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 | /* | |
3f620621 | 21 | =for apidoc_section $string |
78342678 | 22 | =for apidoc AmnU||Nullch |
72d33970 FC |
23 | Null character pointer. (No longer available when C<PERL_CORE> is |
24 | defined.) | |
2307c6d0 | 25 | |
3f620621 | 26 | =for apidoc_section $SV |
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 | |
51b56f5c KW |
31 | |
32 | Below are signatures of functions from config.h which can't easily be gleaned | |
33 | from it, and are very unlikely to change | |
34 | ||
3f620621 | 35 | =for apidoc_section $signals |
51b56f5c KW |
36 | =for apidoc Am|int|Sigsetjmp|jmp_buf env|int savesigs |
37 | =for apidoc Am|void|Siglongjmp|jmp_buf env|int val | |
38 | ||
3f620621 | 39 | =for apidoc_section $filesystem |
51b56f5c KW |
40 | =for apidoc Am|void *|FILE_ptr|FILE * f |
41 | =for apidoc Am|Size_t|FILE_cnt|FILE * f | |
42 | =for apidoc Am|void *|FILE_base|FILE * f | |
b290de04 | 43 | =for apidoc Am|Size_t|FILE_bufsiz|FILE *f |
51b56f5c | 44 | |
3f620621 | 45 | =for apidoc_section $string |
b290de04 KW |
46 | =for apidoc Amu|token|CAT2|token x|token y |
47 | =for apidoc Amu|string|STRINGIFY|token x | |
51b56f5c | 48 | |
3f620621 | 49 | =for apidoc_section $numeric |
b290de04 | 50 | =for apidoc Am|double|Drand01 |
51b56f5c KW |
51 | =for apidoc Am|void|seedDrand01|Rand_seed_t x |
52 | =for apidoc Am|char *|Gconvert|double x|Size_t n|bool t|char * b | |
53 | ||
54 | =cut | |
954c1994 GS |
55 | */ |
56 | ||
24792b8d NC |
57 | # define Nullch Null(char*) |
58 | # define Nullfp Null(PerlIO*) | |
59 | # define Nullsv Null(SV*) | |
60 | #endif | |
8d063cd8 | 61 | |
641d3f0b | 62 | #ifdef TRUE |
63 | #undef TRUE | |
64 | #endif | |
65 | #ifdef FALSE | |
66 | #undef FALSE | |
67 | #endif | |
68 | #define TRUE (1) | |
69 | #define FALSE (0) | |
70 | ||
df87895c | 71 | /* |
3f620621 | 72 | =for apidoc_section $SV |
1607e393 KW |
73 | =for apidoc Am |AV * |MUTABLE_AV |AV * p |
74 | =for apidoc_item |CV * |MUTABLE_CV |CV * p | |
75 | =for apidoc_item |GV * |MUTABLE_GV |GV * p | |
76 | =for apidoc_item |HV * |MUTABLE_HV |HV * p | |
77 | =for apidoc_item |IO * |MUTABLE_IO |IO * p | |
78 | =for apidoc_item |void *|MUTABLE_PTR|void * p | |
79 | =for apidoc_item |SV * |MUTABLE_SV |SV * p | |
df87895c KW |
80 | |
81 | The C<MUTABLE_I<*>>() macros cast pointers to the types shown, in such a way | |
82 | (compiler permitting) that casting away const-ness will give a warning; | |
83 | e.g.: | |
84 | ||
85 | const SV *sv = ...; | |
86 | AV *av1 = (AV*)sv; <== BAD: the const has been silently | |
87 | cast away | |
88 | AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn | |
89 | ||
54ba56f3 KW |
90 | C<MUTABLE_PTR> is the base macro used to derive new casts. The other |
91 | already-built-in ones return pointers to what their names indicate. | |
df87895c KW |
92 | |
93 | =cut | |
cf3f0ffb | 94 | |
ba04fd90 KW |
95 | The brace group version will raise a diagnostic if 'p' is const; the other |
96 | blindly casts away const. | |
97 | */ | |
041c1a23 | 98 | #if defined(PERL_USE_GCC_BRACE_GROUPS) |
6c2255e0 | 99 | # define MUTABLE_PTR(p) ({ void *p_ = (p); p_; }) |
b1bc3f34 NC |
100 | #else |
101 | # define MUTABLE_PTR(p) ((void *) (p)) | |
102 | #endif | |
103 | ||
a062e10d | 104 | #define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p)) |
ea726b52 | 105 | #define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p)) |
159b6efe | 106 | #define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p)) |
dbebbdb4 | 107 | #define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p)) |
a45c7426 | 108 | #define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p)) |
b1bc3f34 | 109 | #define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p)) |
27d4fb96 | 110 | |
b1c011dc | 111 | #ifndef __cplusplus |
bd31be4b | 112 | # include <stdbool.h> |
bd31be4b NC |
113 | #endif |
114 | ||
25ba28ce | 115 | /* |
3f620621 | 116 | =for apidoc_section $casting |
25ba28ce KW |
117 | =for apidoc Am|bool|cBOOL|bool expr |
118 | ||
9fa5fd96 KW |
119 | Cast-to-bool. When Perl was able to be compiled on pre-C99 compilers, a |
120 | C<(bool)> cast didn't necessarily do the right thing, so this macro was | |
121 | created (and made somewhat complicated to work around bugs in old | |
122 | compilers). Now, many years later, and C99 is used, this is no longer | |
123 | required, but is kept for backwards compatibility. | |
25ba28ce KW |
124 | |
125 | =cut | |
126 | */ | |
9fa5fd96 | 127 | #define cBOOL(cbool) ((bool) (cbool)) |
f2338a2e | 128 | |
46c6c7e2 | 129 | /* Try to figure out __func__ or __FUNCTION__ equivalent, if any. |
e352bcff JH |
130 | * XXX Should really be a Configure probe, with HAS__FUNCTION__ |
131 | * and FUNCTION__ as results. | |
132 | * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */ | |
46c6c7e2 JH |
133 | #if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */ |
134 | # define FUNCTION__ __func__ | |
f31da5f4 | 135 | # define SAFE_FUNCTION__ __func__ |
7adf2470 | 136 | #elif (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */ |
f31da5f4 YO |
137 | # define FUNCTION__ ("") |
138 | # define SAFE_FUNCTION__ ("UNKNOWN") | |
46c6c7e2 | 139 | #else |
07798b17 | 140 | # define FUNCTION__ __FUNCTION__ /* Common extension. */ |
f31da5f4 | 141 | # define SAFE_FUNCTION__ __FUNCTION__ /* Common extension. */ |
46c6c7e2 JH |
142 | #endif |
143 | ||
27d4fb96 | 144 | /* XXX A note on the perl source internal type system. The |
145 | original intent was that I32 be *exactly* 32 bits. | |
146 | ||
147 | Currently, we only guarantee that I32 is *at least* 32 bits. | |
148 | Specifically, if int is 64 bits, then so is I32. (This is the case | |
149 | for the Cray.) This has the advantage of meshing nicely with | |
150 | standard library calls (where we pass an I32 and the library is | |
151 | expecting an int), but the disadvantage that an I32 is not 32 bits. | |
152 | Andy Dougherty August 1996 | |
24fef2a7 | 153 | |
dc45a647 MB |
154 | There is no guarantee that there is *any* integral type with |
155 | exactly 32 bits. It is perfectly legal for a system to have | |
156 | sizeof(short) == sizeof(int) == sizeof(long) == 8. | |
693762b4 | 157 | |
dc45a647 MB |
158 | Similarly, there is no guarantee that I16 and U16 have exactly 16 |
159 | bits. | |
693762b4 | 160 | |
8e84507e NIS |
161 | For dealing with issues that may arise from various 32/64-bit |
162 | systems, we will ask Configure to check out | |
8175356b | 163 | |
1604cfb0 MS |
164 | SHORTSIZE == sizeof(short) |
165 | INTSIZE == sizeof(int) | |
166 | LONGSIZE == sizeof(long) | |
167 | LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG) | |
168 | PTRSIZE == sizeof(void *) | |
169 | DOUBLESIZE == sizeof(double) | |
170 | LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE). | |
8175356b | 171 | |
27d4fb96 | 172 | */ |
173 | ||
69512466 JH |
174 | #ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */ |
175 | # include <inttypes.h> | |
dd0eed91 JH |
176 | # ifdef INT32_MIN_BROKEN |
177 | # undef INT32_MIN | |
178 | # define INT32_MIN (-2147483647-1) | |
179 | # endif | |
180 | # ifdef INT64_MIN_BROKEN | |
181 | # undef INT64_MIN | |
182 | # define INT64_MIN (-9223372036854775807LL-1) | |
183 | # endif | |
69512466 JH |
184 | #endif |
185 | ||
8175356b JH |
186 | typedef I8TYPE I8; |
187 | typedef U8TYPE U8; | |
188 | typedef I16TYPE I16; | |
189 | typedef U16TYPE U16; | |
190 | typedef I32TYPE I32; | |
191 | typedef U32TYPE U32; | |
16d89be8 | 192 | |
74b807c7 | 193 | #ifdef QUADKIND |
8175356b JH |
194 | typedef I64TYPE I64; |
195 | typedef U64TYPE U64; | |
16d89be8 | 196 | #endif |
8175356b | 197 | |
5ff3f7a4 GS |
198 | /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type. |
199 | Please search CHAR_MAX in perl.h for further details. */ | |
26e4a9a9 KW |
200 | #ifdef UINT8_MAX |
201 | # define U8_MAX UINT8_MAX | |
0e983133 | 202 | #else |
26e4a9a9 | 203 | # define U8_MAX PERL_UCHAR_MAX |
0e983133 | 204 | #endif |
26e4a9a9 KW |
205 | #ifdef UINT8_MIN |
206 | # define U8_MIN UINT8_MIN | |
5ff3f7a4 | 207 | #else |
26e4a9a9 KW |
208 | # define U8_MIN PERL_UCHAR_MIN |
209 | #endif | |
5ff3f7a4 | 210 | |
26e4a9a9 KW |
211 | #ifdef INT16_MAX |
212 | # define I16_MAX INT16_MAX | |
79072805 | 213 | #else |
26e4a9a9 KW |
214 | # define I16_MAX PERL_SHORT_MAX |
215 | #endif | |
216 | #ifdef INT16_MIN | |
217 | # define I16_MIN INT16_MIN | |
218 | #else | |
219 | # define I16_MIN PERL_SHORT_MIN | |
220 | #endif | |
221 | #ifdef UINT16_MAX | |
222 | # define U16_MAX UINT16_MAX | |
223 | #else | |
224 | # define U16_MAX PERL_USHORT_MAX | |
225 | #endif | |
226 | #ifdef UINT16_MIN | |
227 | # define U16_MIN UINT16_MIN | |
228 | #else | |
229 | # define U16_MIN PERL_USHORT_MIN | |
79072805 LW |
230 | #endif |
231 | ||
26e4a9a9 KW |
232 | #ifdef INT32_MAX |
233 | # define I32_MAX INT32_MAX | |
234 | #elif LONGSIZE > 4 | |
235 | # define I32_MAX PERL_INT_MAX | |
236 | #else | |
237 | # define I32_MAX PERL_LONG_MAX | |
238 | #endif | |
239 | #ifdef INT32_MIN | |
240 | # define I32_MIN INT32_MIN | |
241 | #elif LONGSIZE > 4 | |
242 | # define I32_MIN PERL_INT_MIN | |
243 | #else | |
244 | # define I32_MIN PERL_LONG_MIN | |
245 | #endif | |
246 | #ifdef UINT32_MAX | |
247 | # ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */ | |
248 | # define U32_MAX UINT_MAX | |
249 | # else | |
250 | # define U32_MAX 4294967295U | |
251 | # endif | |
252 | #elif LONGSIZE > 4 | |
253 | # define U32_MAX PERL_UINT_MAX | |
254 | #else | |
255 | # define U32_MAX PERL_ULONG_MAX | |
256 | #endif | |
257 | #ifdef UINT32_MIN | |
258 | # define U32_MIN UINT32_MIN | |
259 | #elif LONGSIZE > 4 | |
260 | # define U32_MIN PERL_UINT_MIN | |
261 | #else | |
262 | # define U32_MIN PERL_ULONG_MIN | |
5ff3f7a4 GS |
263 | #endif |
264 | ||
23332c7d KW |
265 | /* |
266 | =for apidoc_section $integer | |
267 | =for apidoc Ay|| PERL_INT_FAST8_T | |
268 | =for apidoc_item PERL_INT_FAST16_T | |
269 | =for apidoc_item PERL_UINT_FAST8_T | |
270 | =for apidoc_item PERL_UINT_FAST16_T | |
271 | ||
272 | These are equivalent to the correspondingly-named C99 typedefs on platforms | |
273 | that have those; they evaluate to C<int> and C<unsigned int> on platforms that | |
274 | don't, so that you can portably take advantage of this C99 feature. | |
275 | ||
276 | =cut | |
277 | */ | |
934902b8 | 278 | #ifdef I_STDINT |
247cee9f KW |
279 | typedef int_fast8_t PERL_INT_FAST8_T; |
280 | typedef uint_fast8_t PERL_UINT_FAST8_T; | |
281 | typedef int_fast16_t PERL_INT_FAST16_T; | |
282 | typedef uint_fast16_t PERL_UINT_FAST16_T; | |
934902b8 | 283 | #else |
247cee9f KW |
284 | typedef int PERL_INT_FAST8_T; |
285 | typedef unsigned int PERL_UINT_FAST8_T; | |
286 | typedef int PERL_INT_FAST16_T; | |
287 | typedef unsigned int PERL_UINT_FAST16_T; | |
934902b8 | 288 | #endif |
247cee9f | 289 | |
464decb6 | 290 | /* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case |
ab350cbd KW |
291 | * anyone is grepping for it. So BIT_DIGITS gives the number of decimal digits |
292 | * required to represent any possible unsigned number containing N bits. | |
293 | * TYPE_DIGITS gives the number of decimal digits required to represent any | |
294 | * possible unsigned number of type T. */ | |
464decb6 | 295 | #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */ |
fc36a67e | 296 | #define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8) |
297 | #define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */ | |
298 | ||
88794300 | 299 | /* Unused by core; should be deprecated */ |
ff68c719 | 300 | #define Ctl(ch) ((ch) & 037) |
8d063cd8 | 301 | |
98fce2a4 KW |
302 | #if defined(PERL_CORE) || defined(PERL_EXT) |
303 | # ifndef MIN | |
304 | # define MIN(a,b) ((a) < (b) ? (a) : (b)) | |
305 | # endif | |
306 | # ifndef MAX | |
307 | # define MAX(a,b) ((a) > (b) ? (a) : (b)) | |
308 | # endif | |
309 | #endif | |
310 | ||
84ff4fa9 KW |
311 | /* Returns a boolean as to whether the input unsigned number is a power of 2 |
312 | * (2**0, 2**1, etc). In other words if it has just a single bit set. | |
313 | * If not, subtracting 1 would leave the uppermost bit set, so the & would | |
314 | * yield non-zero */ | |
315 | #if defined(PERL_CORE) || defined(PERL_EXT) | |
011b1419 | 316 | # define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0) |
84ff4fa9 KW |
317 | #endif |
318 | ||
d223e1ea | 319 | /* Returns a mask with the lowest n bits set */ |
fae1e72b | 320 | #define nBIT_MASK(n) ((UINTMAX_C(1) << (n)) - 1) |
d223e1ea | 321 | |
1381ccb1 KW |
322 | /* The largest unsigned number that will fit into n bits */ |
323 | #define nBIT_UMAX(n) nBIT_MASK(n) | |
324 | ||
8d9433eb | 325 | /* |
3f620621 | 326 | =for apidoc_section $directives |
d23778e6 | 327 | =for apidoc Am||__ASSERT_|bool expr |
8d9433eb KW |
328 | |
329 | This is a helper macro to avoid preprocessor issues, replaced by nothing | |
330 | unless under DEBUGGING, where it expands to an assert of its argument, | |
331 | followed by a comma (hence the comma operator). If we just used a straight | |
332 | assert(), we would get a comma with nothing before it when not DEBUGGING. | |
333 | ||
334 | =cut | |
335 | ||
803e4935 | 336 | We also use empty definition under Coverity since the __ASSERT_ |
8d9433eb KW |
337 | checks often check for things that Really Cannot Happen, and Coverity |
338 | detects that and gets all excited. */ | |
3e94db23 | 339 | |
e7ae132e KW |
340 | #if defined(DEBUGGING) && !defined(__COVERITY__) \ |
341 | && ! defined(PERL_SMALL_MACRO_BUFFER) | |
0f092d08 KW |
342 | # define __ASSERT_(statement) assert(statement), |
343 | #else | |
344 | # define __ASSERT_(statement) | |
345 | #endif | |
346 | ||
3fe05580 | 347 | /* |
3f620621 | 348 | =for apidoc_section $SV |
3fe05580 | 349 | |
3bb9fd01 | 350 | =for apidoc Ama|SV*|newSVpvs|"literal string" |
1568d13a | 351 | Like C<newSVpvn>, but takes a literal string instead of a |
30a15352 | 352 | string/length pair. |
3fe05580 | 353 | |
3bb9fd01 | 354 | =for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags |
1568d13a | 355 | Like C<newSVpvn_flags>, but takes a literal string instead of |
30a15352 | 356 | a string/length pair. |
84bafc02 | 357 | |
3bb9fd01 | 358 | =for apidoc Ama|SV*|newSVpvs_share|"literal string" |
1568d13a | 359 | Like C<newSVpvn_share>, but takes a literal string instead of |
30a15352 | 360 | a string/length pair and omits the hash parameter. |
3fe05580 | 361 | |
3bb9fd01 | 362 | =for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags |
1568d13a | 363 | Like C<sv_catpvn_flags>, but takes a literal string instead |
30a15352 | 364 | of a string/length pair. |
9dcc53ea | 365 | |
3bb9fd01 | 366 | =for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string" |
1568d13a | 367 | Like C<sv_catpvn_nomg>, but takes a literal string instead of |
0c395ea5 | 368 | a string/length pair. |
9dcc53ea | 369 | |
3bb9fd01 | 370 | =for apidoc Am|void|sv_catpvs|SV* sv|"literal string" |
1568d13a | 371 | Like C<sv_catpvn>, but takes a literal string instead of a |
0c395ea5 | 372 | string/length pair. |
3fe05580 | 373 | |
3bb9fd01 | 374 | =for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string" |
1568d13a | 375 | Like C<sv_catpvn_mg>, but takes a literal string instead of a |
9dcc53ea Z |
376 | string/length pair. |
377 | ||
3bb9fd01 | 378 | =for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string" |
1568d13a | 379 | Like C<sv_setref_pvn>, but takes a literal string instead of |
0c395ea5 | 380 | a string/length pair. |
9dcc53ea | 381 | |
3f620621 | 382 | =for apidoc_section $string |
3fe05580 | 383 | |
3bb9fd01 | 384 | =for apidoc Ama|char*|savepvs|"literal string" |
1568d13a | 385 | Like C<savepvn>, but takes a literal string instead of a |
30a15352 | 386 | string/length pair. |
3fe05580 | 387 | |
3bb9fd01 | 388 | =for apidoc Ama|char*|savesharedpvs|"literal string" |
9dcc53ea Z |
389 | A version of C<savepvs()> which allocates the duplicate string in memory |
390 | which is shared between threads. | |
391 | ||
3f620621 | 392 | =for apidoc_section $GV |
3fe05580 | 393 | |
3bb9fd01 | 394 | =for apidoc Am|HV*|gv_stashpvs|"name"|I32 create |
1568d13a | 395 | Like C<gv_stashpvn>, but takes a literal string instead of a |
0c395ea5 | 396 | string/length pair. |
3fe05580 | 397 | |
3f620621 | 398 | =for apidoc_section $HV |
3fe05580 | 399 | |
3bb9fd01 | 400 | =for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval |
1568d13a | 401 | Like C<hv_fetch>, but takes a literal string instead of a |
0c395ea5 | 402 | string/length pair. |
3f620621 | 403 | =for apidoc_section $lexer |
510966aa | 404 | |
3bb9fd01 | 405 | =for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags |
510966aa | 406 | |
1568d13a | 407 | Like L</lex_stuff_pvn>, but takes a literal string instead of |
0c395ea5 | 408 | a string/length pair. |
510966aa | 409 | |
3fe05580 MHM |
410 | =cut |
411 | */ | |
412 | ||
ca0572d7 KW |
413 | #define ASSERT_IS_LITERAL(s) ("" s "") |
414 | ||
a34e53fc | 415 | /* |
3f620621 | 416 | =for apidoc_section $string |
2efa8cc7 | 417 | |
a34e53fc KW |
418 | =for apidoc Amu|pair|STR_WITH_LEN|"literal string" |
419 | ||
420 | Returns two comma separated tokens of the input literal string, and its length. | |
421 | This is convenience macro which helps out in some API calls. | |
422 | Note that it can't be used as an argument to macros or functions that under | |
423 | some configurations might be macros, which means that it requires the full | |
424 | Perl_xxx(aTHX_ ...) form for any API calls where it's used. | |
425 | ||
426 | =cut | |
427 | */ | |
428 | ||
ca0572d7 | 429 | #define STR_WITH_LEN(s) ASSERT_IS_LITERAL(s), (sizeof(s)-1) |
ba3a79e7 GA |
430 | |
431 | /* STR_WITH_LEN() shortcuts */ | |
432 | #define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str)) | |
84bafc02 NC |
433 | #define newSVpvs_flags(str,flags) \ |
434 | Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags) | |
ba3a79e7 | 435 | #define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0) |
9dcc53ea Z |
436 | #define sv_catpvs_flags(sv, str, flags) \ |
437 | Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags) | |
438 | #define sv_catpvs_nomg(sv, str) \ | |
439 | Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0) | |
440 | #define sv_catpvs(sv, str) \ | |
441 | Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC) | |
442 | #define sv_catpvs_mg(sv, str) \ | |
443 | Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC) | |
3fe05580 | 444 | #define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str)) |
9dcc53ea Z |
445 | #define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str)) |
446 | #define sv_setref_pvs(rv, classname, str) \ | |
447 | Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str)) | |
ba3a79e7 | 448 | #define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str)) |
9dcc53ea Z |
449 | #define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str)) |
450 | #define gv_stashpvs(str, create) \ | |
451 | Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create) | |
3752a9fe KW |
452 | |
453 | #define gv_fetchpvs(namebeg, flags, sv_type) \ | |
454 | Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), flags, sv_type) | |
72659597 | 455 | #define gv_fetchpvn gv_fetchpvn_flags |
9dcc53ea Z |
456 | #define sv_catxmlpvs(dsv, str, utf8) \ |
457 | Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8) | |
4ac46235 | 458 | |
ba3a79e7 | 459 | |
510966aa Z |
460 | #define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags) |
461 | ||
b96d8cd9 | 462 | #define get_cvs(str, flags) \ |
1604cfb0 | 463 | Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags)) |
5c1737d1 | 464 | |
9b6e9510 | 465 | /* internal helpers */ |
4a1bbd3d | 466 | /* Transitional */ |
d67f622b N |
467 | #ifndef PERL_VERSION_MAJOR |
468 | # define PERL_VERSION_MAJOR PERL_REVISION | |
4a1bbd3d KW |
469 | #else |
470 | # undef PERL_REVISION /* We don't want code to be using these */ | |
471 | #endif | |
d67f622b N |
472 | #ifndef PERL_VERSION_MINOR |
473 | # define PERL_VERSION_MINOR PERL_VERSION | |
4a1bbd3d KW |
474 | #else |
475 | # undef PERL_VERSION | |
476 | #endif | |
d67f622b N |
477 | #ifndef PERL_VERSION_PATCH |
478 | # define PERL_VERSION_PATCH PERL_SUBVERSION | |
4a1bbd3d KW |
479 | #else |
480 | # undef PERL_SUBVERSION | |
481 | #endif | |
482 | ||
483 | #define PERL_JNP_TO_DECIMAL_(maJor,miNor,Patch) \ | |
484 | /* '10*' leaves room for things like alpha, beta, releases */ \ | |
485 | (10 * ((maJor) * 1000000) + ((miNor) * 1000) + (Patch)) | |
9b6e9510 | 486 | #define PERL_DECIMAL_VERSION_ \ |
d67f622b N |
487 | PERL_JNP_TO_DECIMAL_(PERL_VERSION_MAJOR, PERL_VERSION_MINOR, \ |
488 | PERL_VERSION_PATCH) | |
9b6e9510 KW |
489 | |
490 | /* | |
3f620621 | 491 | =for apidoc_section $versioning |
4a1bbd3d | 492 | =for apidoc AmR|bool|PERL_VERSION_EQ|const U8 major|const U8 minor|const U8 patch |
9de44d19 | 493 | =for apidoc_item PERL_VERSION_GE |
1607e393 KW |
494 | =for apidoc_item PERL_VERSION_GT |
