |
4541904d | 2425 | #define Newz(x,v,n,t) Newxz(v,n,t) |
a6f6820f | 2426 | #endif |
a02a5408 | 2427 | |
ff68c719 | 2428 | #define Renew(v,n,t) \ |
d1401ee9 | 2429 | (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t)))))) |
ff68c719 | 2430 | #define Renewc(v,n,t,c) \ |
d1401ee9 | 2431 | (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t)))))) |
94010e71 NC |
2432 | |
2433 | #ifdef PERL_POISON | |
2434 | #define Safefree(d) \ | |
06c0cc96 | 2435 | ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0) |
94010e71 | 2436 | #else |
fe4f188c | 2437 | #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d))) |
94010e71 | 2438 | #endif |
55497cff | 2439 | |
dbb57106 YO |
2440 | /* assert that a valid ptr has been supplied - use this instead of assert(ptr) * |
2441 | * as it handles cases like constant string arguments without throwing warnings * | |
2442 | * the cast is required, as is the inequality check, to avoid warnings */ | |
45908e4d | 2443 | #define perl_assert_ptr(p) assert( ((void*)(p)) != 0 ) |
55497cff | 2444 | |
45908e4d YO |
2445 | |
2446 | #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))) | |
2447 | #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))) | |
2448 | #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t))) | |
2449 | ||
bdd1531d | 2450 | /* Like above, but returns a pointer to 'd' */ |
45908e4d YO |
2451 | #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))) |
2452 | #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))) | |
45908e4d | 2453 | #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t))) |
e90e2364 | 2454 | |
7e337ee0 JH |
2455 | #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t))) |
2456 | #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB) | |
2457 | #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF) | |
2458 | #define Poison(d,n,t) PoisonFree(d,n,t) | |
27d5b266 | 2459 | |
caa674f3 DD |
2460 | #ifdef PERL_POISON |
2461 | # define PERL_POISON_EXPR(x) x | |
2462 | #else | |
2463 | # define PERL_POISON_EXPR(x) | |
2464 | #endif | |
2465 | ||
ff68c719 | 2466 | #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s))) |
2cc61e15 | 2467 | |
c3caa5c3 JH |
2468 | /* C_ARRAY_LENGTH is the number of elements in the C array (so you |
2469 | * want your zero-based indices to be less than but not equal to). | |
2470 | * | |
2471 | * C_ARRAY_END is one past the last: half-open/half-closed range, | |
2472 | * not last-inclusive range. */ | |
622913ab | 2473 | #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0])) |
c3caa5c3 | 2474 | #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a)) |
622913ab | 2475 | |
2cc61e15 DD |
2476 | #ifdef NEED_VA_COPY |
2477 | # ifdef va_copy | |
2478 | # define Perl_va_copy(s, d) va_copy(d, s) | |
07798b17 AC |
2479 | # elif defined(__va_copy) |
2480 | # define Perl_va_copy(s, d) __va_copy(d, s) | |
2cc61e15 | 2481 | # else |
07798b17 | 2482 | # define Perl_va_copy(s, d) Copy(s, d, 1, va_list) |
2cc61e15 DD |
2483 | # endif |
2484 | #endif | |
2485 | ||
472d47bc SB |
2486 | /* convenience debug macros */ |
2487 | #ifdef USE_ITHREADS | |
2488 | #define pTHX_FORMAT "Perl interpreter: 0x%p" | |
2489 | #define pTHX__FORMAT ", Perl interpreter: 0x%p" | |
f54cb97a AL |
2490 | #define pTHX_VALUE_ (void *)my_perl, |
2491 | #define pTHX_VALUE (void *)my_perl | |
2492 | #define pTHX__VALUE_ ,(void *)my_perl, | |
2493 | #define pTHX__VALUE ,(void *)my_perl | |
472d47bc | 2494 | #else |
3609ea0d | 2495 | #define pTHX_FORMAT |
472d47bc | 2496 | #define pTHX__FORMAT |
3609ea0d | 2497 | #define pTHX_VALUE_ |
472d47bc | 2498 | #define pTHX_VALUE |
3609ea0d | 2499 | #define pTHX__VALUE_ |
472d47bc SB |
2500 | #define pTHX__VALUE |
2501 | #endif /* USE_ITHREADS */ | |
3609ea0d | 2502 | |
2acdbac1 NC |
2503 | /* Perl_deprecate was not part of the public API, and did not have a deprecate() |
2504 | shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor | |
2505 | CPAN::Unpack show any users outside the core. */ | |
2506 | #ifdef PERL_CORE | |
dc6e8de0 A |
2507 | # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \ |
2508 | "Use of " s " is deprecated") | |
c9680906 A |
2509 | # define deprecate_disappears_in(when,message) \ |
2510 | Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \ | |
2511 | message ", and will disappear in Perl " when) | |
ac641426 A |
2512 | # define deprecate_fatal_in(when,message) \ |
2513 | Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \ | |
2514 | message ". Its use will be fatal in Perl " when) | |
2acdbac1 NC |
2515 | #endif |
2516 | ||
dfff4baf BF |
2517 | /* Internal macros to deal with gids and uids */ |
2518 | #ifdef PERL_CORE | |
2519 | ||
2520 | # if Uid_t_size > IVSIZE | |
2521 | # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid)) | |
2522 | # define SvUID(sv) SvNV(sv) | |
07798b17 AC |
2523 | # elif Uid_t_sign <= 0 |
2524 | # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid)) | |
2525 | # define SvUID(sv) SvIV(sv) | |
dfff4baf | 2526 | # else |
07798b17 AC |
2527 | # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid)) |
2528 | # define SvUID(sv) SvUV(sv) | |
dfff4baf BF |
2529 | # endif /* Uid_t_size */ |
2530 | ||
2531 | # if Gid_t_size > IVSIZE | |
2532 | # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid)) | |
2533 | # define SvGID(sv) SvNV(sv) | |
07798b17 AC |
2534 | # elif Gid_t_sign <= 0 |
2535 | # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid)) | |
2536 | # define SvGID(sv) SvIV(sv) | |
dfff4baf | 2537 | # else |
07798b17 AC |
2538 | # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid)) |
2539 | # define SvGID(sv) SvUV(sv) | |
dfff4baf BF |
2540 | # endif /* Gid_t_size */ |
2541 | ||
2542 | #endif | |
2543 | ||
6a5bc5ac | 2544 | #endif /* PERL_HANDY_H_ */ |
9d745869 | 2545 | |
e9a8c099 | 2546 | /* |
14d04a33 | 2547 | * ex: set ts=8 sts=4 sw=4 et: |
e9a8c099 | 2548 | */ |