495 | =for apidoc_item PERL_VERSION_LE | |
496 | =for apidoc_item PERL_VERSION_LT | |
497 | =for apidoc_item PERL_VERSION_NE | |
9b6e9510 | 498 | |
4a1bbd3d | 499 | Returns whether or not the perl currently being compiled has the specified |
9b6e9510 KW |
500 | relationship to the perl given by the parameters. For example, |
501 | ||
502 | #if PERL_VERSION_GT(5,24,2) | |
503 | code that will only be compiled on perls after v5.24.2 | |
504 | #else | |
505 | fallback code | |
506 | #endif | |
507 | ||
508 | Note that this is usable in making compile-time decisions | |
509 | ||
4a1bbd3d KW |
510 | You may use the special value '*' for the final number to mean ALL possible |
511 | values for it. Thus, | |
512 | ||
513 | #if PERL_VERSION_EQ(5,31,'*') | |
514 | ||
515 | means all perls in the 5.31 series. And | |
516 | ||
517 | #if PERL_VERSION_NE(5,24,'*') | |
518 | ||
519 | means all perls EXCEPT 5.24 ones. And | |
520 | ||
521 | #if PERL_VERSION_LE(5,9,'*') | |
522 | ||
523 | is effectively | |
524 | ||
525 | #if PERL_VERSION_LT(5,10,0) | |
526 | ||
527 | This means you don't have to think so much when converting from the existing | |
528 | deprecated C<PERL_VERSION> to using this macro: | |
529 | ||
530 | #if PERL_VERSION <= 9 | |
531 | ||
532 | becomes | |
533 | ||
534 | #if PERL_VERSION_LE(5,9,'*') | |
535 | ||
9b6e9510 KW |
536 | =cut |
537 | */ | |
538 | ||
4a1bbd3d KW |
539 | /* N.B. These don't work if the patch version is 42 or 92, as those are what |
540 | * '*' is in ASCII and EBCDIC respectively */ | |
541 | # define PERL_VERSION_EQ(j,n,p) \ | |
542 | (((p) == '*') \ | |
d67f622b N |
543 | ? ( (j) == PERL_VERSION_MAJOR \ |
544 | && (n) == PERL_VERSION_MINOR) \ | |
4a1bbd3d KW |
545 | : (PERL_DECIMAL_VERSION_ == PERL_JNP_TO_DECIMAL_(j,n,p))) |
546 | # define PERL_VERSION_NE(j,n,p) (! PERL_VERSION_EQ(j,n,p)) | |
547 | ||
548 | # define PERL_VERSION_LT(j,n,p) /* < '*' effectively means < 0 */ \ | |
549 | (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \ | |
550 | (n), \ | |
551 | (((p) == '*') ? 0 : p))) | |
552 | # define PERL_VERSION_GE(j,n,p) (! PERL_VERSION_LT(j,n,p)) | |
553 | ||
554 | # define PERL_VERSION_LE(j,n,p) /* <= '*' effectively means < n+1 */ \ | |
555 | (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \ | |
556 | (((p) == '*') ? ((n)+1) : (n)), \ | |
557 | (((p) == '*') ? 0 : p))) | |
558 | # define PERL_VERSION_GT(j,n,p) (! PERL_VERSION_LE(j,n,p)) | |
9b6e9510 | 559 | |
954c1994 | 560 | /* |
3f620621 | 561 | =for apidoc_section $string |
ccfc67b7 | 562 | |
954c1994 | 563 | =for apidoc Am|bool|strNE|char* s1|char* s2 |
dc6b0978 KW |
564 | Test two C<NUL>-terminated strings to see if they are different. Returns true |
565 | or false. | |
954c1994 GS |
566 | |
567 | =for apidoc Am|bool|strEQ|char* s1|char* s2 | |
dc6b0978 KW |
568 | Test two C<NUL>-terminated strings to see if they are equal. Returns true or |
569 | false. | |
954c1994 GS |
570 | |
571 | =for apidoc Am|bool|strLT|char* s1|char* s2 | |
dc6b0978 KW |
572 | Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the |
573 | second, C<s2>. Returns true or false. | |
954c1994 GS |
574 | |
575 | =for apidoc Am|bool|strLE|char* s1|char* s2 | |
dc6b0978 KW |
576 | Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or |
577 | equal to the second, C<s2>. Returns true or false. | |
954c1994 GS |
578 | |
579 | =for apidoc Am|bool|strGT|char* s1|char* s2 | |
dc6b0978 KW |
580 | Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than |
581 | the second, C<s2>. Returns true or false. | |
954c1994 GS |
582 | |
583 | =for apidoc Am|bool|strGE|char* s1|char* s2 | |
dc6b0978 KW |
584 | Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than |
585 | or equal to the second, C<s2>. Returns true or false. | |
954c1994 GS |
586 | |
587 | =for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len | |
dc6b0978 KW |
588 | Test two C<NUL>-terminated strings to see if they are different. The C<len> |
589 | parameter indicates the number of bytes to compare. Returns true or false. (A | |
954c1994 GS |
590 | wrapper for C<strncmp>). |
591 | ||
592 | =for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len | |
dc6b0978 KW |
593 | Test two C<NUL>-terminated strings to see if they are equal. The C<len> |
594 | parameter indicates the number of bytes to compare. Returns true or false. (A | |
595 | wrapper for C<strncmp>). | |
954c1994 | 596 | |
bd18bd40 KW |
597 | =for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len |
598 | Test two buffers (which may contain embedded C<NUL> characters, to see if they | |
599 | are equal. The C<len> parameter indicates the number of bytes to compare. | |
b96bd7bf KW |
600 | Returns true or false. It is undefined behavior if either of the buffers |
601 | doesn't contain at least C<len> bytes. | |
bd18bd40 | 602 | |
3bb9fd01 | 603 | =for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2" |
2d8eeddb KW |
604 | Like L</memEQ>, but the second string is a literal enclosed in double quotes, |
605 | C<l1> gives the number of bytes in C<s1>. | |
b96bd7bf | 606 | Returns true or false. |
2d8eeddb | 607 | |
bd18bd40 KW |
608 | =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len |
609 | Test two buffers (which may contain embedded C<NUL> characters, to see if they | |
610 | are not equal. The C<len> parameter indicates the number of bytes to compare. | |
b96bd7bf KW |
611 | Returns true or false. It is undefined behavior if either of the buffers |
612 | doesn't contain at least C<len> bytes. | |
bd18bd40 | 613 | |
3bb9fd01 | 614 | =for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2" |
2d8eeddb KW |
615 | Like L</memNE>, but the second string is a literal enclosed in double quotes, |
616 | C<l1> gives the number of bytes in C<s1>. | |
b96bd7bf | 617 | Returns true or false. |
2d8eeddb | 618 | |
4aada8b9 KW |
619 | =for apidoc Am|bool|memCHRs|"list"|char c |
620 | Returns the position of the first occurence of the byte C<c> in the literal | |
621 | string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are | |
622 | treated as unsigned char. Thus this macro can be used to determine if C<c> is | |
623 | in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c> | |
624 | is C<NUL> (and the set doesn't include C<NUL>). | |
625 | ||
954c1994 | 626 | =cut |
fc169e00 KW |
627 | |
628 | New macros should use the following conventions for their names (which are | |
629 | based on the underlying C library functions): | |
630 | ||
631 | (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s? | |
632 | ||
633 | Each has two main parameters, string-like operands that are compared | |
634 | against each other, as specified by the macro name. Some macros may | |
635 | additionally have one or potentially even two length parameters. If a length | |
636 | parameter applies to both string parameters, it will be positioned third; | |
637 | otherwise any length parameter immediately follows the string parameter it | |
638 | applies to. | |
639 | ||
640 | If the prefix to the name is 'str', the string parameter is a pointer to a C | |
641 | language string. Such a string does not contain embedded NUL bytes; its | |
642 | length may be unknown, but can be calculated by C<strlen()>, since it is | |
643 | terminated by a NUL, which isn't included in its length. | |
644 | ||
a3815e44 | 645 | The optional 'n' following 'str' means that there is a third parameter, |
fc169e00 KW |
646 | giving the maximum number of bytes to look at in each string. Even if both |
647 | strings are longer than the length parameter, those extra bytes will be | |
648 | unexamined. | |
649 | ||
650 | The 's' suffix means that the 2nd byte string parameter is a literal C | |
651 | double-quoted string. Its length will automatically be calculated by the | |
652 | macro, so no length parameter will ever be needed for it. | |
653 | ||
654 | If the prefix is 'mem', the string parameters don't have to be C strings; | |
655 | they may contain embedded NUL bytes, do not necessarily have a terminating | |
656 | NUL, and their lengths can be known only through other means, which in | |
657 | practice are additional parameter(s) passed to the function. All 'mem' | |
658 | functions have at least one length parameter. Barring any 'l' or 's' suffix, | |
659 | there is a single length parameter, in position 3, which applies to both | |
660 | string parameters. The 's' suffix means, as described above, that the 2nd | |
661 | string is a literal double-quoted C string (hence its length is calculated by | |
662 | the macro, and the length parameter to the function applies just to the first | |
663 | string parameter, and hence is positioned just after it). An 'l' suffix | |
664 | means that the 2nd string parameter has its own length parameter, and the | |
665 | signature will look like memFOOl(s1, l1, s2, l2). | |
666 | ||
667 | BEGIN (and END) are for testing if the 2nd string is an initial (or final) | |
668 | substring of the 1st string. 'P' if present indicates that the substring | |
669 | must be a "proper" one in tha mathematical sense that the first one must be | |
670 | strictly larger than the 2nd. | |
671 | ||
954c1994 GS |
672 | */ |
673 | ||
62946e08 | 674 | |
75400963 KW |
675 | #define strNE(s1,s2) (strcmp(s1,s2) != 0) |
676 | #define strEQ(s1,s2) (strcmp(s1,s2) == 0) | |
8d063cd8 LW |
677 | #define strLT(s1,s2) (strcmp(s1,s2) < 0) |
678 | #define strLE(s1,s2) (strcmp(s1,s2) <= 0) | |
679 | #define strGT(s1,s2) (strcmp(s1,s2) > 0) | |
680 | #define strGE(s1,s2) (strcmp(s1,s2) >= 0) | |
62946e08 | 681 | |
75400963 KW |
682 | #define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0) |
683 | #define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0) | |
378cc40b | 684 | |
9d3980bc KW |
685 | #define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0) |
686 | #define memNE(s1,s2,l) (! memEQ(s1,s2,l)) | |
36477c24 | 687 | |
085b7534 | 688 | /* memEQ and memNE where second comparand is a string constant */ |
568a785a | 689 | #define memEQs(s1, l, s2) \ |
ca0572d7 | 690 | (((sizeof(s2)-1) == (l)) && memEQ((s1), ASSERT_IS_LITERAL(s2), (sizeof(s2)-1))) |
5f50c6c9 | 691 | #define memNEs(s1, l, s2) (! memEQs(s1, l, s2)) |
568a785a | 692 | |
fdbb9a7c KW |
693 | /* Keep these private until we decide it was a good idea */ |
694 | #if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX) | |
695 | ||
ca0572d7 | 696 | #define strBEGINs(s1,s2) (strncmp(s1,ASSERT_IS_LITERAL(s2), sizeof(s2)-1) == 0) |
fdbb9a7c | 697 | |
bdb7e3f0 | 698 | #define memBEGINs(s1, l, s2) \ |
30a6480c | 699 | ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \ |
ca0572d7 | 700 | && memEQ(s1, ASSERT_IS_LITERAL(s2), sizeof(s2)-1)) |
de627158 | 701 | #define memBEGINPs(s1, l, s2) \ |
30a6480c | 702 | ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \ |
ca0572d7 | 703 | && memEQ(s1, ASSERT_IS_LITERAL(s2), sizeof(s2)-1)) |
bdb7e3f0 | 704 | #define memENDs(s1, l, s2) \ |
30a6480c | 705 | ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \ |
ca0572d7 | 706 | && memEQ(s1 + (l) - (sizeof(s2) - 1), ASSERT_IS_LITERAL(s2), sizeof(s2)-1)) |
b80f8424 | 707 | #define memENDPs(s1, l, s2) \ |
30a6480c | 708 | ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \ |
ca0572d7 | 709 | && memEQ(s1 + (l) - (sizeof(s2) - 1), ASSERT_IS_LITERAL(s2), sizeof(s2)-1)) |
fdbb9a7c | 710 | #endif /* End of making macros private */ |
bdb7e3f0 | 711 | |
062b6850 KW |
712 | #define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0) |
713 | #define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0) | |
714 | #define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0) | |
715 | #define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0) | |
716 | ||
ca0572d7 | 717 | #define memCHRs(s1,c) ((const char *) memchr(ASSERT_IS_LITERAL(s1) , c, sizeof(s1)-1)) |
4aada8b9 | 718 | |
bbce6d69 | 719 | /* |
720 | * Character classes. | |
721 | * | |
722 | * Unfortunately, the introduction of locales means that we | |
723 | * can't trust isupper(), etc. to tell the truth. And when | |
724 | * it comes to /\w+/ with tainting enabled, we *must* be able | |
725 | * to trust our character classes. | |
726 | * | |
81d43abf KW |
727 | * Therefore, the default tests in the text of Perl will be independent of |
728 | * locale. Any code that wants to depend on the current locale will use the | |
729 | * macros that contain _LC in their names | |
bbce6d69 | 730 | */ |
731 | ||
5bf5e40b | 732 | #ifdef USE_LOCALE_CTYPE |
2304df62 AD |
733 | # ifndef CTYPE256 |
734 | # define CTYPE256 | |
735 | # endif | |
736 | #endif | |
737 | ||
954c1994 | 738 | /* |
ccfc67b7 | 739 | |
dcccc8ff | 740 | =head1 Character classification |
243effed KW |
741 | This section is about functions (really macros) that classify characters |
742 | into types, such as punctuation versus alphabetic, etc. Most of these are | |
743 | analogous to regular expression character classes. (See | |
744 | L<perlrecharclass/POSIX Character Classes>.) There are several variants for | |
745 | each class. (Not all macros have all variants; each item below lists the | |
746 | ones valid for it.) None are affected by C<use bytes>, and only the ones | |
747 | with C<LC> in the name are affected by the current locale. | |
748 | ||
d713f9d9 KW |
749 | The base function, e.g., C<isALPHA()>, takes any signed or unsigned value, |
750 | treating it as a code point, and returns a boolean as to whether or not the | |
751 | character represented by it is (or on non-ASCII platforms, corresponds to) an | |
6aff1f14 KW |
752 | ASCII character in the named class based on platform, Unicode, and Perl rules. |
753 | If the input is a number that doesn't fit in an octet, FALSE is returned. | |
243effed | 754 | |
c98722a4 | 755 | Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function |
550da823 KW |
756 | with no suffix C<"_A">. This variant is used to emphasize by its name that |
757 | only ASCII-range characters can return TRUE. | |
4b9734bf | 758 | |
d60679e1 | 759 | Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set |
4b9734bf KW |
760 | onto the platform. That is, the code points that are ASCII are unaffected, |
761 | since ASCII is a subset of Latin-1. But the non-ASCII code points are treated | |
762 | as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return | |
763 | true when called with the code point 0xDF, which is a word character in both | |
4650c663 | 764 | ASCII and EBCDIC (though it represents different characters in each). |
d713f9d9 KW |
765 | If the input is a number that doesn't fit in an octet, FALSE is returned. |
766 | (Perl's documentation uses a colloquial definition of Latin-1, to include all | |
767 | code points below 256.) | |
243effed | 768 | |
d713f9d9 KW |
769 | Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for |
770 | inputs below 256, but if the code point is larger than 255, Unicode rules are | |
771 | used to determine if it is in the character class. For example, | |
d0da05db | 772 | C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A |
6aff1f14 | 773 | WITH MACRON in Unicode, and is a word character. |
243effed | 774 | |
059703b0 KW |
775 | Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>, |
776 | but are used for UTF-8 encoded strings. The two forms are different names for | |
777 | the same thing. Each call to one of these classifies the first character of | |
778 | the string starting at C<p>. The second parameter, C<e>, points to anywhere in | |
779 | the string beyond the first character, up to one byte past the end of the | |
780 | entire string. Although both variants are identical, the suffix C<_safe> in | |
781 | one name emphasizes that it will not attempt to read beyond S<C<e - 1>>, | |
782 | provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in | |
783 | C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in | |
784 | some way, the program may croak, or the function may return FALSE, at the | |
785 | discretion of the implementation, and subject to change in future releases. | |
243effed | 786 | |
d713f9d9 KW |
787 | Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants, |
788 | but the result is based on the current locale, which is what C<LC> in the name | |
789 | stands for. If Perl can determine that the current locale is a UTF-8 locale, | |
790 | it uses the published Unicode rules; otherwise, it uses the C library function | |
791 | that gives the named classification. For example, C<isDIGIT_LC()> when not in | |
792 | a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always | |
1a83413c KW |
793 | returned if the input won't fit into an octet. On some platforms where the C |
794 | library function is known to be defective, Perl changes its result to follow | |
795 | the POSIX standard's rules. | |
243effed | 796 | |
d713f9d9 KW |
797 | Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less |
798 | than 256, but for larger ones it returns the Unicode classification of the code | |
799 | point. | |
243effed | 800 | |
059703b0 KW |
801 | Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like |
802 | C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms | |
803 | are different names for the same thing. Each call to one of these classifies | |
804 | the first character of the string starting at C<p>. The second parameter, | |
805 | C<e>, points to anywhere in the string beyond the first character, up to one | |
806 | byte past the end of the entire string. Although both variants are identical, | |
807 | the suffix C<_safe> in one name emphasizes that it will not attempt to read | |
808 | beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this | |
809 | is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input | |
810 | character is malformed in some way, the program may croak, or the function may | |
811 | return FALSE, at the discretion of the implementation, and subject to change in | |
812 | future releases. | |
ccfc67b7 | 813 | |
6fdd32c3 KW |
814 | =for apidoc Am|bool|isALPHA|UV ch |
815 | =for apidoc_item ||isALPHA_A|UV ch | |
6fdd32c3 | 816 | =for apidoc_item ||isALPHA_LC|UV ch |
fa470d81 | 817 | =for apidoc_item ||isALPHA_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
818 | =for apidoc_item ||isALPHA_LC_uvchr|UV ch |
819 | =for apidoc_item ||isALPHA_L1|UV ch | |
820 | =for apidoc_item ||isALPHA_utf8|U8 * s|U8 * end | |
821 | =for apidoc_item ||isALPHA_utf8_safe|U8 * s|U8 * end | |
822 | =for apidoc_item ||isALPHA_uvchr|UV ch | |
d713f9d9 KW |
823 | Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>, |
824 | analogous to C<m/[[:alpha:]]/>. | |
dcccc8ff | 825 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 | 826 | the variants. |
8a58bdcf | 827 | |
f16858ed KW |
828 | =cut |
829 | ||
f1460a66 | 830 | Here and below, we add the prototypes of these macros for downstream programs |
f16858ed KW |
831 | that would be interested in them, such as Devel::PPPort |
832 | ||
6fdd32c3 KW |
833 | =for apidoc Am|bool|isALPHANUMERIC|UV ch |
834 | =for apidoc_item ||isALPHANUMERIC_A|UV ch | |
6fdd32c3 | 835 | =for apidoc_item ||isALPHANUMERIC_LC|UV ch |
fa470d81 | 836 | =for apidoc_item ||isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
837 | =for apidoc_item ||isALPHANUMERIC_LC_uvchr|UV ch |
838 | =for apidoc_item ||isALPHANUMERIC_L1|UV ch | |
839 | =for apidoc_item ||isALPHANUMERIC_utf8|U8 * s|U8 * end | |
840 | =for apidoc_item ||isALPHANUMERIC_utf8_safe|U8 * s|U8 * end | |
841 | =for apidoc_item ||isALPHANUMERIC_uvchr|UV ch | |
d713f9d9 KW |
842 | Returns a boolean indicating whether the specified character is one of |
843 | C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>. | |
dcccc8ff | 844 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 | 845 | the variants. |
15861f94 | 846 | |
c1ef4981 KW |
847 | =for apidoc Am|bool|isALNUMC|UV ch |
848 | =for apidoc_item ||isALNUMC_A|UV ch | |
849 | =for apidoc_item ||isALNUMC_LC|UV ch | |
850 | =for apidoc_item ||isALNUMC_LC_uvchr|UV ch | |
851 | =for apidoc_item ||isALNUMC_L1|UV ch | |
852 | These are discouraged, backward compatibility macros for L</C<isALPHANUMERIC>>. | |
853 | That is, each returns a boolean indicating whether the specified character is | |
854 | one of C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>. | |
855 | ||
856 | The C<C> suffix in the names was meant to indicate that they correspond to the | |
857 | C language L<C<isalnum(3)>>. | |
255b632a | 858 | |
6fdd32c3 KW |
859 | =for apidoc Am|bool|isASCII|UV ch |
860 | =for apidoc_item ||isASCII_A|UV ch | |
6fdd32c3 | 861 | =for apidoc_item ||isASCII_LC|UV ch |
fa470d81 | 862 | =for apidoc_item ||isASCII_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
863 | =for apidoc_item ||isASCII_LC_uvchr|UV ch |
864 | =for apidoc_item ||isASCII_L1|UV ch | |
865 | =for apidoc_item ||isASCII_utf8|U8 * s|U8 * end | |
866 | =for apidoc_item ||isASCII_utf8_safe|U8 * s|U8 * end | |
867 | =for apidoc_item ||isASCII_uvchr|UV ch | |
8a58bdcf | 868 | Returns a boolean indicating whether the specified character is one of the 128 |
243effed | 869 | characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>. |
e5ad6aba | 870 | On non-ASCII platforms, it returns TRUE iff this |
8a58bdcf KW |
871 | character corresponds to an ASCII character. Variants C<isASCII_A()> and |
872 | C<isASCII_L1()> are identical to C<isASCII()>. | |
dcccc8ff | 873 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 | 874 | the variants. |
059703b0 KW |
875 | Note, however, that some platforms do not have the C library routine |
876 | C<isascii()>. In these cases, the variants whose names contain C<LC> are the | |
877 | same as the corresponding ones without. | |
243effed | 878 | |
d98532ea KW |
879 | Also note, that because all ASCII characters are UTF-8 invariant (meaning they |
880 | have the exact same representation (always a single byte) whether encoded in | |
881 | UTF-8 or not), C<isASCII> will give the correct results when called with any | |
059703b0 KW |
882 | byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and |
883 | C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8. | |
d98532ea | 884 | |
6fdd32c3 KW |
885 | =for apidoc Am|bool|isBLANK|UV ch |
886 | =for apidoc_item ||isBLANK_A|UV ch | |
6fdd32c3 | 887 | =for apidoc_item ||isBLANK_LC|UV ch |
fa470d81 | 888 | =for apidoc_item ||isBLANK_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
889 | =for apidoc_item ||isBLANK_LC_uvchr|UV ch |
890 | =for apidoc_item ||isBLANK_L1|UV ch | |
891 | =for apidoc_item ||isBLANK_utf8|U8 * s|U8 * end | |
892 | =for apidoc_item ||isBLANK_utf8_safe|U8 * s|U8 * end | |
893 | =for apidoc_item ||isBLANK_uvchr|UV ch | |
243effed | 894 | Returns a boolean indicating whether the specified character is a |
6aff1f14 | 895 | character considered to be a blank, analogous to C<m/[[:blank:]]/>. |
dcccc8ff | 896 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 KW |
897 | the variants. |
898 | Note, | |
da8c1a98 KW |
899 | however, that some platforms do not have the C library routine |
900 | C<isblank()>. In these cases, the variants whose names contain C<LC> are | |
901 | the same as the corresponding ones without. | |
243effed | 902 | |
6fdd32c3 KW |
903 | =for apidoc Am|bool|isCNTRL|UV ch |
904 | =for apidoc_item ||isCNTRL_A|UV ch | |
6fdd32c3 | 905 | =for apidoc_item ||isCNTRL_LC|UV ch |
fa470d81 | 906 | =for apidoc_item ||isCNTRL_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
907 | =for apidoc_item ||isCNTRL_LC_uvchr|UV ch |
908 | =for apidoc_item ||isCNTRL_L1|UV ch | |
909 | =for apidoc_item ||isCNTRL_utf8|U8 * s|U8 * end | |
910 | =for apidoc_item ||isCNTRL_utf8_safe|U8 * s|U8 * end | |
911 | =for apidoc_item ||isCNTRL_uvchr|UV ch | |
f16858ed | 912 | |
243effed | 913 | Returns a boolean indicating whether the specified character is a |
6aff1f14 | 914 | control character, analogous to C<m/[[:cntrl:]]/>. |
dcccc8ff | 915 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 KW |
916 | the variants. |
917 | On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant. | |
918 | ||
6fdd32c3 KW |
919 | =for apidoc Am|bool|isDIGIT|UV ch |
920 | =for apidoc_item ||isDIGIT_A|UV ch | |
6fdd32c3 | 921 | =for apidoc_item ||isDIGIT_LC|UV ch |
fa470d81 | 922 | =for apidoc_item ||isDIGIT_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
923 | =for apidoc_item ||isDIGIT_LC_uvchr|UV ch |
924 | =for apidoc_item ||isDIGIT_L1|UV ch | |
925 | =for apidoc_item ||isDIGIT_utf8|U8 * s|U8 * end | |
926 | =for apidoc_item ||isDIGIT_utf8_safe|U8 * s|U8 * end | |
927 | =for apidoc_item ||isDIGIT_uvchr|UV ch | |
fa470d81 | 928 | |
2787a470 | 929 | Returns a boolean indicating whether the specified character is a |
6aff1f14 | 930 | digit, analogous to C<m/[[:digit:]]/>. |
8a58bdcf | 931 | Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>. |
dcccc8ff | 932 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 KW |
933 | the variants. |
934 | ||
6fdd32c3 KW |
935 | =for apidoc Am|bool|isGRAPH|UV ch |
936 | =for apidoc_item ||isGRAPH_A|UV ch | |
6fdd32c3 | 937 | =for apidoc_item ||isGRAPH_LC|UV ch |
fa470d81 | 938 | =for apidoc_item ||isGRAPH_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
939 | =for apidoc_item ||isGRAPH_LC_uvchr|UV ch |
940 | =for apidoc_item ||isGRAPH_L1|UV ch | |
941 | =for apidoc_item ||isGRAPH_utf8|U8 * s|U8 * end | |
942 | =for apidoc_item ||isGRAPH_utf8_safe|U8 * s|U8 * end | |
943 | =for apidoc_item ||isGRAPH_uvchr|UV ch | |
243effed | 944 | Returns a boolean indicating whether the specified character is a |
6aff1f14 | 945 | graphic character, analogous to C<m/[[:graph:]]/>. |
dcccc8ff | 946 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 KW |
947 | the variants. |
948 | ||
6fdd32c3 KW |
949 | =for apidoc Am|bool|isLOWER|UV ch |
950 | =for apidoc_item ||isLOWER_A|UV ch | |
6fdd32c3 | 951 | =for apidoc_item ||isLOWER_LC|UV ch |
fa470d81 | 952 | =for apidoc_item ||isLOWER_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
953 | =for apidoc_item ||isLOWER_LC_uvchr|UV ch |
954 | =for apidoc_item ||isLOWER_L1|UV ch | |
955 | =for apidoc_item ||isLOWER_utf8|U8 * s|U8 * end | |
956 | =for apidoc_item ||isLOWER_utf8_safe|U8 * s|U8 * end | |
957 | =for apidoc_item ||isLOWER_uvchr|UV ch | |
2787a470 | 958 | Returns a boolean indicating whether the specified character is a |
6aff1f14 | 959 | lowercase character, analogous to C<m/[[:lower:]]/>. |
dcccc8ff | 960 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 KW |
961 | the variants |
962 | ||
6fdd32c3 KW |
963 | =for apidoc Am|bool|isOCTAL|UV ch |
964 | =for apidoc_item ||isOCTAL_A|UV ch | |
965 | =for apidoc_item ||isOCTAL_L1|UV ch | |
2787a470 | 966 | Returns a boolean indicating whether the specified character is an |
6aff1f14 | 967 | octal digit, [0-7]. |
243effed KW |
968 | The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to |
969 | C<isOCTAL>. | |
970 | ||
6fdd32c3 KW |
971 | =for apidoc Am|bool|isPUNCT|UV ch |
972 | =for apidoc_item ||isPUNCT_A|UV ch | |
6fdd32c3 | 973 | =for apidoc_item ||isPUNCT_LC|UV ch |
fa470d81 | 974 | =for apidoc_item ||isPUNCT_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
975 | =for apidoc_item ||isPUNCT_LC_uvchr|UV ch |
976 | =for apidoc_item ||isPUNCT_L1|UV ch | |
977 | =for apidoc_item ||isPUNCT_utf8|U8 * s|U8 * end | |
978 | =for apidoc_item ||isPUNCT_utf8_safe|U8 * s|U8 * end | |
979 | =for apidoc_item ||isPUNCT_uvchr|UV ch | |
243effed | 980 | Returns a boolean indicating whether the specified character is a |
6aff1f14 KW |
981 | punctuation character, analogous to C<m/[[:punct:]]/>. |
982 | Note that the definition of what is punctuation isn't as | |
243effed KW |
983 | straightforward as one might desire. See L<perlrecharclass/POSIX Character |
984 | Classes> for details. | |
dcccc8ff | 985 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 KW |
986 | the variants. |
987 | ||
6fdd32c3 KW |
988 | =for apidoc Am|bool|isSPACE|UV ch |
989 | =for apidoc_item ||isSPACE_A|UV ch | |
6fdd32c3 | 990 | =for apidoc_item ||isSPACE_LC|UV ch |
fa470d81 | 991 | =for apidoc_item ||isSPACE_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
992 | =for apidoc_item ||isSPACE_LC_uvchr|UV ch |
993 | =for apidoc_item ||isSPACE_L1|UV ch | |
994 | =for apidoc_item ||isSPACE_utf8|U8 * s|U8 * end | |
995 | =for apidoc_item ||isSPACE_utf8_safe|U8 * s|U8 * end | |
996 | =for apidoc_item ||isSPACE_uvchr|UV ch | |
2787a470 | 997 | Returns a boolean indicating whether the specified character is a |
6aff1f14 | 998 | whitespace character. This is analogous |
398d098a | 999 | to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18 |
779cf272 | 1000 | this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the |
398d098a KW |
1001 | locale forms of this macro (the ones with C<LC> in their names) matched |
1002 | precisely what C<m/[[:space:]]/> does. In those releases, the only difference, | |
1003 | in the non-locale variants, was that C<isSPACE()> did not match a vertical tab. | |
1004 | (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.) | |
dcccc8ff | 1005 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 KW |
1006 | the variants. |
1007 | ||
6fdd32c3 KW |
1008 | =for apidoc Am|bool|isPSXSPC|UV ch |
1009 | =for apidoc_item ||isPSXSPC_A|UV ch | |
6fdd32c3 | 1010 | =for apidoc_item ||isPSXSPC_LC|UV ch |
fa470d81 | 1011 | =for apidoc_item ||isPSXSPC_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
1012 | =for apidoc_item ||isPSXSPC_LC_uvchr|UV ch |
1013 | =for apidoc_item ||isPSXSPC_L1|UV ch | |
1014 | =for apidoc_item ||isPSXSPC_utf8|U8 * s|U8 * end | |
1015 | =for apidoc_item ||isPSXSPC_utf8_safe|U8 * s|U8 * end | |
1016 | =for apidoc_item ||isPSXSPC_uvchr|UV ch | |
398d098a | 1017 | (short for Posix Space) |
779cf272 KW |
1018 | Starting in 5.18, this is identical in all its forms to the |
1019 | corresponding C<isSPACE()> macros. | |
398d098a KW |
1020 | The locale forms of this macro are identical to their corresponding |
1021 | C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the | |
1022 | non-locale forms differ from their C<isSPACE()> forms only in that the | |
1023 | C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do. | |
1024 | Otherwise they are identical. Thus this macro is analogous to what | |
1025 | C<m/[[:space:]]/> matches in a regular expression. | |
dcccc8ff | 1026 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 KW |
1027 | the variants. |
1028 | ||
6fdd32c3 KW |
1029 | =for apidoc Am|bool|isUPPER|UV ch |
1030 | =for apidoc_item ||isUPPER_A|UV ch | |
6fdd32c3 | 1031 | =for apidoc_item ||isUPPER_LC|UV ch |
fa470d81 | 1032 | =for apidoc_item ||isUPPER_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
1033 | =for apidoc_item ||isUPPER_LC_uvchr|UV ch |
1034 | =for apidoc_item ||isUPPER_L1|UV ch | |
1035 | =for apidoc_item ||isUPPER_utf8|U8 * s|U8 * end | |
1036 | =for apidoc_item ||isUPPER_utf8_safe|U8 * s|U8 * end | |
1037 | =for apidoc_item ||isUPPER_uvchr|UV ch | |
2787a470 | 1038 | Returns a boolean indicating whether the specified character is an |
6aff1f14 | 1039 | uppercase character, analogous to C<m/[[:upper:]]/>. |
dcccc8ff | 1040 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 KW |
1041 | the variants. |
1042 | ||
6fdd32c3 KW |
1043 | =for apidoc Am|bool|isPRINT|UV ch |
1044 | =for apidoc_item ||isPRINT_A|UV ch | |
6fdd32c3 | 1045 | =for apidoc_item ||isPRINT_LC|UV ch |
fa470d81 | 1046 | =for apidoc_item ||isPRINT_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
1047 | =for apidoc_item ||isPRINT_LC_uvchr|UV ch |
1048 | =for apidoc_item ||isPRINT_L1|UV ch | |
1049 | =for apidoc_item ||isPRINT_utf8|U8 * s|U8 * end | |
1050 | =for apidoc_item ||isPRINT_utf8_safe|U8 * s|U8 * end | |
1051 | =for apidoc_item ||isPRINT_uvchr|UV ch | |
8eea39dd | 1052 | Returns a boolean indicating whether the specified character is a |
6aff1f14 | 1053 | printable character, analogous to C<m/[[:print:]]/>. |
dcccc8ff | 1054 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 KW |
1055 | the variants. |
1056 | ||
6fdd32c3 KW |
1057 | =for apidoc Am|bool|isWORDCHAR|UV ch |
1058 | =for apidoc_item ||isWORDCHAR_A|UV ch | |
6fdd32c3 | 1059 | =for apidoc_item ||isWORDCHAR_LC|UV ch |
fa470d81 | 1060 | =for apidoc_item ||isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
1061 | =for apidoc_item ||isWORDCHAR_LC_uvchr|UV ch |
1062 | =for apidoc_item ||isWORDCHAR_L1|UV ch | |
1063 | =for apidoc_item ||isWORDCHAR_utf8|U8 * s|U8 * end | |
1064 | =for apidoc_item ||isWORDCHAR_utf8_safe|U8 * s|U8 * end | |
1065 | =for apidoc_item ||isWORDCHAR_uvchr|UV ch | |
243effed KW |
1066 | Returns a boolean indicating whether the specified character is a character |
1067 | that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match | |
1068 | in a regular expression. A word character is an alphabetic character, a | |
1069 | decimal digit, a connecting punctuation character (such as an underscore), or | |
1070 | a "mark" character that attaches to one of those (like some sort of accent). | |
c1ef4981 | 1071 | |
dcccc8ff | 1072 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 | 1073 | the variants. |
1607e393 | 1074 | |
fa470d81 KW |
1075 | C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, |
1076 | C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and | |
1077 | C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally | |
1078 | include the platform's native underscore. | |
1079 | ||
c1ef4981 KW |
1080 | =for apidoc Am|bool|isALNUM |UV ch |
1081 | =for apidoc_item ||isALNUM_A |UV ch | |
1082 | =for apidoc_item ||isALNUM_LC |UV ch | |
1083 | =for apidoc_item ||isALNUM_LC_uvchr|UV ch | |
1084 | These are each a synonym for their respectively named L</C<isWORDCHAR>> | |
1085 | variant. | |
1086 | ||
1087 | They are provided for backward compatibility, even though a word character | |
1088 | includes more than the standard C language meaning of alphanumeric. | |
1089 | To get the C language definition, use the corresponding L</C<isALPHANUMERIC>> | |
1090 | variant. | |
1091 | ||
6fdd32c3 KW |
1092 | =for apidoc Am|bool|isXDIGIT|UV ch |
1093 | =for apidoc_item ||isXDIGIT_A|UV ch | |
6fdd32c3 | 1094 | =for apidoc_item ||isXDIGIT_LC|UV ch |
fa470d81 | 1095 | =for apidoc_item ||isXDIGIT_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
1096 | =for apidoc_item ||isXDIGIT_LC_uvchr|UV ch |
1097 | =for apidoc_item ||isXDIGIT_L1|UV ch | |
1098 | =for apidoc_item ||isXDIGIT_utf8|U8 * s|U8 * end | |
1099 | =for apidoc_item ||isXDIGIT_utf8_safe|U8 * s|U8 * end | |
1100 | =for apidoc_item ||isXDIGIT_uvchr|UV ch | |
8a58bdcf | 1101 | Returns a boolean indicating whether the specified character is a hexadecimal |
243effed KW |
1102 | digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()> |
1103 | and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>. | |
dcccc8ff | 1104 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 KW |
1105 | the variants. |
1106 | ||
6fdd32c3 KW |
1107 | =for apidoc Am|bool|isIDFIRST|UV ch |
1108 | =for apidoc_item ||isIDFIRST_A|UV ch | |
6fdd32c3 | 1109 | =for apidoc_item ||isIDFIRST_LC|UV ch |
fa470d81 | 1110 | =for apidoc_item ||isIDFIRST_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
1111 | =for apidoc_item ||isIDFIRST_LC_uvchr|UV ch |
1112 | =for apidoc_item ||isIDFIRST_L1|UV ch | |
1113 | =for apidoc_item ||isIDFIRST_utf8|U8 * s|U8 * end | |
1114 | =for apidoc_item ||isIDFIRST_utf8_safe|U8 * s|U8 * end | |
1115 | =for apidoc_item ||isIDFIRST_uvchr|UV ch | |
3c3ecf18 KW |
1116 | Returns a boolean indicating whether the specified character can be the first |
1117 | character of an identifier. This is very close to, but not quite the same as | |
1118 | the official Unicode property C<XID_Start>. The difference is that this | |
1119 | returns true only if the input character also matches L</isWORDCHAR>. | |
dcccc8ff | 1120 | See the L<top of this section|/Character classification> for an explanation of |
fa470d81 KW |
1121 | the variants. |
1122 | ||
6fdd32c3 KW |
1123 | =for apidoc Am|bool|isIDCONT|UV ch |
1124 | =for apidoc_item ||isIDCONT_A|UV ch | |
6fdd32c3 | 1125 | =for apidoc_item ||isIDCONT_LC|UV ch |
fa470d81 | 1126 | =for apidoc_item ||isIDCONT_LC_utf8_safe|U8 * s| U8 *end |
1607e393 KW |
1127 | =for apidoc_item ||isIDCONT_LC_uvchr|UV ch |
1128 | =for apidoc_item ||isIDCONT_L1|UV ch | |
1129 | =for apidoc_item ||isIDCONT_utf8|U8 * s|U8 * end | |
1130 | =for apidoc_item ||isIDCONT_utf8_safe|U8 * s|U8 * end | |
1131 | =for apidoc_item ||isIDCONT_uvchr|UV ch | |
3c3ecf18 KW |
1132 | Returns a boolean indicating whether the specified character can be the |
1133 | second or succeeding character of an identifier. This is very close to, but | |
1134 | not quite the same as the official Unicode property C<XID_Continue>. The | |
1135 | difference is that this returns true only if the input character also matches | |
dcccc8ff | 1136 | L</isWORDCHAR>. See the L<top of this section|/Character classification> for |
fa470d81 | 1137 | an explanation of the variants. |
f16858ed | 1138 | |
3f620621 | 1139 | =for apidoc_section $numeric |
8eea39dd | 1140 | |
95a59cab | 1141 | =for apidoc Am|U8|READ_XDIGIT|char str* |
243effed | 1142 | Returns the value of an ASCII-range hex digit and advances the string pointer. |
95a59cab YO |
1143 | Behaviour is only well defined when isXDIGIT(*str) is true. |
1144 | ||
e7c1e6c1 | 1145 | =head1 Character case changing |
21da7284 KW |
1146 | Perl uses "full" Unicode case mappings. This means that converting a single |
1147 | character to another case may result in a sequence of more than one character. | |
1148 | For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two | |
1149 | character sequence C<SS>. This presents some complications The lowercase of | |
1150 | all characters in the range 0..255 is a single character, and thus | |
1151 | C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't | |
1152 | return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has | |
1153 | an API that does allow every possible legal result to be returned.) Likewise | |
1154 | no other function that is crippled by not being able to give the correct | |
1155 | results for the full range of possible inputs has been implemented here. | |
e7c1e6c1 | 1156 | |
45d6cb5e KW |
1157 | =for apidoc Am|UV|toUPPER|UV cp |
1158 | =for apidoc_item |UV|toUPPER_A|UV cp | |
45d6cb5e KW |
1159 | =for apidoc_item |UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp |
1160 | =for apidoc_item |UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp | |
1607e393 | 1161 | =for apidoc_item |UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp |
45d6cb5e KW |
1162 | |
1163 | These all return the uppercase of a character. The differences are what domain | |
1164 | they operate on, and whether the input is specified as a code point (those | |
1165 | forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter | |
1166 | case, the code point to use is the first one in the buffer of UTF-8 encoded | |
1167 | code points, delineated by the arguments S<C<p .. e - 1>>. | |
1168 | ||
1169 | C<toUPPER> and C<toUPPER_A> are synonyms of each other. They return the | |
1170 | uppercase of any lowercase ASCII-range code point. All other inputs are | |
1171 | returned unchanged. Since these are macros, the input type may be any integral | |
1172 | one, and the output will occupy the same number of bits as the input. | |
1173 | ||
1174 | There is no C<toUPPER_L1> nor C<toUPPER_LATIN1> as the uppercase of some code | |
1175 | points in the 0..255 range is above that range or consists of multiple | |
1176 | characters. Instead use C<toUPPER_uvchr>. | |
1177 | ||
1178 | C<toUPPER_uvchr> returns the uppercase of any Unicode code point. The return | |
1179 | value is identical to that of C<toUPPER_A> for input code points in the ASCII | |
1180 | range. The uppercase of the vast majority of Unicode code points is the same | |
1181 | as the code point itself. For these, and for code points above the legal | |
1182 | Unicode maximum, this returns the input code point unchanged. It additionally | |
1183 | stores the UTF-8 of the result into the buffer beginning at C<s>, and its | |
1184 | length in bytes into C<*lenp>. The caller must have made C<s> large enough to | |
1185 | contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow. | |
1186 | ||
1187 | NOTE: the uppercase of a code point may be more than one code point. The | |
1188 | return value of this function is only the first of these. The entire uppercase | |
1189 | is returned in C<s>. To determine if the result is more than a single code | |
1190 | point, you can do something like this: | |
1191 | ||
1192 | uc = toUPPER_uvchr(cp, s, &len); | |
1193 | if (len > UTF8SKIP(s)) { is multiple code points } | |
1194 | else { is a single code point } | |
1195 | ||
1196 | C<toUPPER_utf8> and C<toUPPER_utf8_safe> are synonyms of each other. The only | |
1197 | difference between these and C<toUPPER_uvchr> is that the source for these is | |
1198 | encoded in UTF-8, instead of being a code point. It is passed as a buffer | |
1199 | starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p> | |
1200 | buffer may certainly contain more than one code point; but only the first one | |
1201 | (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is | |
1202 | malformed in some way, the program may croak, or the function may return the | |
1203 | REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to | |
1204 | change in future releases. | |
1205 | ||
1206 | =for apidoc Am|UV|toFOLD|UV cp | |
1207 | =for apidoc_item |UV|toFOLD_A|UV cp | |
45d6cb5e KW |
1208 | =for apidoc_item |UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp |
1209 | =for apidoc_item |UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp | |
1607e393 | 1210 | =for apidoc_item |UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp |
45d6cb5e KW |
1211 | |
1212 | These all return the foldcase of a character. "foldcase" is an internal case | |
1213 | for C</i> pattern matching. If the foldcase of character A and the foldcase of | |
1214 | character B are the same, they match caselessly; otherwise they don't. | |
1215 | ||
1216 | The differences in the forms are what domain they operate on, and whether the | |
1217 | input is specified as a code point (those forms with a C<cp> parameter) or as a | |
1218 | UTF-8 string (the others). In the latter case, the code point to use is the | |
1219 | first one in the buffer of UTF-8 encoded code points, delineated by the | |
1220 | arguments S<C<p .. e - 1>>. | |
1221 | ||
1222 | C<toFOLD> and C<toFOLD_A> are synonyms of each other. They return the | |
1223 | foldcase of any ASCII-range code point. In this range, the foldcase is | |
1224 | identical to the lowercase. All other inputs are returned unchanged. Since | |
1225 | these are macros, the input type may be any integral one, and the output will | |
1226 | occupy the same number of bits as the input. | |
1227 | ||
1228 | There is no C<toFOLD_L1> nor C<toFOLD_LATIN1> as the foldcase of some code | |
1229 | points in the 0..255 range is above that range or consists of multiple | |
1230 | characters. Instead use C<toFOLD_uvchr>. | |
1231 | ||
1232 | C<toFOLD_uvchr> returns the foldcase of any Unicode code point. The return | |
1233 | value is identical to that of C<toFOLD_A> for input code points in the ASCII | |
1234 | range. The foldcase of the vast majority of Unicode code points is the same | |
1235 | as the code point itself. For these, and for code points above the legal | |
1236 | Unicode maximum, this returns the input code point unchanged. It additionally | |
1237 | stores the UTF-8 of the result into the buffer beginning at C<s>, and its | |
1238 | length in bytes into C<*lenp>. The caller must have made C<s> large enough to | |
1239 | contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow. | |
1240 | ||
1241 | NOTE: the foldcase of a code point may be more than one code point. The | |
1242 | return value of this function is only the first of these. The entire foldcase | |
1243 | is returned in C<s>. To determine if the result is more than a single code | |
1244 | point, you can do something like this: | |
1245 | ||
1246 | uc = toFOLD_uvchr(cp, s, &len); | |
1247 | if (len > UTF8SKIP(s)) { is multiple code points } | |
1248 | else { is a single code point } | |
1249 | ||
1250 | C<toFOLD_utf8> and C<toFOLD_utf8_safe> are synonyms of each other. The only | |
1251 | difference between these and C<toFOLD_uvchr> is that the source for these is | |
1252 | encoded in UTF-8, instead of being a code point. It is passed as a buffer | |
1253 | starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p> | |
1254 | buffer may certainly contain more than one code point; but only the first one | |
1255 | (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is | |
1256 | malformed in some way, the program may croak, or the function may return the | |
1257 | REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to | |
1258 | change in future releases. | |
1f607577 | 1259 | |
3cb048e5 KW |
1260 | =for apidoc Am|UV|toLOWER|UV cp |
1261 | =for apidoc_item |UV|toLOWER_A|UV cp | |
3cb048e5 KW |
1262 | =for apidoc_item |UV|toLOWER_LATIN1|UV cp |
1263 | =for apidoc_item |UV|toLOWER_LC|UV cp | |
1607e393 | 1264 | =for apidoc_item |UV|toLOWER_L1|UV cp |
3cb048e5 KW |
1265 | =for apidoc_item |UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp |
1266 | =for apidoc_item |UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp | |
1607e393 | 1267 | =for apidoc_item |UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp |
3cb048e5 KW |
1268 | |
1269 | These all return the lowercase of a character. The differences are what domain | |
1270 | they operate on, and whether the input is specified as a code point (those | |
1271 | forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter | |
1272 | case, the code point to use is the first one in the buffer of UTF-8 encoded | |
1273 | code points, delineated by the arguments S<C<p .. e - 1>>. | |
1274 | ||
1275 | C<toLOWER> and C<toLOWER_A> are synonyms of each other. They return the | |
1276 | lowercase of any uppercase ASCII-range code point. All other inputs are | |
1277 | returned unchanged. Since these are macros, the input type may be any integral | |
1278 | one, and the output will occupy the same number of bits as the input. | |
1279 | ||
1280 | C<toLOWER_L1> and C<toLOWER_LATIN1> are synonyms of each other. They behave | |
1281 | identically as C<toLOWER> for ASCII-range input. But additionally will return | |
1282 | the lowercase of any uppercase code point in the entire 0..255 range, assuming | |
1283 | a Latin-1 encoding (or the EBCDIC equivalent on such platforms). | |
1284 | ||
1285 | C<toLOWER_LC> returns the lowercase of the input code point according to the | |
1286 | rules of the current POSIX locale. Input code points outside the range 0..255 | |
1287 | are returned unchanged. | |
1288 | ||
1289 | C<toLOWER_uvchr> returns the lowercase of any Unicode code point. The return | |
1290 | value is identical to that of C<toLOWER_L1> for input code points in the 0..255 | |
1291 | range. The lowercase of the vast majority of Unicode code points is the same | |
1292 | as the code point itself. For these, and for code points above the legal | |
1293 | Unicode maximum, this returns the input code point unchanged. It additionally | |
1294 | stores the UTF-8 of the result into the buffer beginning at C<s>, and its | |
1295 | length in bytes into C<*lenp>. The caller must have made C<s> large enough to | |
1296 | contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow. | |
1297 | ||
1298 | NOTE: the lowercase of a code point may be more than one code point. The | |
1299 | return value of this function is only the first of these. The entire lowercase | |
1300 | is returned in C<s>. To determine if the result is more than a single code | |
1301 | point, you can do something like this: | |
1302 | ||
1303 | uc = toLOWER_uvchr(cp, s, &len); | |
1304 | if (len > UTF8SKIP(s)) { is multiple code points } | |
1305 | else { is a single code point } | |
1306 | ||
1307 | C<toLOWER_utf8> and C<toLOWER_utf8_safe> are synonyms of each other. The only | |
1308 | difference between these and C<toLOWER_uvchr> is that the source for these is | |
1309 | encoded in UTF-8, instead of being a code point. It is passed as a buffer | |
1310 | starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p> | |
1311 | buffer may certainly contain more than one code point; but only the first one | |
1312 | (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is | |
1313 | malformed in some way, the program may croak, or the function may return the | |
1314 | REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to | |
1315 | change in future releases. | |
1f607577 | 1316 | |
45d6cb5e KW |
1317 | =for apidoc Am|UV|toTITLE|UV cp |
1318 | =for apidoc_item |UV|toTITLE_A|UV cp | |
45d6cb5e KW |
1319 | =for apidoc_item |UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp |
1320 | =for apidoc_item |UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp | |
1607e393 | 1321 | =for apidoc_item |UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp |
45d6cb5e KW |
1322 | |
1323 | These all return the titlecase of a character. The differences are what domain | |
1324 | they operate on, and whether the input is specified as a code point (those | |
1325 | forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter | |
1326 | case, the code point to use is the first one in the buffer of UTF-8 encoded | |
1327 | code points, delineated by the arguments S<C<p .. e - 1>>. | |
1328 | ||
1329 | C<toTITLE> and C<toTITLE_A> are synonyms of each other. They return the | |
1330 | titlecase of any lowercase ASCII-range code point. In this range, the | |
1331 | titlecase is identical to the uppercase. All other inputs are returned | |
1332 | unchanged. Since these are macros, the input type may be any integral one, and | |
1333 | the output will occupy the same number of bits as the input. | |
1334 | ||
1335 | There is no C<toTITLE_L1> nor C<toTITLE_LATIN1> as the titlecase of some code | |
1336 | points in the 0..255 range is above that range or consists of multiple | |
1337 | characters. Instead use C<toTITLE_uvchr>. | |
1338 | ||
1339 | C<toTITLE_uvchr> returns the titlecase of any Unicode code point. The return | |
1340 | value is identical to that of C<toTITLE_A> for input code points in the ASCII | |
1341 | range. The titlecase of the vast majority of Unicode code points is the same | |
1342 | as the code point itself. For these, and for code points above the legal | |
1343 | Unicode maximum, this returns the input code point unchanged. It additionally | |
1344 | stores the UTF-8 of the result into the buffer beginning at C<s>, and its | |
1345 | length in bytes into C<*lenp>. The caller must have made C<s> large enough to | |
1346 | contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow. | |
1347 | ||
1348 | NOTE: the titlecase of a code point may be more than one code point. The | |
1349 | return value of this function is only the first of these. The entire titlecase | |
1350 | is returned in C<s>. To determine if the result is more than a single code | |
1351 | point, you can do something like this: | |
1352 | ||
1353 | uc = toTITLE_uvchr(cp, s, &len); | |
1354 | if (len > UTF8SKIP(s)) { is multiple code points } | |
1355 | else { is a single code point } | |
1356 | ||
1357 | C<toTITLE_utf8> and C<toTITLE_utf8_safe> are synonyms of each other. The only | |
1358 | difference between these and C<toTITLE_uvchr> is that the source for these is | |
1359 | encoded in UTF-8, instead of being a code point. It is passed as a buffer | |
1360 | starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p> | |
1361 | buffer may certainly contain more than one code point; but only the first one | |
1362 | (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is | |
1363 | malformed in some way, the program may croak, or the function may return the | |
1364 | REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to | |
1365 | change in future releases. | |
1f607577 | 1366 | |
954c1994 | 1367 | =cut |
353c9b6f | 1368 | |
d0da05db | 1369 | XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names |
1e222e4f KW |
1370 | really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change, |
1371 | and aren't general purpose as they don't work on U+DF, and assert against that. | |
8fd8ea43 | 1372 | and isCASED_LC, as it really is more of an internal thing. |
243effed | 1373 | |
8a58bdcf | 1374 | Note that these macros are repeated in Devel::PPPort, so should also be |
62fa66b6 KW |
1375 | patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc |
1376 | ||
954c1994 GS |
1377 | */ |
1378 | ||
8f5283f4 KW |
1379 | /* |
1380 | void below because that's the best fit, and works for Devel::PPPort | |
3f620621 | 1381 | =for apidoc_section $integer |
545bca17 | 1382 | =for apidoc AyT||WIDEST_UTYPE |
8f5283f4 KW |
1383 | |
1384 | Yields the widest unsigned integer type on the platform, currently either | |
326c768d | 1385 | C<U32> or C<U64>. This can be used in declarations such as |
8f5283f4 KW |
1386 | |
1387 | WIDEST_UTYPE my_uv; | |
1388 | ||
1389 | or casts | |
1390 | ||
1391 | my_uv = (WIDEST_UTYPE) val; | |
1392 | ||
1393 | =cut | |
1394 | ||
1395 | */ | |
bbe73871 | 1396 | #define WIDEST_UTYPE PERL_UINTMAX_T |
7c062697 | 1397 | |
de40ad3f KW |
1398 | /* Where there could be some confusion, use this as a static assert in macros |
1399 | * to make sure that a parameter isn't a pointer. But some compilers can't | |
1400 | * handle this. The only one known so far that doesn't is gcc 3.3.6; the check | |
1401 | * below isn't thorough for such an old compiler, so may have to be revised if | |
1402 | * experience so dictates. */ | |
1403 | #if ! PERL_IS_GCC || PERL_GCC_VERSION_GT(3,3,6) | |
1404 | # define ASSERT_NOT_PTR(x) ((x) | 0) | |
1405 | #else | |
1406 | # define ASSERT_NOT_PTR(x) (x) | |
1407 | #endif | |
296969d3 | 1408 | |
08e4e7bf | 1409 | /* Likewise, this is effectively a static assert to be used to guarantee the |
43131e10 KW |
1410 | * parameter is a pointer |
1411 | * | |
1412 | * NOT suitable for void* | |
1413 | */ | |
08e4e7bf KW |
1414 | #define ASSERT_IS_PTR(x) (__ASSERT_(sizeof(*(x))) (x)) |
1415 | ||
3912bc88 KW |
1416 | /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in |
1417 | * the lower 8. It is designed to be hopefully bomb-proof, making sure that no | |
1418 | * bits of information are lost even on a 64-bit machine, but to get the | |
1419 | * compiler to optimize it out if possible. This is because Configure makes | |
1420 | * sure that the machine has an 8-bit byte, so if c is stored in a byte, the | |
1421 | * sizeof() guarantees that this evaluates to a constant true at compile time. | |
7e75d1a1 JH |
1422 | * |
1423 | * For Coverity, be always true, because otherwise Coverity thinks | |
1424 | * it finds several expressions that are always true, independent | |
1425 | * of operands. Well, they are, but that is kind of the point. | |
220c71bf | 1426 | */ |
7e75d1a1 | 1427 | #ifndef __COVERITY__ |
231a6d16 KW |
1428 | /* The '| 0' part in ASSERT_NOT_PTR ensures a compiler error if c is not |
1429 | * integer (like e.g., a pointer) */ | |
1430 | # define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \ | |
1431 | || (((WIDEST_UTYPE) ASSERT_NOT_PTR(c)) >> 8) == 0) | |
7e75d1a1 | 1432 | #else |
9555181b | 1433 | # define FITS_IN_8_BITS(c) (1) |
7e75d1a1 | 1434 | #endif |
cf301eb7 | 1435 | |
45f4bb73 | 1436 | /* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative |
833b0f46 | 1437 | * Written this way so that after optimization, only one conditional test is |
76d3ad4c KW |
1438 | * needed. (The NV casts stop any warnings about comparison always being true |
1439 | * if called with an unsigned. The cast preserves the sign, which is all we | |
1440 | * care about.) */ | |
92a0bb24 KW |
1441 | #define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \ |
1442 | __ASSERT_((NV) (n) >= 0) \ | |
1443 | withinCOUNT_KNOWN_VALID_((c), (l), (n))) | |
1444 | ||
1445 | /* For internal use only, this can be used in places where it is known that the | |
1446 | * parameters to withinCOUNT() are valid, to avoid the asserts. For example, | |
1447 | * inRANGE() below, calls this several times, but does all the necessary | |
1448 | * asserts itself, once. The reason that this is necessary is that the | |
1449 | * duplicate asserts were exceeding the internal limits of some compilers */ | |
1450 | #define withinCOUNT_KNOWN_VALID_(c, l, n) \ | |
296969d3 KW |
1451 | ((((WIDEST_UTYPE) (c)) - ASSERT_NOT_PTR(l)) \ |
1452 | <= ((WIDEST_UTYPE) ASSERT_NOT_PTR(n))) | |
833b0f46 | 1453 | |
94250c4f KW |
1454 | /* Returns true if c is in the range l..u, where 'l' is non-negative |
1455 | * Written this way so that after optimization, only one conditional test is | |
4758c20d | 1456 | * needed. */ |
92a0bb24 KW |
1457 | #define inRANGE(c, l, u) (__ASSERT_((NV) (l) >= 0) __ASSERT_((u) >= (l)) \ |
1458 | ( (sizeof(c) == sizeof(U8)) ? inRANGE_helper_(U8, (c), (l), ((u))) \ | |
1442da54 | 1459 | : (sizeof(c) == sizeof(U16)) ? inRANGE_helper_(U16,(c), (l), ((u))) \ |
92a0bb24 KW |
1460 | : (sizeof(c) == sizeof(U32)) ? inRANGE_helper_(U32,(c), (l), ((u))) \ |
1461 | : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \ | |
1462 | inRANGE_helper_(WIDEST_UTYPE,(c), (l), ((u)))))) | |
1463 | ||
1464 | /* For internal use, this is used by machine-generated code which generates | |
1465 | * known valid calls, with a known sizeof(). This avoids the extra code and | |
1466 | * asserts that were exceeding internal limits of some compilers. */ | |
1467 | #define inRANGE_helper_(cast, c, l, u) \ | |
1468 | withinCOUNT_KNOWN_VALID_(((cast) (c)), (l), ((u) - (l))) | |
305fe86e | 1469 | |
41f43cc2 | 1470 | #ifdef EBCDIC |
b6340bd0 | 1471 | # ifndef _ALL_SOURCE |
0852beac KW |
1472 | /* The native libc isascii() et.al. functions return the wrong results |
1473 | * on at least z/OS unless this is defined. */ | |
b6340bd0 KW |
1474 | # error _ALL_SOURCE should probably be defined |
1475 | # endif | |
41f43cc2 | 1476 | #else |
0852beac KW |
1477 | /* There is a simple definition of ASCII for ASCII platforms. But the |
1478 | * EBCDIC one isn't so simple, so is defined using table look-up like the | |
9c903d59 | 1479 | * other macros below. |
3f3c579d KW |
1480 | * |
1481 | * The cast here is used instead of '(c) >= 0', because some compilers emit | |
1482 | * a warning that that test is always true when the parameter is an | |
1483 | * unsigned type. khw supposes that it could be written as | |
1484 | * && ((c) == '\0' || (c) > 0) | |
1485 | * to avoid the message, but the cast will likely avoid extra branches even | |
296969d3 KW |
1486 | * with stupid compilers. */ |
1487 | # define isASCII(c) (((WIDEST_UTYPE) ASSERT_NOT_PTR(c)) < 128) | |
41f43cc2 KW |
1488 | #endif |
1489 | ||
38694112 KW |
1490 | /* Take the eight possible bit patterns of the lower 3 bits and you get the |
1491 | * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits | |
1492 | * can be ignored. If the rest match '0', we have an octal */ | |
296969d3 | 1493 | #define isOCTAL_A(c) ((((WIDEST_UTYPE) ASSERT_NOT_PTR(c)) & ~7) == '0') |
c2da0b36 | 1494 | |
9fb1bf9d | 1495 | #ifdef H_PERL /* If have access to perl.h, lookup in its table */ |
f4cdb42c | 1496 | |
a500dc72 KW |
1497 | /* Character class numbers. For internal core Perl use only. The ones less |
1498 | * than 32 are used in PL_charclass[] and the ones up through the one that | |
91456fff | 1499 | * corresponds to <HIGHEST_REGCOMP_DOT_H_SYNC_> are used by regcomp.h and |
a500dc72 KW |
1500 | * related files. PL_charclass ones use names used in l1_char_class_tab.h but |
1501 | * their actual definitions are here. If that file has a name not used here, | |
1502 | * it won't compile. | |
1709d539 KW |
1503 | * |
1504 | * The first group of these is ordered in what I (khw) estimate to be the | |
31c7f561 | 1505 | * frequency of their use. This gives a slight edge to exiting a loop earlier |
58a3ba2c KW |
1506 | * (in reginclass() in regexec.c). Except \v should be last, as it isn't a |
1507 | * real Posix character class, and some (small) inefficiencies in regular | |
1508 | * expression handling would be introduced by putting it in the middle of those | |
1509 | * that are. Also, cntrl and ascii come after the others as it may be useful | |
1510 | * to group these which have no members that match above Latin1, (or above | |
1511 | * ASCII in the latter case) */ | |
1512 | ||
91456fff KW |
1513 | # define CC_WORDCHAR_ 0 /* \w and [:word:] */ |
1514 | # define CC_DIGIT_ 1 /* \d and [:digit:] */ | |
1515 | # define CC_ALPHA_ 2 /* [:alpha:] */ | |
1516 | # define CC_LOWER_ 3 /* [:lower:] */ | |
1517 | # define CC_UPPER_ 4 /* [:upper:] */ | |
1518 | # define CC_PUNCT_ 5 /* [:punct:] */ | |
1519 | # define CC_PRINT_ 6 /* [:print:] */ | |
1520 | # define CC_ALPHANUMERIC_ 7 /* [:alnum:] */ | |
1521 | # define CC_GRAPH_ 8 /* [:graph:] */ | |
1522 | # define CC_CASED_ 9 /* [:lower:] or [:upper:] under /i */ | |
1523 | # define CC_SPACE_ 10 /* \s, [:space:] */ | |
1524 | # define CC_BLANK_ 11 /* [:blank:] */ | |
1525 | # define CC_XDIGIT_ 12 /* [:xdigit:] */ | |
1526 | # define CC_CNTRL_ 13 /* [:cntrl:] */ | |
1527 | # define CC_ASCII_ 14 /* [:ascii:] */ | |
1528 | # define CC_VERTSPACE_ 15 /* \v */ | |
1529 | ||
1530 | # define HIGHEST_REGCOMP_DOT_H_SYNC_ CC_VERTSPACE_ | |
a0947d7b | 1531 | |
1709d539 | 1532 | /* The members of the third group below do not need to be coordinated with data |
3ffc8c70 | 1533 | * structures in regcomp.[ch] and regexec.c. */ |
91456fff KW |
1534 | # define CC_IDFIRST_ 16 |
1535 | # define CC_CHARNAME_CONT_ 17 | |
1536 | # define CC_NONLATIN1_FOLD_ 18 | |
1537 | # define CC_NONLATIN1_SIMPLE_FOLD_ 19 | |
1538 | # define CC_QUOTEMETA_ 20 | |
1539 | # define CC_NON_FINAL_FOLD_ 21 | |
1540 | # define CC_IS_IN_SOME_FOLD_ 22 | |
1541 | # define CC_BINDIGIT_ 23 | |
1542 | # define CC_OCTDIGIT_ 24 | |
1543 | # define CC_MNEMONIC_CNTRL_ 25 | |
073c22b3 | 1544 | |
51b58dba | 1545 | /* Unused: 26-31 |
f4cdb42c KW |
1546 | * If more bits are needed, one could add a second word for non-64bit |
1547 | * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd | |
37ede926 KW |
1548 | * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it |
1549 | * is used only for optimization (as of this writing), and differs in the | |
1550 | * Latin1 range from the ALPHA bit only in two relatively unimportant | |
a500dc72 | 1551 | * characters: the masculine and feminine ordinal indicators, so removing it |
073c22b3 KW |
1552 | * would just cause /i regexes which match them to run less efficiently. |
1553 | * Similarly the EBCDIC-only bits are used just for speed, and could be | |
1554 | * replaced by other means */ | |
96ac0975 | 1555 | |
3a371f2f KW |
1556 | #if defined(PERL_CORE) || defined(PERL_EXT) |
1557 | /* An enum version of the character class numbers, to help compilers | |
1558 | * optimize */ | |
1559 | typedef enum { | |
91456fff KW |
1560 | CC_ENUM_ALPHA_ = CC_ALPHA_, |
1561 | CC_ENUM_ALPHANUMERIC_ = CC_ALPHANUMERIC_, | |
1562 | CC_ENUM_ASCII_ = CC_ASCII_, | |
1563 | CC_ENUM_BLANK_ = CC_BLANK_, | |
1564 | CC_ENUM_CASED_ = CC_CASED_, | |
1565 | CC_ENUM_CNTRL_ = CC_CNTRL_, | |
1566 | CC_ENUM_DIGIT_ = CC_DIGIT_, | |
1567 | CC_ENUM_GRAPH_ = CC_GRAPH_, | |
1568 | CC_ENUM_LOWER_ = CC_LOWER_, | |
1569 | CC_ENUM_PRINT_ = CC_PRINT_, | |
1570 | CC_ENUM_PUNCT_ = CC_PUNCT_, | |
1571 | CC_ENUM_SPACE_ = CC_SPACE_, | |
1572 | CC_ENUM_UPPER_ = CC_UPPER_, | |
1573 | CC_ENUM_VERTSPACE_ = CC_VERTSPACE_, | |
1574 | CC_ENUM_WORDCHAR_ = CC_WORDCHAR_, | |
1575 | CC_ENUM_XDIGIT_ = CC_XDIGIT_ | |
1576 | } char_class_number_; | |
3a371f2f KW |
1577 | #endif |
1578 | ||
91456fff | 1579 | #define POSIX_CC_COUNT (HIGHEST_REGCOMP_DOT_H_SYNC_ + 1) |
63c61c3f | 1580 | |
6635f04f | 1581 | START_EXTERN_C |
96ac0975 NC |
1582 | # ifdef DOINIT |
1583 | EXTCONST U32 PL_charclass[] = { | |
1584 | # include "l1_char_class_tab.h" | |
1585 | }; | |
1586 | ||
1587 | # else /* ! DOINIT */ | |
1588 | EXTCONST U32 PL_charclass[]; | |
1589 | # endif | |
6635f04f | 1590 | END_EXTERN_C |
96ac0975 | 1591 | |
265c1f46 | 1592 | /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */ |
91456fff | 1593 | # define CC_mask_(classnum) (1U << (classnum)) |
4650c663 KW |
1594 | |
1595 | /* For internal core Perl use only: the base macro for defining macros like | |
1596 | * isALPHA */ | |
6eb62d23 | 1597 | # define generic_isCC_(c, classnum) cBOOL(FITS_IN_8_BITS(c) \ |
91456fff | 1598 | && (PL_charclass[(U8) (c)] & CC_mask_(classnum))) |
4eeeb416 | 1599 | |
f4cdb42c KW |
1600 | /* The mask for the _A versions of the macros; it just adds in the bit for |
1601 | * ASCII. */ | |
91456fff | 1602 | # define CC_mask_A_(classnum) (CC_mask_(classnum) | CC_mask_(CC_ASCII_)) |
f4cdb42c | 1603 | |
4650c663 KW |
1604 | /* For internal core Perl use only: the base macro for defining macros like |
1605 | * isALPHA_A. The foo_A version makes sure that both the desired bit and | |
1606 | * the ASCII bit are present */ | |
6eb62d23 | 1607 | # define generic_isCC_A_(c, classnum) (FITS_IN_8_BITS(c) \ |
91456fff KW |
1608 | && ((PL_charclass[(U8) (c)] & CC_mask_A_(classnum)) \ |
1609 | == CC_mask_A_(classnum))) | |
f4cdb42c | 1610 | |
26c1d9d8 KW |
1611 | /* On ASCII platforms certain classes form a single range. It's faster to |
1612 | * special case these. isDIGIT is a single range on all platforms */ | |
b877c1ff | 1613 | # ifdef EBCDIC |
91456fff KW |
1614 | # define isALPHA_A(c) generic_isCC_A_(c, CC_ALPHA_) |
1615 | # define isGRAPH_A(c) generic_isCC_A_(c, CC_GRAPH_) | |
1616 | # define isLOWER_A(c) generic_isCC_A_(c, CC_LOWER_) | |
1617 | # define isPRINT_A(c) generic_isCC_A_(c, CC_PRINT_) | |
1618 | # define isUPPER_A(c) generic_isCC_A_(c, CC_UPPER_) | |
b877c1ff | 1619 | # else |
26c1d9d8 | 1620 | /* By folding the upper and lowercase, we can use a single range */ |
b877c1ff | 1621 | # define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z') |
26c1d9d8 | 1622 | # define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e) |
b877c1ff KW |
1623 | # define isLOWER_A(c) inRANGE(c, 'a', 'z') |
1624 | # define isPRINT_A(c) inRANGE(c, ' ', 0x7e) | |
1625 | # define isUPPER_A(c) inRANGE(c, 'A', 'Z') | |
1626 | # endif | |
91456fff KW |
1627 | # define isALPHANUMERIC_A(c) generic_isCC_A_(c, CC_ALPHANUMERIC_) |
1628 | # define isBLANK_A(c) generic_isCC_A_(c, CC_BLANK_) | |
1629 | # define isCNTRL_A(c) generic_isCC_A_(c, CC_CNTRL_) | |
b877c1ff | 1630 | # define isDIGIT_A(c) inRANGE(c, '0', '9') |
91456fff KW |
1631 | # define isPUNCT_A(c) generic_isCC_A_(c, CC_PUNCT_) |
1632 | # define isSPACE_A(c) generic_isCC_A_(c, CC_SPACE_) | |
1633 | # define isWORDCHAR_A(c) generic_isCC_A_(c, CC_WORDCHAR_) | |
1634 | # define isXDIGIT_A(c) generic_isCC_(c, CC_XDIGIT_) /* No non-ASCII xdigits | |
b7d90381 | 1635 | */ |
91456fff KW |
1636 | # define isIDFIRST_A(c) generic_isCC_A_(c, CC_IDFIRST_) |
1637 | # define isALPHA_L1(c) generic_isCC_(c, CC_ALPHA_) | |
1638 | # define isALPHANUMERIC_L1(c) generic_isCC_(c, CC_ALPHANUMERIC_) | |
1639 | # define isBLANK_L1(c) generic_isCC_(c, CC_BLANK_) | |
3ded5eb0 KW |
1640 | |
1641 | /* continuation character for legal NAME in \N{NAME} */ | |
91456fff | 1642 | # define isCHARNAME_CONT(c) generic_isCC_(c, CC_CHARNAME_CONT_) |
3ded5eb0 | 1643 | |
91456fff KW |
1644 | # define isCNTRL_L1(c) generic_isCC_(c, CC_CNTRL_) |
1645 | # define isGRAPH_L1(c) generic_isCC_(c, CC_GRAPH_) | |
1646 | # define isLOWER_L1(c) generic_isCC_(c, CC_LOWER_) | |
1647 | # define isPRINT_L1(c) generic_isCC_(c, CC_PRINT_) | |
b7d90381 | 1648 | # define isPSXSPC_L1(c) isSPACE_L1(c) |
91456fff KW |
1649 | # define isPUNCT_L1(c) generic_isCC_(c, CC_PUNCT_) |
1650 | # define isSPACE_L1(c) generic_isCC_(c, CC_SPACE_) | |
1651 | # define isUPPER_L1(c) generic_isCC_(c, CC_UPPER_) | |
1652 | # define isWORDCHAR_L1(c) generic_isCC_(c, CC_WORDCHAR_) | |
1653 | # define isIDFIRST_L1(c) generic_isCC_(c, CC_IDFIRST_) | |
f4cdb42c | 1654 | |
0852beac | 1655 | # ifdef EBCDIC |
91456fff | 1656 | # define isASCII(c) generic_isCC_(c, CC_ASCII_) |
0852beac KW |
1657 | # endif |
1658 | ||
f12c0118 | 1659 | /* Participates in a single-character fold with a character above 255 */ |
c62fdeb7 | 1660 | # if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_REGEXEC_C) |
81d43abf | 1661 | # define HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(c) \ |
c62fdeb7 | 1662 | (( ! cBOOL(FITS_IN_8_BITS(c))) \ |
91456fff | 1663 | || (PL_charclass[(U8) (c)] & CC_mask_(CC_NONLATIN1_SIMPLE_FOLD_))) |
c62fdeb7 | 1664 | |
91456fff KW |
1665 | # define IS_NON_FINAL_FOLD(c) generic_isCC_(c, CC_NON_FINAL_FOLD_) |
1666 | # define IS_IN_SOME_FOLD_L1(c) generic_isCC_(c, CC_IS_IN_SOME_FOLD_) | |
c62fdeb7 | 1667 | # endif |
f12c0118 KW |
1668 | |
1669 | /* Like the above, but also can be part of a multi-char fold */ | |
c62fdeb7 KW |
1670 | # define HAS_NONLATIN1_FOLD_CLOSURE(c) \ |
1671 | ( (! cBOOL(FITS_IN_8_BITS(c))) \ | |
91456fff | 1672 | || (PL_charclass[(U8) (c)] & CC_mask_(CC_NONLATIN1_FOLD_))) |
430b7c70 | 1673 | |
91456fff | 1674 | # define _isQUOTEMETA(c) generic_isCC_(c, CC_QUOTEMETA_) |
5e6ebb12 KW |
1675 | |
1676 | /* is c a control character for which we have a mnemonic? */ | |
1677 | # if defined(PERL_CORE) || defined(PERL_EXT) | |
91456fff | 1678 | # define isMNEMONIC_CNTRL(c) generic_isCC_(c, CC_MNEMONIC_CNTRL_) |
5e6ebb12 | 1679 | # endif |
687c8d01 | 1680 | #else /* else we don't have perl.h H_PERL */ |
3ded5eb0 KW |
1681 | |
1682 | /* If we don't have perl.h, we are compiling a utility program. Below we | |
1683 | * hard-code various macro definitions that wouldn't otherwise be available | |
fc273927 | 1684 | * to it. Most are coded based on first principles. These are written to |
74665a89 | 1685 | * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */ |
182c4ace | 1686 | # define isDIGIT_A(c) inRANGE(c, '0', '9') |
0852beac | 1687 | # define isBLANK_A(c) ((c) == ' ' || (c) == '\t') |
74665a89 KW |
1688 | # define isSPACE_A(c) (isBLANK_A(c) \ |
1689 | || (c) == '\n' \ | |
1690 | || (c) == '\r' \ | |
1691 | || (c) == '\v' \ | |
0852beac | 1692 | || (c) == '\f') |
74665a89 KW |
1693 | /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for |
1694 | * uppercase. The tests for those aren't necessary on ASCII, but hurt only | |
1695 | * performance (if optimization isn't on), and allow the same code to be | |
1696 | * used for both platform types */ | |
182c4ace KW |
1697 | # define isLOWER_A(c) inRANGE((c), 'a', 'i') \ |
1698 | || inRANGE((c), 'j', 'r') \ | |
1699 | || inRANGE((c), 's', 'z') | |
1700 | # define isUPPER_A(c) inRANGE((c), 'A', 'I') \ | |
1701 | || inRANGE((c), 'J', 'R') \ | |
1702 | || inRANGE((c), 'S', 'Z') | |
a4d7a999 KW |
1703 | # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c)) |
1704 | # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c)) | |
3ded5eb0 | 1705 | # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_') |
0852beac | 1706 | # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_') |
182c4ace KW |
1707 | # define isXDIGIT_A(c) ( isDIGIT_A(c) \ |
1708 | || inRANGE((c), 'a', 'f') \ | |
1709 | || inRANGE((c), 'A', 'F') | |
74665a89 KW |
1710 | # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \ |
1711 | || (c) == '#' || (c) == '$' || (c) == '%' \ | |
1712 | || (c) == '&' || (c) == '\'' || (c) == '(' \ | |
1713 | || (c) == ')' || (c) == '*' || (c) == '+' \ | |
1714 | || (c) == ',' || (c) == '.' || (c) == '/' \ | |
1715 | || (c) == ':' || (c) == ';' || (c) == '<' \ | |
1716 | || (c) == '=' || (c) == '>' || (c) == '?' \ | |
1717 | || (c) == '@' || (c) == '[' || (c) == '\\' \ | |
1718 | || (c) == ']' || (c) == '^' || (c) == '_' \ | |
1719 | || (c) == '`' || (c) == '{' || (c) == '|' \ | |
1720 | || (c) == '}' || (c) == '~') | |
1721 | # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c)) | |
1722 | # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ') | |
3ded5eb0 | 1723 | |
0852beac | 1724 | # ifdef EBCDIC |
74665a89 KW |
1725 | /* The below is accurate for the 3 EBCDIC code pages traditionally |
1726 | * supported by perl. The only difference between them in the controls | |
1727 | * is the position of \n, and that is represented symbolically below */ | |
1728 | # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \ | |
1729 | || (c) == '\f' || (c) == '\n' || (c) == '\r' \ | |
1730 | || (c) == '\t' || (c) == '\v' \ | |
182c4ace | 1731 | || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \ |
8ec0a736 | 1732 | || (c) == 7F /* U+7F DEL */ \ |
182c4ace KW |
1733 | || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \ |
1734 | DC[1-3] */ \ | |
74665a89 KW |
1735 | || (c) == 0x18 /* U+18 CAN */ \ |
1736 | || (c) == 0x19 /* U+19 EOM */ \ | |
182c4ace | 1737 | || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \ |
74665a89 KW |
1738 | || (c) == 0x26 /* U+17 ETB */ \ |
1739 | || (c) == 0x27 /* U+1B ESC */ \ | |
1740 | || (c) == 0x2D /* U+05 ENQ */ \ | |
1741 | || (c) == 0x2E /* U+06 ACK */ \ | |
1742 | || (c) == 0x32 /* U+16 SYN */ \ | |
1743 | || (c) == 0x37 /* U+04 EOT */ \ | |
1744 | || (c) == 0x3C /* U+14 DC4 */ \ | |
1745 | || (c) == 0x3D /* U+15 NAK */ \ | |
1746 | || (c) == 0x3F)/* U+1A SUB */ | |
0852beac | 1747 | # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c)) |
74665a89 KW |
1748 | # else /* isASCII is already defined for ASCII platforms, so can use that to |
1749 | define isCNTRL */ | |
1750 | # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c)) | |
0852beac KW |
1751 | # endif |
1752 | ||
3ffc8c70 | 1753 | /* The _L1 macros may be unnecessary for the utilities; I (khw) added them |
caa94d35 KW |
1754 | * during debugging, and it seems best to keep them. We may be called |
1755 | * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't | |
1756 | * do anything anyway, so make it not a problem */ | |
1757 | # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1) | |
1758 | # define NATIVE_TO_LATIN1(ch) (ch) | |
1759 | # endif | |
3ded5eb0 KW |
1760 | # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c)) |
1761 | # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c)) | |
1762 | # define isBLANK_L1(c) (isBLANK_A(c) \ | |
1763 | || (FITS_IN_8_BITS(c) \ | |
1764 | && NATIVE_TO_LATIN1((U8) c) == 0xA0)) | |
1765 | # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c))) | |
1766 | # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c))) | |
1767 | # define isLOWER_L1(c) (isLOWER_A(c) \ | |
1768 | || (FITS_IN_8_BITS(c) \ | |
ae683a5f | 1769 | && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \ |
3ded5eb0 KW |
1770 | && NATIVE_TO_LATIN1((U8) c) != 0xF7) \ |
1771 | || NATIVE_TO_LATIN1((U8) c) == 0xAA \ | |
1772 | || NATIVE_TO_LATIN1((U8) c) == 0xBA \ | |
1773 | || NATIVE_TO_LATIN1((U8) c) == 0xB5))) | |
1774 | # define isPRINT_L1(c) (isPRINT_A(c) \ | |
1775 | || (FITS_IN_8_BITS(c) \ | |
1776 | && NATIVE_TO_LATIN1((U8) c) >= 0xA0)) | |
3ded5eb0 KW |
1777 | # define isPUNCT_L1(c) (isPUNCT_A(c) \ |
1778 | || (FITS_IN_8_BITS(c) \ | |
ae683a5f | 1779 | && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \ |
3ded5eb0 KW |
1780 | || NATIVE_TO_LATIN1((U8) c) == 0xA7 \ |
1781 | || NATIVE_TO_LATIN1((U8) c) == 0xAB \ | |
1782 | || NATIVE_TO_LATIN1((U8) c) == 0xB6 \ | |
1783 | || NATIVE_TO_LATIN1((U8) c) == 0xB7 \ | |
1784 | || NATIVE_TO_LATIN1((U8) c) == 0xBB \ | |
1785 | || NATIVE_TO_LATIN1((U8) c) == 0xBF))) | |
1786 | # define isSPACE_L1(c) (isSPACE_A(c) \ | |
1787 | || (FITS_IN_8_BITS(c) \ | |
ae683a5f | 1788 | && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \ |
3ded5eb0 KW |
1789 | || NATIVE_TO_LATIN1((U8) c) == 0xA0))) |
1790 | # define isUPPER_L1(c) (isUPPER_A(c) \ | |
1791 | || (FITS_IN_8_BITS(c) \ | |
182c4ace KW |
1792 | && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \ |
1793 | 0xC0, 0xDE) \ | |
3ded5eb0 KW |
1794 | && NATIVE_TO_LATIN1((U8) c) != 0xD7))) |
1795 | # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c)) | |
1796 | # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_') | |
1797 | # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \ | |
1798 | || isBLANK_L1(c) \ | |
1799 | || (c) == '-' \ | |
1800 | || (c) == '(' \ | |
1801 | || (c) == ')') | |
1802 | /* The following are not fully accurate in the above-ASCII range. I (khw) | |
1803 | * don't think it's necessary to be so for the purposes where this gets | |
1804 | * compiled */ | |
6eb62d23 | 1805 | # define isQUOTEMETA_(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c)) |
3ded5eb0 KW |
1806 | # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c) |
1807 | ||
1808 | /* And these aren't accurate at all. They are useful only for above | |
1809 | * Latin1, which utilities and bootstrapping don't deal with */ | |
1810 | # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0 | |
6838b41e | 1811 | # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0 |
3ded5eb0 KW |
1812 | # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0 |
1813 | ||
1814 | /* Many of the macros later in this file are defined in terms of these. By | |
1815 | * implementing them with a function, which converts the class number into | |
1816 | * a call to the desired macro, all of the later ones work. However, that | |
1817 | * function won't be actually defined when building a utility program (no | |
1818 | * perl.h), and so a compiler error will be generated if one is attempted | |
1819 | * to be used. And the above-Latin1 code points require Unicode tables to | |
1820 | * be present, something unlikely to be the case when bootstrapping */ | |
6eb62d23 | 1821 | # define generic_isCC_(c, classnum) \ |
3ded5eb0 | 1822 | (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE)) |
6eb62d23 | 1823 | # define generic_isCC_A_(c, classnum) \ |
3ded5eb0 | 1824 | (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE)) |
687c8d01 | 1825 | #endif /* End of no perl.h H_PERL */ |
8a58bdcf | 1826 | |
e66b99e9 KW |
1827 | #define isALPHANUMERIC(c) isALPHANUMERIC_A(c) |
1828 | #define isALPHA(c) isALPHA_A(c) | |
0852beac KW |
1829 | #define isASCII_A(c) isASCII(c) |
1830 | #define isASCII_L1(c) isASCII(c) | |
e66b99e9 KW |
1831 | #define isBLANK(c) isBLANK_A(c) |
1832 | #define isCNTRL(c) isCNTRL_A(c) | |
1833 | #define isDIGIT(c) isDIGIT_A(c) | |
1834 | #define isGRAPH(c) isGRAPH_A(c) | |
1835 | #define isIDFIRST(c) isIDFIRST_A(c) | |
1836 | #define isLOWER(c) isLOWER_A(c) | |
1837 | #define isPRINT(c) isPRINT_A(c) | |
779cf272 | 1838 | #define isPSXSPC_A(c) isSPACE_A(c) |
e66b99e9 | 1839 | #define isPSXSPC(c) isPSXSPC_A(c) |
779cf272 | 1840 | #define isPSXSPC_L1(c) isSPACE_L1(c) |
e66b99e9 KW |
1841 | #define isPUNCT(c) isPUNCT_A(c) |
1842 | #define isSPACE(c) isSPACE_A(c) | |
1843 | #define isUPPER(c) isUPPER_A(c) | |
1844 | #define isWORDCHAR(c) isWORDCHAR_A(c) | |
1845 | #define isXDIGIT(c) isXDIGIT_A(c) | |
1846 | ||
1847 | /* ASCII casing. These could also be written as | |
1848 | #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c)) | |
1849 | #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c)) | |
1850 | which uses table lookup and mask instead of subtraction. (This would | |
c5e9991e KW |
1851 | work because the _MOD does not apply in the ASCII range). |
1852 | ||
1853 | These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII | |
1854 | UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are | |
1855 | there non-ASCII invariants, and all of them are controls.) */ | |
68067e4e DM |
1856 | #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c)) |
1857 | #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c)) | |
bbce6d69 | 1858 | |
25200305 KW |
1859 | /* In the ASCII range, these are equivalent to what they're here defined to be. |
1860 | * But by creating these definitions, other code doesn't have to be aware of | |
c5e9991e KW |
1861 | * this detail. Actually this works for all UTF-8 invariants, not just the |
1862 | * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */ | |
25200305 | 1863 | #define toFOLD(c) toLOWER(c) |
25200305 KW |
1864 | #define toTITLE(c) toUPPER(c) |
1865 | ||
c753c8d3 KW |
1866 | #define toLOWER_A(c) toLOWER(c) |
1867 | #define toUPPER_A(c) toUPPER(c) | |
25200305 KW |
1868 | #define toFOLD_A(c) toFOLD(c) |
1869 | #define toTITLE_A(c) toTITLE(c) | |
1a0901db | 1870 | |
4650c663 | 1871 | /* Use table lookup for speed; returns the input itself if is out-of-range */ |
b2bf251f | 1872 | #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \ |
8e7c6e7d | 1873 | ? (c) \ |
f4cd282c | 1874 | : PL_latin1_lc[ (U8) (c) ]) |
c753c8d3 KW |
1875 | #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */ |
1876 | ||
1a0901db | 1877 | /* Modified uc. Is correct uc except for three non-ascii chars which are |
4650c663 KW |
1878 | * all mapped to one of them, and these need special handling; returns the |
1879 | * input itself if is out-of-range */ | |
b2bf251f | 1880 | #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \ |
8e7c6e7d | 1881 | ? (c) \ |
f4cd282c | 1882 | : PL_mod_latin1_uc[ (U8) (c) ]) |
31f05a37 | 1883 | #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale |
84061b6a | 1884 | |
beab9ebe KW |
1885 | /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */ |
1886 | ||
1887 | /* For internal core Perl use only: the base macro for defining macros like | |
1888 | * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point | |
31f05a37 | 1889 | * (0-255) to check. In a UTF-8 locale, the result is the same as calling |
53049083 KW |
1890 | * isFOO_L1(); 'classnum' is something like CC_UPPER_, which gives the class |
1891 | * number for doing this. For non-UTF-8 locales, the code to actually do the | |
1892 | * test this is passed in 'non_utf8'. If 'c' is above 255, 0 is returned. For | |
1893 | * accessing the full range of possible code points under locale rules, use the | |
1894 | * macros based on generic_LC_uvchr_ instead of this. */ | |
1895 | #define generic_LC_base_(c, classnum, non_utf8_func) \ | |
81d43abf KW |
1896 | (! FITS_IN_8_BITS(c) \ |
1897 | ? 0 \ | |
1898 | : IN_UTF8_CTYPE_LOCALE \ | |
53049083 KW |
1899 | ? cBOOL(PL_charclass[(U8) (c)] & CC_mask_(classnum)) \ |
1900 | : cBOOL(non_utf8_func(c))) | |
beab9ebe | 1901 | |
ef620431 KW |
1902 | /* A helper macro for defining macros like isALPHA_LC. On systems without |
1903 | * proper locales, these reduce to, e.g., isALPHA_A */ | |
1904 | #ifdef CTYPE256 | |
1905 | # define generic_LC_(c, classnum, non_utf8_func) \ | |
1906 | generic_LC_base_(c, classnum, non_utf8_func) | |
1907 | #else | |
1908 | # define generic_LC_(c, classnum, non_utf8_func) \ | |
1909 | generic_isCC_A_(c, classnum) | |
1910 | #endif | |
beab9ebe | 1911 | |
33bdb9d3 KW |
1912 | /* Below are the definitions for the locale-sensitive character classification |
1913 | * macros whose input domain is a byte, and the locale isn't UTF-8. These are | |
1914 | * as close as possible to the bare versions on the platform and still yield | |
1915 | * POSIX Standard-compliant results. | |
1916 | * | |
1917 | * There is currently only one place these definitions should be used, in | |
1918 | * certain function calls like Perl_iswordchar_() in inline.h. | |
1919 | * | |
1920 | * Most likely you want to use the macros a ways below with names like | |
1921 | * isALPHA_LC(). Rarely, you may want isU8_ALPHA_LC(), somewhat below. | |
1922 | * | |
1923 | * The first two aren't in C89, so the fallback is to use the non-locale | |
1924 | * sensitive versions; these are the same for all platforms */ | |
f05550c0 | 1925 | #if defined(HAS_ISASCII) |
cbc5b6f1 | 1926 | # define is_base_ASCII(c) isascii((U8) (c)) |
84061b6a | 1927 | #else |
cbc5b6f1 | 1928 | # define is_base_ASCII(c) isASCII(c) |
84061b6a KW |
1929 | #endif |
1930 | ||
f05550c0 | 1931 | #if defined(HAS_ISBLANK) |
cbc5b6f1 | 1932 | # define is_base_BLANK(c) isblank((U8) (c)) |
6d432bcf | 1933 | #else |
cbc5b6f1 | 1934 | # define is_base_BLANK(c) isBLANK(c) |
84061b6a KW |
1935 | #endif |
1936 | ||
6d432bcf | 1937 | /* The next few are the same in all platforms. */ |
cbc5b6f1 KW |
1938 | #define is_base_CNTRL(c) iscntrl((U8) (c)) |
1939 | #define is_base_IDFIRST(c) (UNLIKELY((c) == '_') || is_base_ALPHA(c)) | |
1940 | #define is_base_SPACE(c) isspace((U8) (c)) | |
1941 | #define is_base_WORDCHAR(c) (UNLIKELY((c) == '_') || is_base_ALPHANUMERIC(c)) | |
3f0486a3 | 1942 | |
33bdb9d3 | 1943 | /* The base-level case changing macros are also the same in all platforms */ |
cbc5b6f1 KW |
1944 | #define to_base_LOWER(c) tolower((U8) (c)) |
1945 | #define to_base_UPPER(c) toupper((U8) (c)) | |
1946 | #define to_base_FOLD(c) to_base_LOWER(c) | |
33bdb9d3 KW |
1947 | |
1948 | #ifdef WIN32 | |
81d43abf KW |
1949 | |
1950 | /* The Windows functions don't bother to follow the POSIX standard, which for | |
1951 | * example says that something can't both be a printable and a control. But | |
6d432bcf KW |
1952 | * Windows treats \t as both a control and a printable, and does such things as |
1953 | * making superscripts into both digits and punctuation. These #defines tame | |
1954 | * these flaws by assuming that the definitions of controls (and the other few | |
1955 | * ones defined above) are correct, and then making sure that other definitions | |
1956 | * don't have weirdnesses, by adding a check that \w and its subsets aren't | |
1957 | * ispunct(), and things that are \W, like ispunct(), arent't controls. Not | |
1958 | * all possible weirdnesses are checked for, just ones that were detected on | |
1959 | * actual Microsoft code pages */ | |
cbc5b6f1 KW |
1960 | # define is_base_ALPHA(c) \ |
1961 | (isalpha((U8) (c)) && ! is_base_PUNCT(c)) | |
1962 | # define is_base_ALPHANUMERIC(c) \ | |
1963 | (isalnum((U8) (c)) && ! is_base_PUNCT(c)) | |
1964 | # define is_base_CASED(c) \ | |
1965 | ((isupper((U8) (c)) || islower((U8) (c))) && ! is_base_PUNCT(c)) | |
1966 | # define is_base_DIGIT(c) \ | |
1967 | (isdigit((U8) (c)) && ! is_base_PUNCT(c)) | |
1968 | # define is_base_GRAPH(c) \ | |
1969 | (isgraph((U8) (c)) && ! is_base_CNTRL(c)) | |
1970 | # define is_base_LOWER(c) \ | |
1971 | (islower((U8) (c)) && ! is_base_PUNCT(c)) | |
1972 | # define is_base_PRINT(c) \ | |
1973 | (isprint((U8) (c)) && ! is_base_CNTRL(c)) | |
1974 | # define is_base_PUNCT(c) \ | |
1975 | (ispunct((U8) (c)) && ! is_base_CNTRL(c)) | |
1976 | # define is_base_UPPER(c) \ | |
1977 | (isupper((U8) (c)) && ! is_base_PUNCT(c)) | |
1978 | # define is_base_XDIGIT(c) \ | |
1979 | (isxdigit((U8) (c)) && ! is_base_PUNCT(c)) | |
6d432bcf KW |
1980 | #else |
1981 | ||
33bdb9d3 KW |
1982 | /* For all other platforms, as far as we know, isdigit(), etc. work sanely |
1983 | * enough */ | |
cbc5b6f1 KW |
1984 | # define is_base_ALPHA(c) isalpha((U8) (c)) |
1985 | # define is_base_ALPHANUMERIC(c) isalnum((U8) (c)) | |
1986 | # define is_base_CASED(c) (islower((U8) (c)) || isupper((U8) (c))) | |
1987 | # define is_base_DIGIT(c) isdigit((U8) (c)) | |
6d432bcf KW |
1988 | |
1989 | /* ... But it seems that IBM products treat NBSP as both a space and a | |
1990 | * graphic; these are the two platforms that we have active test beds for. | |
1991 | */ | |
1992 | # if defined(OS390) || defined(_AIX) | |
cbc5b6f1 | 1993 | # define is_base_GRAPH(c) (isgraph((U8) (c)) && ! isspace((U8) (c))) |
6d432bcf | 1994 | # else |
cbc5b6f1 | 1995 | # define is_base_GRAPH(c) isgraph((U8) (c)) |
3f0486a3 | 1996 | # endif |
cbc5b6f1 KW |
1997 | # define is_base_LOWER(c) islower((U8) (c)) |
1998 | # define is_base_PRINT(c) isprint((U8) (c)) | |
1999 | # define is_base_PUNCT(c) ispunct((U8) (c)) | |
2000 | # define is_base_UPPER(c) isupper((U8) (c)) | |
2001 | # define is_base_XDIGIT(c) isxdigit((U8) (c)) | |
6d432bcf KW |
2002 | #endif |
2003 | ||
33bdb9d3 KW |
2004 | /* Below is the next level up, which currently expands to nothing more |
2005 | * than the previous layer. These are the macros to use if you really need | |
2006 | * something whose input domain is a byte, and the locale isn't UTF-8; that is, | |
2007 | * where you normally would have to use things like bare isalnum(). | |
2008 | * | |
2009 | * But most likely you should instead use the layer defined further below which | |
2010 | * has names like isALPHA_LC. They deal with larger-than-byte inputs, and | |
2011 | * UTF-8 locales. | |
2012 | * | |
2013 | * (Note, proper general operation of the bare libc functons requires you to | |
2014 | * cast to U8. These do that for you automatically.) */ | |
2015 | ||
cbc5b6f1 | 2016 | # define WRAP_U8_LC_(c, classnum, base) base(c) |
33bdb9d3 KW |
2017 | |
2018 | #define isU8_ALPHANUMERIC_LC(c) \ | |
cbc5b6f1 KW |
2019 | WRAP_U8_LC_((c), CC_ALPHANUMERIC_, is_base_ALPHANUMERIC) |
2020 | #define isU8_ALPHA_LC(c) WRAP_U8_LC_((c), CC_ALPHA_, is_base_ALPHA) | |
2021 | #define isU8_ASCII_LC(c) WRAP_U8_LC_((c), CC_ASCII_, is_base_ASCII) | |
2022 | #define isU8_BLANK_LC(c) WRAP_U8_LC_((c), CC_BLANK_, is_base_BLANK) | |
2023 | #define isU8_CASED_LC(c) WRAP_U8_LC_((c), CC_CASED_, is_base_CASED) | |
2024 | #define isU8_CNTRL_LC(c) WRAP_U8_LC_((c), CC_CNTRL_, is_base_CNTRL) | |
2025 | #define isU8_DIGIT_LC(c) WRAP_U8_LC_((c), CC_DIGIT_, is_base_DIGIT) | |
2026 | #define isU8_GRAPH_LC(c) WRAP_U8_LC_((c), CC_GRAPH_, is_base_GRAPH) | |
2027 | #define isU8_IDFIRST_LC(c) WRAP_U8_LC_((c), CC_IDFIRST_, is_base_IDFIRST) | |
2028 | #define isU8_LOWER_LC(c) WRAP_U8_LC_((c), CC_LOWER_, is_base_LOWER) | |
2029 | #define isU8_PRINT_LC(c) WRAP_U8_LC_((c), CC_PRINT_, is_base_PRINT) | |
2030 | #define isU8_PUNCT_LC(c) WRAP_U8_LC_((c), CC_PUNCT_, is_base_PUNCT) | |
2031 | #define isU8_SPACE_LC(c) WRAP_U8_LC_((c), CC_SPACE_, is_base_SPACE) | |
2032 | #define isU8_UPPER_LC(c) WRAP_U8_LC_((c), CC_UPPER_, is_base_UPPER) | |
2033 | #define isU8_WORDCHAR_LC(c) WRAP_U8_LC_((c), CC_WORDCHAR_, is_base_WORDCHAR) | |
2034 | #define isU8_XDIGIT_LC(c) WRAP_U8_LC_((c), CC_XDIGIT_, is_base_XDIGIT) | |
2035 | ||
2036 | #define toU8_LOWER_LC(c) WRAP_U8_LC_((c), CC_TOLOWER_, to_base_LOWER) | |
2037 | #define toU8_UPPER_LC(c) WRAP_U8_LC_((c), CC_TOUPPER_, to_base_UPPER) | |
33bdb9d3 KW |
2038 | #define toU8_FOLD_LC(c) toU8_LOWER_LC(c) |
2039 | ||
6d432bcf KW |
2040 | /* The definitions below use the ones above to create versions in which the |
2041 | * input domain isn't restricted to bytes (though always returning false if the | |
2042 | * input doesn't fit in a byte), and to behave properly should the locale be | |
33bdb9d3 KW |
2043 | * UTF-8. These are the documented ones, suitable for general use (though |
2044 | * toUPPER_LC and toFOLD_LC aren't documented because they need special | |
2045 | * handling to deal with SHARP S expanding to two characters). */ | |
2046 | ||
6d432bcf KW |
2047 | #define isASCII_LC(c) (FITS_IN_8_BITS(c) && isU8_ASCII_LC(c)) |
2048 | #define isALPHA_LC(c) generic_LC_(c, CC_ALPHA_, isU8_ALPHA_LC) | |
2049 | #define isALPHANUMERIC_LC(c) \ | |
2050 | generic_LC_(c, CC_ALPHANUMERIC_, isU8_ALPHANUMERIC_LC) | |
2051 | #define isBLANK_LC(c) generic_LC_(c, CC_BLANK_, isU8_BLANK_LC) | |
8fd8ea43 | 2052 | #define isCASED_LC(c) generic_LC_(c, CC_CASED_, isU8_CASED_LC) |
6d432bcf KW |
2053 | #define isCNTRL_LC(c) generic_LC_(c, CC_CNTRL_, isU8_CNTRL_LC) |
2054 | #define isDIGIT_LC(c) generic_LC_(c, CC_DIGIT_, isU8_DIGIT_LC) | |
2055 | #define isGRAPH_LC(c) generic_LC_(c, CC_GRAPH_, isU8_GRAPH_LC) | |
2056 | #define isIDFIRST_LC(c) generic_LC_(c, CC_IDFIRST_, isU8_IDFIRST_LC) | |
2057 | #define isLOWER_LC(c) generic_LC_(c, CC_LOWER_, isU8_LOWER_LC) | |
2058 | #define isPRINT_LC(c) generic_LC_(c, CC_PRINT_, isU8_PRINT_LC) | |
2059 | #define isPUNCT_LC(c) generic_LC_(c, CC_PUNCT_, isU8_PUNCT_LC) | |
2060 | #define isSPACE_LC(c) generic_LC_(c, CC_SPACE_, isU8_SPACE_LC) | |
2061 | #define isUPPER_LC(c) generic_LC_(c, CC_UPPER_, isU8_UPPER_LC) | |
2062 | #define isWORDCHAR_LC(c) generic_LC_(c, CC_WORDCHAR_, isU8_WORDCHAR_LC) | |
2063 | #define isXDIGIT_LC(c) generic_LC_(c, CC_XDIGIT_, isU8_XDIGIT_LC) | |
4a283f4f | 2064 | |
ef620431 | 2065 | #ifndef CTYPE256 |
d277535a KW |
2066 | # define toLOWER_LC(c) toLOWER_A(c) |
2067 | # define toUPPER_LC(c) toUPPER_A(c) | |
2068 | # define toFOLD_LC(c) toFOLD_A(c) | |
ef620431 KW |
2069 | #else |
2070 | ||
4a283f4f KW |
2071 | /* In the next three macros, the reason for using the PL_latin arrays is in |
2072 | * case the system function is defective; it ensures uniform results that | |
2073 | * conform to the Unicode standard. */ | |
2074 | ||
2075 | /* This does not handle the anomalies in UTF-8 Turkic locales. */ | |
31e89ad7 | 2076 | # define toLOWER_LC(c) ((! FITS_IN_8_BITS(c)) \ |
4a283f4f KW |
2077 | ? (c) \ |
2078 | : ((IN_UTF8_CTYPE_LOCALE) \ | |
2079 | ? PL_latin1_lc[ (U8) (c) ] \ | |
6d432bcf | 2080 | : ((U8) toU8_LOWER_LC(c)))) |
4a283f4f KW |
2081 | |
2082 | /* In this macro, note that the result can be larger than a byte in a UTF-8 | |
2083 | * locale. It returns a single value, so can't adequately return the upper | |
2084 | * case of LATIN SMALL LETTER SHARP S in a UTF-8 locale (which should be a | |
2085 | * string of two values "SS"); instead it asserts against that under | |
2086 | * DEBUGGING, and otherwise returns its input. It does not handle the | |
2087 | * anomalies in UTF-8 Turkic locales. */ | |
31e89ad7 | 2088 | # define toUPPER_LC(c) \ |
4a283f4f KW |
2089 | ((! FITS_IN_8_BITS(c)) \ |
2090 | ? (c) \ | |
2091 | : ((! IN_UTF8_CTYPE_LOCALE) \ | |
6d432bcf | 2092 | ? ((U8) toU8_UPPER_LC(c)) \ |
4a283f4f KW |
2093 | : (UNLIKELY(((U8)(c)) == MICRO_SIGN) \ |
2094 | ? GREEK_CAPITAL_LETTER_MU \ | |
2095 | : ((UNLIKELY(((U8) (c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \ | |
2096 | ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \ | |
2097 | : (UNLIKELY(((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \ | |
2098 | ? (__ASSERT_(0) (c)) /* Fail on Sharp S in DEBUGGING */ \ | |
2099 | : PL_mod_latin1_uc[ (U8) (c) ])))))) | |
2100 | ||
2101 | /* In this macro, note that the result can be larger than a byte in a UTF-8 | |
2102 | * locale. It returns a single value, so can't adequately return the fold case | |
2103 | * of LATIN SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of | |
2104 | * two values "ss"); instead it asserts against that under DEBUGGING, and | |
2105 | * otherwise returns its input. It does not handle the anomalies in UTF-8 | |
2106 | * Turkic locales */ | |
31e89ad7 | 2107 | # define toFOLD_LC(c) \ |
4a283f4f KW |
2108 | ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \ |
2109 | ? GREEK_SMALL_LETTER_MU \ | |
2110 | : (__ASSERT_( ! IN_UTF8_CTYPE_LOCALE \ | |
2111 | || LIKELY((c) != LATIN_SMALL_LETTER_SHARP_S)) \ | |
31e89ad7 | 2112 | toLOWER_LC(c))) |
f05550c0 | 2113 | #endif |
55204971 | 2114 | |
eba68aa0 KW |
2115 | #define isIDCONT(c) isWORDCHAR(c) |
2116 | #define isIDCONT_A(c) isWORDCHAR_A(c) | |
2117 | #define isIDCONT_L1(c) isWORDCHAR_L1(c) | |
2118 | #define isIDCONT_LC(c) isWORDCHAR_LC(c) | |
13380643 | 2119 | #define isPSXSPC_LC(c) isSPACE_LC(c) |
aaa51d5e | 2120 | |
4650c663 | 2121 | /* For internal core Perl use only: the base macros for defining macros like |
d0da05db | 2122 | * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class |
6eb62d23 | 2123 | * number defined earlier in this file. generic_uvchr_() is used for POSIX |
4650c663 KW |
2124 | * classes where there is a macro or function 'above_latin1' that takes the |
2125 | * single argument 'c' and returns the desired value. These exist for those | |
2366ba44 | 2126 | * classes which have simple definitions, avoiding the overhead of an inversion |
6eb62d23 | 2127 | * list binary search. generic_invlist_uvchr_() can be used |
4650c663 | 2128 | * for classes where that overhead is faster than a direct lookup. |
6eb62d23 KW |
2129 | * generic_uvchr_() won't compile if 'c' isn't unsigned, as it won't match the |
2130 | * 'above_latin1' prototype. generic_isCC_() macro does bounds checking, so | |
4650c663 KW |
2131 | * have duplicate checks here, so could create versions of the macros that |
2132 | * don't, but experiments show that gcc optimizes them out anyway. */ | |
66c17564 KW |
2133 | |
2134 | /* Note that all ignore 'use bytes' */ | |
6eb62d23 KW |
2135 | #define generic_uvchr_(classnum, above_latin1, c) ((c) < 256 \ |
2136 | ? generic_isCC_(c, classnum) \ | |
cd500f2f | 2137 | : above_latin1(c)) |
81d43abf | 2138 | #define generic_invlist_uvchr_(classnum, c) ((c) < 256 \ |
6eb62d23 | 2139 | ? generic_isCC_(c, classnum) \ |
922e8cb4 | 2140 | : _is_uni_FOO(classnum, c)) |
91456fff KW |
2141 | #define isALPHA_uvchr(c) generic_invlist_uvchr_(CC_ALPHA_, c) |
2142 | #define isALPHANUMERIC_uvchr(c) generic_invlist_uvchr_(CC_ALPHANUMERIC_, c) | |
d0da05db | 2143 | #define isASCII_uvchr(c) isASCII(c) |
91456fff | 2144 | #define isBLANK_uvchr(c) generic_uvchr_(CC_BLANK_, is_HORIZWS_cp_high, c) |
d0da05db | 2145 | #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */ |
91456fff KW |
2146 | #define isDIGIT_uvchr(c) generic_invlist_uvchr_(CC_DIGIT_, c) |
2147 | #define isGRAPH_uvchr(c) generic_invlist_uvchr_(CC_GRAPH_, c) | |
1e222e4f | 2148 | #define isIDCONT_uvchr(c) \ |
91456fff | 2149 | generic_uvchr_(CC_WORDCHAR_, _is_uni_perl_idcont, c) |
1e222e4f | 2150 | #define isIDFIRST_uvchr(c) \ |
91456fff KW |
2151 | generic_uvchr_(CC_IDFIRST_, _is_uni_perl_idstart, c) |
2152 | #define isLOWER_uvchr(c) generic_invlist_uvchr_(CC_LOWER_, c) | |
2153 | #define isPRINT_uvchr(c) generic_invlist_uvchr_(CC_PRINT_, c) | |
d0da05db | 2154 | |
91456fff KW |
2155 | #define isPUNCT_uvchr(c) generic_invlist_uvchr_(CC_PUNCT_, c) |
2156 | #define isSPACE_uvchr(c) generic_uvchr_(CC_SPACE_, is_XPERLSPACE_cp_high, c) | |
d0da05db KW |
2157 | #define isPSXSPC_uvchr(c) isSPACE_uvchr(c) |
2158 | ||
91456fff KW |
2159 | #define isUPPER_uvchr(c) generic_invlist_uvchr_(CC_UPPER_, c) |
2160 | #define isVERTWS_uvchr(c) generic_uvchr_(CC_VERTSPACE_, is_VERTWS_cp_high, c) | |
2161 | #define isWORDCHAR_uvchr(c) generic_invlist_uvchr_(CC_WORDCHAR_, c) | |
2162 | #define isXDIGIT_uvchr(c) generic_uvchr_(CC_XDIGIT_, is_XDIGIT_cp_high, c) | |
d0da05db KW |
2163 | |
2164 | #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l) | |
2165 | #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l) | |
2166 | #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l) | |
2167 | #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l) | |
2168 | ||
2169 | /* For backwards compatibility, even though '_uni' should mean official Unicode | |
2170 | * code points, in Perl it means native for those below 256 */ | |
2171 | #define isALPHA_uni(c) isALPHA_uvchr(c) | |
2172 | #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c) | |
2173 | #define isASCII_uni(c) isASCII_uvchr(c) | |
2174 | #define isBLANK_uni(c) isBLANK_uvchr(c) | |
2175 | #define isCNTRL_uni(c) isCNTRL_uvchr(c) | |
2176 | #define isDIGIT_uni(c) isDIGIT_uvchr(c) | |
2177 | #define isGRAPH_uni(c) isGRAPH_uvchr(c) | |
2178 | #define isIDCONT_uni(c) isIDCONT_uvchr(c) | |
2179 | #define isIDFIRST_uni(c) isIDFIRST_uvchr(c) | |
2180 | #define isLOWER_uni(c) isLOWER_uvchr(c) | |
2181 | #define isPRINT_uni(c) isPRINT_uvchr(c) | |
2182 | #define isPUNCT_uni(c) isPUNCT_uvchr(c) | |
2183 | #define isSPACE_uni(c) isSPACE_uvchr(c) | |
2184 | #define isPSXSPC_uni(c) isPSXSPC_uvchr(c) | |
2185 | #define isUPPER_uni(c) isUPPER_uvchr(c) | |
2186 | #define isVERTWS_uni(c) isVERTWS_uvchr(c) | |
2187 | #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c) | |
2188 | #define isXDIGIT_uni(c) isXDIGIT_uvchr(c) | |
2189 | #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l) | |
2190 | #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l) | |
2191 | #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l) | |
2192 | #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l) | |
a0ed51b3 | 2193 | |
4650c663 KW |
2194 | /* For internal core Perl use only: the base macros for defining macros like |
2195 | * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code | |
6eb62d23 | 2196 | * point, not just 0-255. Like generic_uvchr_, there are two versions, one for |
4650c663 | 2197 | * simple class definitions; the other for more complex. These are like |
6eb62d23 KW |
2198 | * generic_uvchr_, so see it for more info. */ |
2199 | #define generic_LC_uvchr_(latin1, above_latin1, c) \ | |
cd500f2f | 2200 | (c < 256 ? latin1(c) : above_latin1(c)) |
81d43abf | 2201 | #define generic_LC_invlist_uvchr_(latin1, classnum, c) \ |
cd500f2f KW |
2202 | (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c)) |
2203 | ||
91456fff | 2204 | #define isALPHA_LC_uvchr(c) generic_LC_invlist_uvchr_(isALPHA_LC, CC_ALPHA_, c) |
6eb62d23 | 2205 | #define isALPHANUMERIC_LC_uvchr(c) generic_LC_invlist_uvchr_(isALPHANUMERIC_LC, \ |
91456fff | 2206 | CC_ALPHANUMERIC_, c) |
b7d90381 | 2207 | #define isASCII_LC_uvchr(c) isASCII_LC(c) |
6eb62d23 | 2208 | #define isBLANK_LC_uvchr(c) generic_LC_uvchr_(isBLANK_LC, \ |
b7d90381 | 2209 | is_HORIZWS_cp_high, c) |
feeab5a9 | 2210 | #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0) |
91456fff KW |
2211 | #define isDIGIT_LC_uvchr(c) generic_LC_invlist_uvchr_(isDIGIT_LC, CC_DIGIT_, c) |
2212 | #define isGRAPH_LC_uvchr(c) generic_LC_invlist_uvchr_(isGRAPH_LC, CC_GRAPH_, c) | |
6eb62d23 | 2213 | #define isIDCONT_LC_uvchr(c) generic_LC_uvchr_(isIDCONT_LC, \ |
eba68aa0 | 2214 | _is_uni_perl_idcont, c) |
6eb62d23 | 2215 | #define isIDFIRST_LC_uvchr(c) generic_LC_uvchr_(isIDFIRST_LC, \ |
cd500f2f | 2216 | _is_uni_perl_idstart, c) |
91456fff KW |
2217 | #define isLOWER_LC_uvchr(c) generic_LC_invlist_uvchr_(isLOWER_LC, CC_LOWER_, c) |
2218 | #define isPRINT_LC_uvchr(c) generic_LC_invlist_uvchr_(isPRINT_LC, CC_PRINT_, c) | |
b7d90381 | 2219 | #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c) |
91456fff | 2220 | #define isPUNCT_LC_uvchr(c) generic_LC_invlist_uvchr_(isPUNCT_LC, CC_PUNCT_, c) |
6eb62d23 | 2221 | #define isSPACE_LC_uvchr(c) generic_LC_uvchr_(isSPACE_LC, \ |
509fb054 | 2222 | is_XPERLSPACE_cp_high, c) |
91456fff | 2223 | #define isUPPER_LC_uvchr(c) generic_LC_invlist_uvchr_(isUPPER_LC, CC_UPPER_, c) |
81d43abf | 2224 | #define isWORDCHAR_LC_uvchr(c) generic_LC_invlist_uvchr_(isWORDCHAR_LC, \ |
91456fff | 2225 | CC_WORDCHAR_, c) |
81d43abf | 2226 | #define isXDIGIT_LC_uvchr(c) generic_LC_uvchr_(isXDIGIT_LC, \ |
b7d90381 | 2227 | is_XDIGIT_cp_high, c) |
e712593e | 2228 | |
b7d90381 | 2229 | #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c)) |
aaa51d5e | 2230 | |
da8c1a98 KW |
2231 | /* The "_safe" macros make sure that we don't attempt to read beyond 'e', but |
2232 | * they don't otherwise go out of their way to look for malformed UTF-8. If | |
2233 | * they can return accurate results without knowing if the input is otherwise | |
2234 | * malformed, they do so. For example isASCII is accurate in spite of any | |
2235 | * non-length malformations because it looks only at a single byte. Likewise | |
2236 | * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8 | |
2237 | * variant ones return FALSE. But, if the input has to be well-formed in order | |
2238 | * for the results to be accurate, the macros will test and if malformed will | |
2239 | * call a routine to die | |
2240 | * | |
2241 | * Except for toke.c, the macros do assume that e > p, asserting that on | |
2242 | * DEBUGGING builds. Much code that calls these depends on this being true, | |
2243 | * for other reasons. toke.c is treated specially as using the regular | |
2244 | * assertion breaks it in many ways. All strings that these operate on there | |
2245 | * are supposed to have an extra NUL character at the end, so that *e = \0. A | |
2246 | * bunch of code in toke.c assumes that this is true, so the assertion allows | |
2247 | * for that */ | |
2248 | #ifdef PERL_IN_TOKE_C | |
2249 | # define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0')) | |
2250 | #else | |
2251 | # define _utf8_safe_assert(p,e) ((e) > (p)) | |
2252 | #endif | |
2253 | ||
6eb62d23 | 2254 | #define generic_utf8_safe_(classnum, p, e, above_latin1) \ |
c81b3562 KW |
2255 | ((! _utf8_safe_assert(p, e)) \ |
2256 | ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\ | |
2257 | : (UTF8_IS_INVARIANT(*(p))) \ | |
6eb62d23 | 2258 | ? generic_isCC_(*(p), classnum) \ |
da8c1a98 KW |
2259 | : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \ |
2260 | ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \ | |
6eb62d23 | 2261 | ? generic_isCC_(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \ |
da8c1a98 KW |
2262 | classnum) \ |
2263 | : (_force_out_malformed_utf8_message( \ | |
2264 | (U8 *) (p), (U8 *) (e), 0, 1), 0)) \ | |
2265 | : above_latin1)) | |
b7d90381 KW |
2266 | /* Like the above, but calls 'above_latin1(p)' to get the utf8 value. |
2267 | * 'above_latin1' can be a macro */ | |
6eb62d23 KW |
2268 | #define generic_func_utf8_safe_(classnum, above_latin1, p, e) \ |
2269 | generic_utf8_safe_(classnum, p, e, above_latin1(p, e)) | |
81d43abf | 2270 | #define generic_non_invlist_utf8_safe_(classnum, above_latin1, p, e) \ |
6eb62d23 | 2271 | generic_utf8_safe_(classnum, p, e, \ |
da8c1a98 KW |
2272 | (UNLIKELY((e) - (p) < UTF8SKIP(p)) \ |
2273 | ? (_force_out_malformed_utf8_message( \ | |
2274 | (U8 *) (p), (U8 *) (e), 0, 1), 0) \ | |
2275 | : above_latin1(p))) | |
2366ba44 KW |
2276 | /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an |
2277 | * 'above_latin1' parameter */ | |
81d43abf | 2278 | #define generic_invlist_utf8_safe_(classnum, p, e) \ |
6eb62d23 | 2279 | generic_utf8_safe_(classnum, p, e, _is_utf8_FOO(classnum, p, e)) |
922e8cb4 | 2280 | |
cc8ab7c0 | 2281 | /* Like the above, but should be used only when it is known that there are no |
ff7ecfc3 KW |
2282 | * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the |
2283 | * class is TRUE for. Hence it can skip the tests for this range. | |
2284 | * 'above_latin1' should include its arguments */ | |
6eb62d23 | 2285 | #define generic_utf8_safe_no_upper_latin1_(classnum, p, e, above_latin1) \ |
da8c1a98 | 2286 | (__ASSERT_(_utf8_safe_assert(p, e)) \ |
2d8dd9eb | 2287 | (isASCII(*(p))) \ |
6eb62d23 | 2288 | ? generic_isCC_(*(p), classnum) \ |
da8c1a98 KW |
2289 | : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \ |
2290 | ? 0 /* Note that doesn't check validity for latin1 */ \ | |
2291 | : above_latin1) | |
2292 | ||
84238efa | 2293 | |
059703b0 KW |
2294 | #define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e) |
2295 | #define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e) | |
2296 | #define isASCII_utf8(p, e) isASCII_utf8_safe(p, e) | |
2297 | #define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e) | |
2298 | #define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e) | |
2299 | #define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e) | |
2300 | #define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e) | |
2301 | #define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e) | |
2302 | #define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e) | |
2303 | #define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e) | |
2304 | #define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e) | |
2305 | #define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e) | |
2306 | #define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e) | |
2307 | #define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e) | |
2308 | #define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e) | |
2309 | #define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e) | |
2310 | #define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e) | |
2311 | #define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e) | |
e8fa43e2 | 2312 | |
91456fff | 2313 | #define isALPHA_utf8_safe(p, e) generic_invlist_utf8_safe_(CC_ALPHA_, p, e) |
da8c1a98 | 2314 | #define isALPHANUMERIC_utf8_safe(p, e) \ |
91456fff | 2315 | generic_invlist_utf8_safe_(CC_ALPHANUMERIC_, p, e) |
da8c1a98 KW |
2316 | #define isASCII_utf8_safe(p, e) \ |
2317 | /* Because ASCII is invariant under utf8, the non-utf8 macro \ | |
2318 | * works */ \ | |
2319 | (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p))) | |
2320 | #define isBLANK_utf8_safe(p, e) \ | |
91456fff | 2321 | generic_non_invlist_utf8_safe_(CC_BLANK_, is_HORIZWS_high, p, e) |
da8c1a98 | 2322 | |
e8fa43e2 KW |
2323 | #ifdef EBCDIC |
2324 | /* Because all controls are UTF-8 invariants in EBCDIC, we can use this | |
2325 | * more efficient macro instead of the more general one */ | |
da8c1a98 | 2326 | # define isCNTRL_utf8_safe(p, e) \ |
56d02b8c | 2327 | (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p))) |
e8fa43e2 | 2328 | #else |
91456fff | 2329 | # define isCNTRL_utf8_safe(p, e) generic_utf8_safe_(CC_CNTRL_, p, e, 0) |
e8fa43e2 KW |
2330 | #endif |
2331 | ||
da8c1a98 | 2332 | #define isDIGIT_utf8_safe(p, e) \ |
91456fff KW |
2333 | generic_utf8_safe_no_upper_latin1_(CC_DIGIT_, p, e, \ |
2334 | _is_utf8_FOO(CC_DIGIT_, p, e)) | |
2335 | #define isGRAPH_utf8_safe(p, e) generic_invlist_utf8_safe_(CC_GRAPH_, p, e) | |
2336 | #define isIDCONT_utf8_safe(p, e) generic_func_utf8_safe_(CC_WORDCHAR_, \ | |
dd1a3ba7 | 2337 | _is_utf8_perl_idcont, p, e) |
e5dcd934 | 2338 | |
c11ff943 KW |
2339 | /* To prevent S_scan_word in toke.c from hanging, we have to make sure that |
2340 | * IDFIRST is an alnum. See | |
8034715d | 2341 | * https://github.com/Perl/perl5/issues/10275 for more detail than you |
f91dcd13 KW |
2342 | * ever wanted to know about. (In the ASCII range, there isn't a difference.) |
2343 | * This used to be not the XID version, but we decided to go with the more | |
2344 | * modern Unicode definition */ | |
da8c1a98 | 2345 | #define isIDFIRST_utf8_safe(p, e) \ |
91456fff | 2346 | generic_func_utf8_safe_(CC_IDFIRST_, \ |
dd1a3ba7 | 2347 | _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e)) |
da8c1a98 | 2348 | |
91456fff KW |
2349 | #define isLOWER_utf8_safe(p, e) generic_invlist_utf8_safe_(CC_LOWER_, p, e) |
2350 | #define isPRINT_utf8_safe(p, e) generic_invlist_utf8_safe_(CC_PRINT_, p, e) | |
da8c1a98 | 2351 | #define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e) |
91456fff | 2352 | #define isPUNCT_utf8_safe(p, e) generic_invlist_utf8_safe_(CC_PUNCT_, p, e) |
da8c1a98 | 2353 | #define isSPACE_utf8_safe(p, e) \ |
91456fff KW |
2354 | generic_non_invlist_utf8_safe_(CC_SPACE_, is_XPERLSPACE_high, p, e) |
2355 | #define isUPPER_utf8_safe(p, e) generic_invlist_utf8_safe_(CC_UPPER_, p, e) | |
da8c1a98 | 2356 | #define isVERTWS_utf8_safe(p, e) \ |
91456fff | 2357 | generic_non_invlist_utf8_safe_(CC_VERTSPACE_, is_VERTWS_high, p, e) |
da8c1a98 | 2358 | #define isWORDCHAR_utf8_safe(p, e) \ |
91456fff | 2359 | generic_invlist_utf8_safe_(CC_WORDCHAR_, p, e) |
da8c1a98 | 2360 | #define isXDIGIT_utf8_safe(p, e) \ |
91456fff | 2361 | generic_utf8_safe_no_upper_latin1_(CC_XDIGIT_, p, e, \ |
da8c1a98 KW |
2362 | (UNLIKELY((e) - (p) < UTF8SKIP(p)) \ |
2363 | ? (_force_out_malformed_utf8_message( \ | |
2364 | (U8 *) (p), (U8 *) (e), 0, 1), 0) \ | |
2365 | : is_XDIGIT_high(p))) | |
a0ed51b3 | 2366 | |
059703b0 KW |
2367 | #define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l) |
2368 | #define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l) | |
2369 | #define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l) | |
2370 | #define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l) | |
2e8adce6 | 2371 | |
567b353c | 2372 | /* For internal core use only, subject to change */ |
059703b0 KW |
2373 | #define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f) |
2374 | #define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f) | |
2375 | #define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f) | |
2376 | #define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f) | |
a1a5ec35 KW |
2377 | |
2378 | #define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL) | |
2379 | #define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0) | |
2380 | #define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0) | |
2381 | #define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0) | |
567b353c | 2382 | |
059703b0 KW |
2383 | #define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e) |
2384 | #define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e) | |
2385 | #define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e) | |
2386 | #define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e) | |
2387 | #define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e) | |
2388 | #define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e) | |
2389 | #define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e) | |
2390 | #define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e) | |
2391 | #define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e) | |
2392 | #define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e) | |
2393 | #define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e) | |
2394 | #define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e) | |
2395 | #define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e) | |
2396 | #define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e) | |
2397 | #define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e) | |
2398 | #define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e) | |
2399 | #define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e) | |
34aeb2e9 | 2400 | |
da8c1a98 | 2401 | /* For internal core Perl use only: the base macros for defining macros like |
6eb62d23 | 2402 | * isALPHA_LC_utf8_safe. These are like generic_utf8_, but if the first code |
da8c1a98 KW |
2403 | * point in 'p' is within the 0-255 range, it uses locale rules from the |
2404 | * passed-in 'macro' parameter */ | |
6eb62d23 | 2405 | #define generic_LC_utf8_safe_(macro, p, e, above_latin1) \ |
da8c1a98 KW |
2406 | (__ASSERT_(_utf8_safe_assert(p, e)) \ |
2407 | (UTF8_IS_INVARIANT(*(p))) \ | |
2408 | ? macro(*(p)) \ | |
2409 | : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \ | |
2410 | ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \ | |
2411 | ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \ | |
2412 | : (_force_out_malformed_utf8_message( \ | |
2413 | (U8 *) (p), (U8 *) (e), 0, 1), 0)) \ | |
2414 | : above_latin1)) | |
2415 | ||
6eb62d23 KW |
2416 | #define generic_LC_invlist_utf8_safe_(macro, classnum, p, e) \ |
2417 | generic_LC_utf8_safe_(macro, p, e, \ | |
2366ba44 | 2418 | _is_utf8_FOO(classnum, p, e)) |
da8c1a98 | 2419 | |
6eb62d23 KW |
2420 | #define generic_LC_func_utf8_safe_(macro, above_latin1, p, e) \ |
2421 | generic_LC_utf8_safe_(macro, p, e, above_latin1(p, e)) | |
da8c1a98 | 2422 | |
6eb62d23 KW |
2423 | #define generic_LC_non_invlist_utf8_safe_(classnum, above_latin1, p, e) \ |
2424 | generic_LC_utf8_safe_(classnum, p, e, \ | |
da8c1a98 KW |
2425 | (UNLIKELY((e) - (p) < UTF8SKIP(p)) \ |
2426 | ? (_force_out_malformed_utf8_message( \ | |
2427 | (U8 *) (p), (U8 *) (e), 0, 1), 0) \ | |
2428 | : above_latin1(p))) | |
2429 | ||
2430 | #define isALPHANUMERIC_LC_utf8_safe(p, e) \ | |
81d43abf | 2431 | generic_LC_invlist_utf8_safe_(isALPHANUMERIC_LC, \ |
91456fff | 2432 | CC_ALPHANUMERIC_, p, e) |
da8c1a98 | 2433 | #define isALPHA_LC_utf8_safe(p, e) \ |
91456fff | 2434 | generic_LC_invlist_utf8_safe_(isALPHA_LC, CC_ALPHA_, p, e) |
da8c1a98 KW |
2435 | #define isASCII_LC_utf8_safe(p, e) \ |
2436 | (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p))) | |
2437 | #define isBLANK_LC_utf8_safe(p, e) \ | |
6eb62d23 | 2438 | generic_LC_non_invlist_utf8_safe_(isBLANK_LC, is_HORIZWS_high, p, e) |
da8c1a98 | 2439 | #define isCNTRL_LC_utf8_safe(p, e) \ |
6eb62d23 | 2440 | generic_LC_utf8_safe_(isCNTRL_LC, p, e, 0) |
da8c1a98 | 2441 | #define isDIGIT_LC_utf8_safe(p, e) \ |
91456fff | 2442 | generic_LC_invlist_utf8_safe_(isDIGIT_LC, CC_DIGIT_, p, e) |
da8c1a98 | 2443 | #define isGRAPH_LC_utf8_safe(p, e) \ |
91456fff | 2444 | generic_LC_invlist_utf8_safe_(isGRAPH_LC, CC_GRAPH_, p, e) |
da8c1a98 | 2445 | #define isIDCONT_LC_utf8_safe(p, e) \ |
6eb62d23 | 2446 | generic_LC_func_utf8_safe_(isIDCONT_LC, \ |
dd1a3ba7 | 2447 | _is_utf8_perl_idcont, p, e) |
da8c1a98 | 2448 | #define isIDFIRST_LC_utf8_safe(p, e) \ |
6eb62d23 | 2449 | generic_LC_func_utf8_safe_(isIDFIRST_LC, \ |
dd1a3ba7 | 2450 | _is_utf8_perl_idstart, p, e) |
da8c1a98 | 2451 | #define isLOWER_LC_utf8_safe(p, e) \ |
91456fff | 2452 | generic_LC_invlist_utf8_safe_(isLOWER_LC, CC_LOWER_, p, e) |
da8c1a98 | 2453 | #define isPRINT_LC_utf8_safe(p, e) \ |
91456fff | 2454 | generic_LC_invlist_utf8_safe_(isPRINT_LC, CC_PRINT_, p, e) |
da8c1a98 KW |
2455 | #define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e) |
2456 | #define isPUNCT_LC_utf8_safe(p, e) \ | |
91456fff | 2457 | generic_LC_invlist_utf8_safe_(isPUNCT_LC, CC_PUNCT_, p, e) |
da8c1a98 | 2458 | #define isSPACE_LC_utf8_safe(p, e) \ |
6eb62d23 | 2459 | generic_LC_non_invlist_utf8_safe_(isSPACE_LC, is_XPERLSPACE_high, p, e) |
da8c1a98 | 2460 | #define isUPPER_LC_utf8_safe(p, e) \ |
91456fff | 2461 | generic_LC_invlist_utf8_safe_(isUPPER_LC, CC_UPPER_, p, e) |
da8c1a98 | 2462 | #define isWORDCHAR_LC_utf8_safe(p, e) \ |
91456fff | 2463 | generic_LC_invlist_utf8_safe_(isWORDCHAR_LC, CC_WORDCHAR_, p, e) |
da8c1a98 | 2464 | #define isXDIGIT_LC_utf8_safe(p, e) \ |
6eb62d23 | 2465 | generic_LC_non_invlist_utf8_safe_(isXDIGIT_LC, is_XDIGIT_high, p, e) |
aaa51d5e | 2466 | |
fbc19f27 KW |
2467 | /* Macros for backwards compatibility and for completeness when the ASCII and |
2468 | * Latin1 values are identical */ | |
b7d90381 KW |
2469 | #define isALPHAU(c) isALPHA_L1(c) |
2470 | #define isDIGIT_L1(c) isDIGIT_A(c) | |
2471 | #define isOCTAL(c) isOCTAL_A(c) | |
2472 | #define isOCTAL_L1(c) isOCTAL_A(c) | |
2473 | #define isXDIGIT_L1(c) isXDIGIT_A(c) | |
2474 | #define isALNUM(c) isWORDCHAR(c) | |
a377c856 | 2475 | #define isALNUM_A(c) isALNUM(c) |
b7d90381 KW |
2476 | #define isALNUMU(c) isWORDCHAR_L1(c) |
2477 | #define isALNUM_LC(c) isWORDCHAR_LC(c) | |
2478 | #define isALNUM_uni(c) isWORDCHAR_uni(c) | |
2e28f0b9 | 2479 | #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c) |
059703b0 | 2480 | #define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e) |
4c1d9526 | 2481 | #define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e) |
059703b0 | 2482 | #define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e) |
4c1d9526 | 2483 | #define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e) |
b7d90381 KW |
2484 | #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */ |
2485 | #define isALNUMC_L1(c) isALPHANUMERIC_L1(c) | |
2486 | #define isALNUMC(c) isALPHANUMERIC(c) | |
2487 | #define isALNUMC_LC(c) isALPHANUMERIC_LC(c) | |
2488 | #define isALNUMC_uni(c) isALPHANUMERIC_uni(c) | |
15861f94 | 2489 | #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c) |
059703b0 | 2490 | #define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e) |
4c1d9526 KW |
2491 | #define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e) |
2492 | #define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e) | |
fbc19f27 | 2493 | |
2bd1cbf6 KW |
2494 | /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII, |
2495 | * except that they don't necessarily mean the same characters, e.g. CTRL-D is | |
2496 | * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC. | |
2497 | * '?' is special-cased on EBCDIC to APC, which is the control there that is | |
2498 | * the outlier from the block that contains the other controls, just like | |
2499 | * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0 | |
2500 | * block. If it weren't special cased, it would yield a non-control. | |
88794300 KW |
2501 | * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D, |
2502 | * etc. */ | |
2bd1cbf6 | 2503 | #ifndef EBCDIC |
75763b3a | 2504 | # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64) |
2bd1cbf6 | 2505 | #else |
75763b3a KW |
2506 | # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ |
2507 | ((isPRINT_A(c)) \ | |
2508 | ? (UNLIKELY((c) == '?') \ | |
2509 | ? QUESTION_MARK_CTRL \ | |
2510 | : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \ | |
2511 | : (UNLIKELY((c) == QUESTION_MARK_CTRL) \ | |
2512 | ? '?' \ | |
2513 | : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64))))) | |
2bd1cbf6 | 2514 | #endif |
bbce6d69 | 2515 | |
837781cc KW |
2516 | /* |
2517 | =for apidoc Ay||line_t | |
2518 | The typedef to use to declare variables that are to hold line numbers. | |
2519 | ||
2520 | =cut | |
2521 | ||
2522 | Line numbers are unsigned, 32 bits. | |
2523 | */ | |
dea28490 | 2524 | typedef U32 line_t; |
ee178617 | 2525 | #define LINE_Tf U32uf |
e5dcd934 | 2526 | #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */ |
378cc40b | 2527 | |
91152fc1 DG |
2528 | /* Helpful alias for version prescan */ |
2529 | #define is_LAX_VERSION(a,b) \ | |
1604cfb0 | 2530 | (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL)) |
91152fc1 DG |
2531 | |
2532 | #define is_STRICT_VERSION(a,b) \ | |
1604cfb0 | 2533 | (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL)) |
91152fc1 DG |
2534 | |
2535 | #define BADVERSION(a,b,c) \ | |
1604cfb0 MS |
2536 | if (b) { \ |
2537 | *b = c; \ | |
2538 | } \ | |
2539 | return a; | |
8c52afec | 2540 | |
1ce77b7d KW |
2541 | /* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or |
2542 | * a-f) to its numeric value without using any branches. The input is | |
2543 | * validated only by an assert() in DEBUGGING builds. | |
2544 | * | |
2545 | * It works by right shifting and isolating the bit that is 0 for the digits, | |
2546 | * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones | |
2547 | * position, and then to the eights position. Both are added together to form | |
2548 | * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the | |
2549 | * final four bits of the input to form the correct value. */ | |
2550 | #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \ | |
2551 | ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \ | |
2552 | + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \ | |
2553 | + ((c) & 0xF)) /* 0-9 if input valid hex digit */ | |
2554 | ||
2555 | /* The argument is a string pointer, which is advanced. */ | |
2556 | #define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1))) | |
95a59cab | 2557 | |
cb27eebd KW |
2558 | /* Converts a character known to represent an octal digit (0-7) to its numeric |
2559 | * value. The input is validated only by an assert() in DEBUGGING builds. In | |
2560 | * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */ | |
2561 | #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c))) | |
2562 | ||
305b8651 | 2563 | /* Efficiently returns a boolean as to if two native characters are equivalent |
f1460a66 | 2564 | * case-insensitively. At least one of the characters must be one of [A-Za-z]; |
305b8651 KW |
2565 | * the ALPHA in the name is to remind you of that. This is asserted() in |
2566 | * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro | |
2567 | * works (on valid input) for both non- and UTF-8-encoded bytes. | |
2568 | * | |
2569 | * When one of the inputs is a compile-time constant and gets folded by the | |
2570 | * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII | |
2571 | * machines, 'A' and 'a' differ by a single bit; the same with the upper and | |
2572 | * lower case of all other ASCII-range alphabetics. On ASCII platforms, they | |
96ca48da KW |
2573 | * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an |
2574 | * exclusive 'or' to find that bit and then inverts it to form a mask, with | |
2575 | * just a single 0, in the bit position where the upper- and lowercase differ. | |
2576 | * */ | |
305b8651 KW |
2577 | #define isALPHA_FOLD_EQ(c1, c2) \ |
2578 | (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \ | |
2579 | ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a'))) | |
2580 | #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2))) | |
2581 | ||
8e84507e | 2582 | /* |
3f620621 | 2583 | =for apidoc_section $memory |
ccfc67b7 | 2584 | |
a02a5408 | 2585 | =for apidoc Am|void|Newx|void* ptr|int nitems|type |
25a3e84c | 2586 | =for apidoc_item |void*|safemalloc|size_t size |
c372fad0 | 2587 | |
954c1994 GS |
2588 | The XSUB-writer's interface to the C C<malloc> function. |
2589 | ||
596f7718 | 2590 | Memory obtained by this should B<ONLY> be freed with L</"Safefree">. |
0d7b2759 | 2591 | |
c5008215 JC |
2592 | In 5.9.3, Newx() and friends replace the older New() API, and drops |
2593 | the first parameter, I<x>, a debug aid which allowed callers to identify | |
37b8b4c9 | 2594 | themselves. This aid has been superseded by a new build option, |
d10b4965 | 2595 | PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still |
c5008215 JC |
2596 | there for use in XS modules supporting older perls. |
2597 | ||
a02a5408 | 2598 | =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast |
954c1994 | 2599 | The XSUB-writer's interface to the C C<malloc> function, with |
fbe13c60 | 2600 | cast. See also C<L</Newx>>. |
954c1994 | 2601 | |
596f7718 | 2602 | Memory obtained by this should B<ONLY> be freed with L</"Safefree">. |
0d7b2759 | 2603 | |
a02a5408 | 2604 | =for apidoc Am|void|Newxz|void* ptr|int nitems|type |
c372fad0 TK |
2605 | =for apidoc_item |void*|safecalloc|size_t nitems|size_t item_size |
2606 | ||
954c1994 | 2607 | The XSUB-writer's interface to the C C<malloc> function. The allocated |
fbe13c60 | 2608 | memory is zeroed with C<memzero>. See also C<L</Newx>>. |
a02a5408 | 2609 | |
596f7718 | 2610 | Memory obtained by this should B<ONLY> be freed with L</"Safefree">. |
0d7b2759 | 2611 | |
954c1994 | 2612 | =for apidoc Am|void|Renew|void* ptr|int nitems|type |
c372fad0 TK |
2613 | =for apidoc_item |void*|saferealloc|void *ptr|size_t size |
2614 | ||
954c1994 GS |
2615 | The XSUB-writer's interface to the C C<realloc> function. |
2616 | ||
596f7718 | 2617 | Memory obtained by this should B<ONLY> be freed with L</"Safefree">. |
0d7b2759 | 2618 | |
954c1994 GS |
2619 | =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast |
2620 | The XSUB-writer's interface to the C C<realloc> function, with | |
2621 | cast. | |
2622 | ||
596f7718 | 2623 | Memory obtained by this should B<ONLY> be freed with L</"Safefree">. |
0d7b2759 | 2624 | |
49b8b560 | 2625 | =for apidoc Am|void|Safefree|void* ptr |
954c1994 GS |
2626 | The XSUB-writer's interface to the C C<free> function. |
2627 | ||
596f7718 | 2628 | This should B<ONLY> be used on memory obtained using L</"Newx"> and friends. |
0d7b2759 | 2629 | |
3f620621 | 2630 | =for apidoc_section $string |
75b94e77 KW |
2631 | =for apidoc Am|void |Move |void* src|void* dest|int nitems|type |
2632 | =for apidoc_item |void *|MoveD|void* src|void* dest|int nitems|type | |
954c1994 | 2633 | The XSUB-writer's interface to the C C<memmove> function. The C<src> is the |
926bb54c | 2634 | source, C<dest> is the destination, C<nitems> is the number of items, and |
fbe13c60 | 2635 | C<type> is the type. Can do overlapping moves. See also C<L</Copy>>. |
954c1994 | 2636 | |
75b94e77 | 2637 | C<MoveD> is like C<Move> but returns C<dest>. Useful |
72d33970 | 2638 | for encouraging compilers to tail-call |
e90e2364 NC |
2639 | optimise. |
2640 | ||
75b94e77 KW |
2641 | =for apidoc Am|void |Copy |void* src|void* dest|int nitems|type |
2642 | =for apidoc_item |void *|CopyD|void* src|void* dest|int nitems|type | |
954c1994 | 2643 | The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the |
926bb54c | 2644 | source, C<dest> is the destination, C<nitems> is the number of items, and |
fbe13c60 | 2645 | C<type> is the type. May fail on overlapping copies. See also C<L</Move>>. |
954c1994 | 2646 | |
75b94e77 | 2647 | C<CopyD> is like C<Copy> but returns C<dest>. Useful |
72d33970 | 2648 | for encouraging compilers to tail-call |
e90e2364 NC |
2649 | optimise. |
2650 | ||
75b94e77 KW |
2651 | =for apidoc Am|void |Zero |void* dest|int nitems|type |
2652 | =for apidoc_item |void *|ZeroD|void* dest|int nitems|type | |
954c1994 GS |
2653 | |
2654 | The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the | |
2655 | destination, C<nitems> is the number of items, and C<type> is the type. | |
2656 | ||
75b94e77 | 2657 | C<ZeroD> is like C<Zero> but returns C<dest>. Useful |
72d33970 | 2658 | for encouraging compilers to tail-call |
e90e2364 NC |
2659 | optimise. |
2660 | ||
3f620621 | 2661 | =for apidoc_section $utility |
e538104b | 2662 | =for apidoc Amu|void|StructCopy|type *src|type *dest|type |
4375e838 | 2663 | This is an architecture-independent macro to copy one structure to another. |
954c1994 | 2664 | |
7e337ee0 JH |
2665 | =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte |
2666 | ||
2667 | Fill up memory with a byte pattern (a byte repeated over and over | |
2668 | again) that hopefully catches attempts to access uninitialized memory. | |
2669 | ||
2670 | =for apidoc Am|void|PoisonNew|void* dest|int nitems|type | |
2671 | ||
2672 | PoisonWith(0xAB) for catching access to allocated but uninitialized memory. | |
2673 | ||
1c12ffb4 | 2674 | =for apidoc Am|void|PoisonFree|void* dest|int nitems|type |
7e337ee0 JH |
2675 | |
2676 | PoisonWith(0xEF) for catching access to freed memory. | |
2677 | ||
9965345d JH |
2678 | =for apidoc Am|void|Poison|void* dest|int nitems|type |
2679 | ||
7e337ee0 | 2680 | PoisonWith(0xEF) for catching access to freed memory. |
9965345d JH |
2681 | |
2682 | =cut */ | |
954c1994 | 2683 | |
561b68a9 SH |
2684 | /* Maintained for backwards-compatibility only. Use newSV() instead. */ |
2685 | #ifndef PERL_CORE | |
ff06c60c | 2686 | #define NEWSV(x,len) newSV(len) |
561b68a9 | 2687 | #endif |
ff06c60c | 2688 | |
b7112dce | 2689 | #define MEM_SIZE_MAX ((MEM_SIZE)-1) |
19a94d75 | 2690 | |
a500027b | 2691 | #define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1)) |
e6bdf523 | 2692 | |
27d5b266 | 2693 | #ifdef PERL_MALLOC_WRAP |
e6bdf523 DM |
2694 | |
2695 | /* This expression will be constant-folded at compile time. It checks | |
2696 | * whether or not the type of the count n is so small (e.g. U8 or U16, or | |
2697 | * U32 on 64-bit systems) that there's no way a wrap-around could occur. | |
2698 | * As well as avoiding the need for a run-time check in some cases, it's | |
2699 | * designed to avoid compiler warnings like: | |
2700 | * comparison is always false due to limited range of data type | |
73e8ff00 DM |
2701 | * It's mathematically equivalent to |
2702 | * max(n) * sizeof(t) > MEM_SIZE_MAX | |
e6bdf523 DM |
2703 | */ |
2704 | ||
2705 | # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \ | |
445198b9 LM |
2706 | ( sizeof(MEM_SIZE) < sizeof(n) \ |
2707 | || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n)))) | |
e6bdf523 | 2708 | |
88f9f128 | 2709 | /* This is written in a slightly odd way to avoid various spurious |
d98e5cde DM |
2710 | * compiler warnings. We *want* to write the expression as |
2711 | * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C) | |
2712 | * (for some compile-time constant C), but even when the LHS | |
2713 | * constant-folds to false at compile-time, g++ insists on emitting | |
2714 | * warnings about the RHS (e.g. "comparison is always false"), so instead | |
2715 | * we write it as | |
e6bdf523 | 2716 | * |
d98e5cde | 2717 | * (cond ? n : X) > C |
88f9f128 | 2718 | * |
d98e5cde DM |
2719 | * where X is a constant with X > C always false. Choosing a value for X |
2720 | * is tricky. If 0, some compilers will complain about 0 > C always being | |
2721 | * false; if 1, Coverity complains when n happens to be the constant value | |
2722 | * '1', that cond ? 1 : 1 has the same value on both branches; so use C | |
2723 | * for X and hope that nothing else whines. | |
e6bdf523 DM |
2724 | */ |
2725 | ||
2726 | # define _MEM_WRAP_WILL_WRAP(n,t) \ | |
88f9f128 DM |
2727 | ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \ |
2728 | MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t)) | |
e6bdf523 DM |
2729 | |
2730 | # define MEM_WRAP_CHECK(n,t) \ | |
1604cfb0 | 2731 | (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \ |
e6bdf523 DM |
2732 | && (croak_memory_wrap(),0)) |
2733 | ||
2734 | # define MEM_WRAP_CHECK_1(n,t,a) \ | |
1604cfb0 MS |
2735 | (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \ |
2736 | && (Perl_croak_nocontext("%s",(a)),0)) | |
e6bdf523 | 2737 | |
814eedc8 DD |
2738 | /* "a" arg must be a string literal */ |
2739 | # define MEM_WRAP_CHECK_s(n,t,a) \ | |
ca0572d7 KW |
2740 | ( (void) (UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \ |
2741 | && (Perl_croak_nocontext(ASSERT_IS_LITERAL(a)), 0))) | |
814eedc8 | 2742 | |
0622ec7a | 2743 | # define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t), |
27d5b266 | 2744 | |
0622ec7a | 2745 | # 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 |
2746 | #else |
2747 | ||
0622ec7a KW |
2748 | # define MEM_WRAP_CHECK(n,t) |
2749 | # define MEM_WRAP_CHECK_1(n,t,a) | |
2750 | # define MEM_WRAP_CHECK_s(n,t,a) | |
2751 | # define MEM_WRAP_CHECK_(n,t) | |
8b44ba4c | 2752 | |
0622ec7a | 2753 | # define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n) |
27d5b266 | 2754 | |
1936d2a7 | 2755 | #endif |
8b44ba4c | 2756 | |
fe4f188c | 2757 | #ifdef PERL_MEM_LOG |
46c6c7e2 | 2758 | /* |
9f653bb5 | 2759 | * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s |
46c6c7e2 JH |
2760 | * go through functions, which are handy for debugging breakpoints, but |
2761 | * which more importantly get the immediate calling environment (file and | |
e352bcff JH |
2762 | * line number, and C function name if available) passed in. This info can |
2763 | * then be used for logging the calls, for which one gets a sample | |
73d1d973 | 2764 | * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined. |
3609ea0d | 2765 | * |
46c6c7e2 | 2766 | * Known problems: |
94e892a6 | 2767 | * - not all memory allocs get logged, only those |
46c6c7e2 | 2768 | * that go through Newx() and derivatives (while all |
94e892a6 | 2769 | * Safefrees do get logged) |
46c6c7e2 JH |
2770 | * - __FILE__ and __LINE__ do not work everywhere |
2771 | * - __func__ or __FUNCTION__ even less so | |
2772 | * - I think more goes on after the perlio frees but | |
2773 | * the thing is that STDERR gets closed (as do all | |
2774 | * the file descriptors) | |
2775 | * - no deeper calling stack than the caller of the Newx() | |
2776 | * or the kind, but do I look like a C reflection/introspection | |
2777 | * utility to you? | |
2778 | * - the function prototypes for the logging functions | |
2779 | * probably should maybe be somewhere else than handy.h | |
2780 | * - one could consider inlining (macrofying) the logging | |
2781 | * for speed, but I am too lazy | |
2782 | * - one could imagine recording the allocations in a hash, | |
2783 | * (keyed by the allocation address?), and maintain that | |
2784 | * through reallocs and frees, but how to do that without | |
2785 | * any News() happening...? | |
73d1d973 | 2786 | * - lots of -Ddefines to get useful/controllable output |
b953482e | 2787 | * - lots of ENV reads |
46c6c7e2 JH |
2788 | */ |
2789 | ||
0b0ab801 | 2790 | # ifdef PERL_CORE |
73d1d973 | 2791 | # ifndef PERL_MEM_LOG_NOIMPL |
0b0ab801 MHM |
2792 | enum mem_log_type { |
2793 | MLT_ALLOC, | |
2794 | MLT_REALLOC, | |
d7a2c63c MHM |
2795 | MLT_FREE, |
2796 | MLT_NEW_SV, | |
2797 | MLT_DEL_SV | |
0b0ab801 MHM |
2798 | }; |
2799 | # endif | |
2800 | # endif | |
2801 | ||
fe4f188c JH |
2802 | #endif |
2803 | ||
2804 | #ifdef PERL_MEM_LOG | |
d1401ee9 MHM |
2805 | #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__) |
2806 | #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__) | |
46c6c7e2 | 2807 | #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__) |
fe4f188c JH |
2808 | #endif |
2809 | ||
2810 | #ifndef MEM_LOG_ALLOC | |
2811 | #define MEM_LOG_ALLOC(n,t,a) (a) | |
2812 | #endif | |
2813 | #ifndef MEM_LOG_REALLOC | |
2814 | #define MEM_LOG_REALLOC(n,t,v,a) (a) | |
2815 | #endif | |
2816 | #ifndef MEM_LOG_FREE | |
2817 | #define MEM_LOG_FREE(a) (a) | |
2818 | #endif | |
2819 | ||
d1401ee9 MHM |
2820 | #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t)))))) |
2821 | #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t)))))) | |
2822 | #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t))))) | |
a6f6820f NC |
2823 | |
2824 | #ifndef PERL_CORE | |
a02a5408 JC |
2825 | /* pre 5.9.x compatibility */ |
2826 | #define New(x,v,n,t) Newx(v,n,t) | |
2827 | #define Newc(x,v,n,t,c) Newxc(v,n,t,c) | |
4541904d | 2828 | #define Newz(x,v,n,t) Newxz(v,n,t) |
a6f6820f | 2829 | #endif |
a02a5408 | 2830 | |
ff68c719 | 2831 | #define Renew(v,n,t) \ |
1604cfb0 | 2832 | (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t)))))) |
ff68c719 | 2833 | #define Renewc(v,n,t,c) \ |
1604cfb0 | 2834 | (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t)))))) |
94010e71 NC |
2835 | |
2836 | #ifdef PERL_POISON | |
2837 | #define Safefree(d) \ | |
06c0cc96 | 2838 | ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0) |
94010e71 | 2839 | #else |
fe4f188c | 2840 | #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d))) |
94010e71 | 2841 | #endif |
55497cff | 2842 | |
dbb57106 YO |
2843 | /* assert that a valid ptr has been supplied - use this instead of assert(ptr) * |
2844 | * as it handles cases like constant string arguments without throwing warnings * | |
2845 | * the cast is required, as is the inequality check, to avoid warnings */ | |
45908e4d | 2846 | #define perl_assert_ptr(p) assert( ((void*)(p)) != 0 ) |
55497cff | 2847 | |
45908e4d YO |
2848 | |
2849 | #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))) | |
2850 | #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))) | |
2851 | #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t))) | |
2852 | ||
bdd1531d | 2853 | /* Like above, but returns a pointer to 'd' */ |
45908e4d YO |
2854 | #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))) |
2855 | #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 | 2856 | #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t))) |
e90e2364 | 2857 | |
7e337ee0 JH |
2858 | #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t))) |
2859 | #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB) | |
2860 | #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF) | |
2861 | #define Poison(d,n,t) PoisonFree(d,n,t) | |
27d5b266 | 2862 | |
caa674f3 DD |
2863 | #ifdef PERL_POISON |
2864 | # define PERL_POISON_EXPR(x) x | |
2865 | #else | |
2866 | # define PERL_POISON_EXPR(x) | |
2867 | #endif | |
2868 | ||
be319be3 | 2869 | /* Shallow copy */ |
ff68c719 | 2870 | #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s))) |
2cc61e15 | 2871 | |
1b7e2294 | 2872 | /* |
3f620621 | 2873 | =for apidoc_section $utility |
1b7e2294 KW |
2874 | |
2875 | =for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a | |
2876 | ||
2877 | Returns the number of elements in the input C array (so you want your | |
2878 | zero-based indices to be less than but not equal to). | |
2879 | ||
2880 | =for apidoc Am|void *|C_ARRAY_END|void *a | |
2881 | ||
2882 | Returns a pointer to one element past the final element of the input C array. | |
2883 | ||
2884 | =cut | |
2885 | ||
2886 | C_ARRAY_END is one past the last: half-open/half-closed range, not | |
2887 | last-inclusive range. | |
2888 | */ | |
622913ab | 2889 | #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0])) |
c3caa5c3 | 2890 | #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a)) |
622913ab | 2891 | |
ff00d73b KW |
2892 | #if defined(PERL_CORE) || defined(PERL_EXT_RE_BUILD) |
2893 | /* strlen() of a literal string constant. Restricting this to core, in part | |
2894 | * because it can generate compiler warnings about comparing unlike signs */ | |
2895 | # define STRLENs(s) (sizeof("" s "") - 1) | |
2896 | #endif | |
2897 | ||
2cc61e15 DD |
2898 | #ifdef NEED_VA_COPY |
2899 | # ifdef va_copy | |
2900 | # define Perl_va_copy(s, d) va_copy(d, s) | |
07798b17 AC |
2901 | # elif defined(__va_copy) |
2902 | # define Perl_va_copy(s, d) __va_copy(d, s) | |
2cc61e15 | 2903 | # else |
07798b17 | 2904 | # define Perl_va_copy(s, d) Copy(s, d, 1, va_list) |
2cc61e15 DD |
2905 | # endif |
2906 | #endif | |
2907 | ||
472d47bc SB |
2908 | /* convenience debug macros */ |
2909 | #ifdef USE_ITHREADS | |
2910 | #define pTHX_FORMAT "Perl interpreter: 0x%p" | |
2911 | #define pTHX__FORMAT ", Perl interpreter: 0x%p" | |
f54cb97a AL |
2912 | #define pTHX_VALUE_ (void *)my_perl, |
2913 | #define pTHX_VALUE (void *)my_perl | |
2914 | #define pTHX__VALUE_ ,(void *)my_perl, | |
2915 | #define pTHX__VALUE ,(void *)my_perl | |
472d47bc | 2916 | #else |
3609ea0d | 2917 | #define pTHX_FORMAT |
472d47bc | 2918 | #define pTHX__FORMAT |
3609ea0d | 2919 | #define pTHX_VALUE_ |
472d47bc | 2920 | #define pTHX_VALUE |
3609ea0d | 2921 | #define pTHX__VALUE_ |
472d47bc SB |
2922 | #define pTHX__VALUE |
2923 | #endif /* USE_ITHREADS */ | |
3609ea0d | 2924 | |
2acdbac1 NC |
2925 | /* Perl_deprecate was not part of the public API, and did not have a deprecate() |
2926 | shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor | |
2927 | CPAN::Unpack show any users outside the core. */ | |
2928 | #ifdef PERL_CORE | |
dc6e8de0 A |
2929 | # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \ |
2930 | "Use of " s " is deprecated") | |
c9680906 A |
2931 | # define deprecate_disappears_in(when,message) \ |
2932 | Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \ | |
81a93f7e | 2933 | message " is deprecated, and will disappear in Perl " when) |
ac641426 A |
2934 | # define deprecate_fatal_in(when,message) \ |
2935 | Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \ | |
81a93f7e | 2936 | message " is deprecated, and will become fatal in Perl " when) |
2acdbac1 NC |
2937 | #endif |
2938 | ||
dfff4baf BF |
2939 | /* Internal macros to deal with gids and uids */ |
2940 | #ifdef PERL_CORE | |
2941 | ||
2942 | # if Uid_t_size > IVSIZE | |
2943 | # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid)) | |
2944 | # define SvUID(sv) SvNV(sv) | |
07798b17 AC |
2945 | # elif Uid_t_sign <= 0 |
2946 | # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid)) | |
2947 | # define SvUID(sv) SvIV(sv) | |
dfff4baf | 2948 | # else |
07798b17 AC |
2949 | # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid)) |
2950 | # define SvUID(sv) SvUV(sv) | |
dfff4baf BF |
2951 | # endif /* Uid_t_size */ |
2952 | ||
2953 | # if Gid_t_size > IVSIZE | |
2954 | # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid)) | |
2955 | # define SvGID(sv) SvNV(sv) | |
07798b17 AC |
2956 | # elif Gid_t_sign <= 0 |
2957 | # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid)) | |
2958 | # define SvGID(sv) SvIV(sv) | |
dfff4baf | 2959 | # else |
07798b17 AC |
2960 | # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid)) |
2961 | # define SvGID(sv) SvUV(sv) | |
dfff4baf BF |
2962 | # endif /* Gid_t_size */ |
2963 | ||
2964 | #endif | |
2965 | ||
08da5deb YO |
2966 | /* These are simple Marsaglia XOR-SHIFT RNG's for 64 and 32 bits. These |
2967 | * RNG's are of reasonable quality, very fast, and have the interesting | |
2968 | * property that provided 'x' is non-zero they create a cycle of 2^32-1 | |
2969 | * or 2^64-1 "random" like numbers, with the exception of 0. Thus they | |
2970 | * are very useful when you want an integer to "dance" in a random way, | |
2971 | * but you also never want it to become 0 and thus false. | |
2972 | * | |
bf2a3dae YO |
2973 | * Obviously they leave x unchanged if it starts out as 0. |
2974 | * | |
2975 | * We have two variants just because that can be helpful in certain | |
2976 | * places. There is no advantage to either, they are equally bad as each | |
2977 | * other as far RNG's go. Sufficiently random for many purposes, but | |
2978 | * insufficiently random for serious use as they fail important tests in | |
2979 | * the Test01 BigCrush RNG test suite by L’Ecuyer and Simard. (Note | |
2980 | * that Drand48 also fails BigCrush). The main point is they produce | |
2981 | * different sequences and in places where we want some randomlike | |
2982 | * behavior they are cheap and easy. | |
2983 | * | |
2984 | * Marsaglia was one of the early researchers into RNG testing and wrote | |
2985 | * the Diehard RNG test suite, which after his death become the | |
2986 | * Dieharder RNG suite, and was generally supplanted by the Test01 suite | |
2987 | * by L'Ecruyer and associates. | |
2988 | * | |
2989 | * There are dozens of shift parameters that create a pseudo random ring | |
2990 | * of integers 1..2^N-1, if you need a different sequence just read the | |
2991 | * paper and select a set of parameters. In fact, simply reversing the | |
2992 | * shift order from L/R/L to R/L/R should result in another valid | |
2993 | * example, but read the paper before you do that. | |
2994 | * | |
2995 | * PDF of the original paper: | |
2996 | * https://www.jstatsoft.org/article/download/v008i14/916 | |
2997 | * Wikipedia: | |
2998 | * https://en.wikipedia.org/wiki/Xorshift | |
2999 | * Criticism: | |
3000 | * https://www.iro.umontreal.ca/~lecuyer/myftp/papers/xorshift.pdf | |
3001 | * Test01: | |
3002 | * http://simul.iro.umontreal.ca/testu01/tu01.html | |
3003 | * Diehard: | |
3004 | * https://en.wikipedia.org/wiki/Diehard_tests | |
3005 | * Dieharder: | |
3006 | * https://webhome.phy.duke.edu/~rgb/General/rand_rate/rand_rate.abs | |
3007 | * | |
3008 | */ | |
08da5deb | 3009 | |
bf2a3dae YO |
3010 | /* 32 bit version */ |
3011 | #define PERL_XORSHIFT32_A(x) \ | |
08da5deb | 3012 | STMT_START { \ |
bf2a3dae YO |
3013 | (x) ^= ((x) << 13); \ |
3014 | (x) ^= ((x) >> 17); \ | |
3015 | (x) ^= ((x) << 5); \ | |
3016 | } STMT_END | |
3017 | ||
3018 | /* 64 bit version */ | |
3019 | #define PERL_XORSHIFT64_A(x) \ | |
3020 | STMT_START { \ | |
3021 | (x) ^= ((x) << 13); \ | |
3022 | (x) ^= ((x) >> 7); \ | |
3023 | (x) ^= ((x) << 17); \ | |
08da5deb YO |
3024 | } STMT_END |
3025 | ||
3026 | /* 32 bit version */ | |
bf2a3dae YO |
3027 | #define PERL_XORSHIFT32_B(x) \ |
3028 | STMT_START { \ | |
3029 | (x) ^= ((x) << 5); \ | |
3030 | (x) ^= ((x) >> 27); \ | |
3031 | (x) ^= ((x) << 8); \ | |
3032 | } STMT_END | |
3033 | ||
3034 | /* 64 bit version - currently this is unused, | |
3035 | * it is provided here to complement the 32 bit _B | |
3036 | * variant which IS used. */ | |
3037 | #define PERL_XORSHIFT64_B(x) \ | |
08da5deb | 3038 | STMT_START { \ |
bf2a3dae YO |
3039 | (x) ^= ((x) << 15); \ |
3040 | (x) ^= ((x) >> 49); \ | |
3041 | (x) ^= ((x) << 26); \ | |
08da5deb YO |
3042 | } STMT_END |
3043 | ||
3044 | ||
6a5bc5ac | 3045 | #endif /* PERL_HANDY_H_ */ |
9d745869 | 3046 | |
e9a8c099 | 3047 | /* |
14d04a33 | 3048 | * ex: set ts=8 sts=4 sw=4 et: |
e9a8c099 | 3049 | */ |