This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regen/regcharclass.pl: Add ability to restrict platforms
[perl5.git] / utf8.h
CommitLineData
a0ed51b3
LW
1/* utf8.h
2 *
2eee27d7
SS
3 * Copyright (C) 2000, 2001, 2002, 2005, 2006, 2007, 2009,
4 * 2010, 2011 by Larry Wall and others
a0ed51b3
LW
5 *
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.
8 *
9 */
10
39e02b42 11/* Use UTF-8 as the default script encoding?
1e54db1a 12 * Turning this on will break scripts having non-UTF-8 binary
39e02b42
JH
13 * data (such as Latin-1) in string literals. */
14#ifdef USE_UTF8_SCRIPTS
15# define USE_UTF8_IN_NAMES (!IN_BYTES)
16#else
17# define USE_UTF8_IN_NAMES (PL_hints & HINT_UTF8)
18#endif
19
051a06d4
KW
20/* For to_utf8_fold_flags, q.v. */
21#define FOLD_FLAGS_LOCALE 0x1
22#define FOLD_FLAGS_FULL 0x2
a0270393 23#define FOLD_FLAGS_NOMIX_ASCII 0x4
051a06d4 24
83199d38
KW
25/* For _core_swash_init(), internal core use only */
26#define _CORE_SWASH_INIT_USER_DEFINED_PROPERTY 0x1
5d3d13d1 27#define _CORE_SWASH_INIT_RETURN_IF_UNDEF 0x2
87367d5f 28#define _CORE_SWASH_INIT_ACCEPT_INVLIST 0x4
83199d38 29
a0270393 30#define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
051a06d4
KW
31#define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, \
32 FOLD_FLAGS_FULL, NULL)
33#define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0, NULL)
34#define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0, NULL)
35#define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0, NULL)
36bb2ab6 36
fd7cb289
RGS
37/* Source backward compatibility. */
38#define uvuni_to_utf8(d, uv) uvuni_to_utf8_flags(d, uv, 0)
39#define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
40
eda9cac1
KW
41#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
42 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
a33c29bc 43#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
5e64d0fa 44#define FOLDEQ_UTF8_LOCALE (1 << 1)
18f762c3
KW
45#define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
46#define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
a33c29bc 47
e6226b18
KW
48/*
49=for apidoc ibcmp_utf8
50
51This is a synonym for (! foldEQ_utf8())
52
53=cut
54*/
55#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
56 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
57
1d72bdf6
NIS
58#ifdef EBCDIC
59/* The equivalent of these macros but implementing UTF-EBCDIC
60 are in the following header file:
61 */
62
63#include "utfebcdic.h"
fd7cb289 64
d06134e5 65#else /* ! EBCDIC */
73c4f7a1
GS
66START_EXTERN_C
67
a0ed51b3 68#ifdef DOINIT
6f06b55f 69EXTCONST unsigned char PL_utf8skip[] = {
b2635aa8
KW
70/* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
71/* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
72/* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
73/* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
74/* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
75/* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
76/* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
77/* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
78/* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
79/* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
80/* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
81/* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
82/* 0xC0 */ 2,2, /* overlong */
83/* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
84/* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */
85/* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */
86/* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */
87/* 0xFE */ 7,13, /* Perl extended (never was official UTF-8). Up to 72bit
88 allowed (64-bit + reserved). */
a0ed51b3
LW
89};
90#else
6f06b55f 91EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
92#endif
93
73c4f7a1 94END_EXTERN_C
7e2040f0 95
b96a92fb 96#include "regcharclass.h"
ec34087a
KW
97#include "unicode_constants.h"
98
1d72bdf6
NIS
99/* Native character to iso-8859-1 */
100#define NATIVE_TO_ASCII(ch) (ch)
101#define ASCII_TO_NATIVE(ch) (ch)
102/* Transform after encoding */
103#define NATIVE_TO_UTF(ch) (ch)
428921e5 104#define NATIVE_TO_I8(ch) NATIVE_TO_UTF(ch) /* a clearer synonym */
1d72bdf6 105#define UTF_TO_NATIVE(ch) (ch)
428921e5 106#define I8_TO_NATIVE(ch) UTF_TO_NATIVE(ch)
1d72bdf6
NIS
107/* Transforms in wide UV chars */
108#define UNI_TO_NATIVE(ch) (ch)
109#define NATIVE_TO_UNI(ch) (ch)
110/* Transforms in invariant space */
111#define NATIVE_TO_NEED(enc,ch) (ch)
112#define ASCII_TO_NEED(enc,ch) (ch)
d7578b48 113
d06134e5 114/* As there are no translations, avoid the function wrapper */
1754c1a1 115#define utf8n_to_uvchr utf8n_to_uvuni
6dd9dce9 116#define valid_utf8_to_uvchr valid_utf8_to_uvuni
1754c1a1 117#define uvchr_to_utf8 uvuni_to_utf8
2b9d42f0 118
877d9f0d 119/*
9041c2e3 120
8c007b5a 121 The following table is from Unicode 3.2.
877d9f0d
JH
122
123 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
124
375122d7 125 U+0000..U+007F 00..7F
e1b711da 126 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 127 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 128 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 129 U+D000..U+D7FF ED 80..9F 80..BF
37e2e78e 130 U+D800..U+DFFF +++++++ utf16 surrogates, not legal utf8 +++++++
375122d7 131 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 132 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH
133 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
134 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
b2635aa8
KW
135 Below are non-Unicode code points
136 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF
137 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF
138 U+200000: F8.. * 88..BF 80..BF 80..BF 80..BF
877d9f0d 139
e1b711da 140Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
141caused by legal UTF-8 avoiding non-shortest encodings: it is technically
142possible to UTF-8-encode a single code point in different ways, but that is
143explicitly forbidden, and the shortest possible encoding should always be used
144(and that is what Perl does).
8c007b5a 145
877d9f0d
JH
146 */
147
8c007b5a
JH
148/*
149 Another way to look at it, as bits:
150
b2635aa8 151 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
8c007b5a 152
b2635aa8
KW
153 0aaa aaaa 0aaa aaaa
154 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa
155 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa
156 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa
8c007b5a
JH
157
158As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 159leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
160encoded character.
161
65ab9279
TC
162Perl's extended UTF-8 means we can have start bytes up to FF.
163
8c007b5a
JH
164*/
165
c4d5f83a 166#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
0447e8df 167#define UTF8_IS_START(c) (((U8)c) >= 0xc2)
e021c6e6 168#define UTF8_IS_CONTINUATION(c) ((((U8)c) & 0xC0) == 0x80)
c512ce4f 169#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
0ae1fa71
CH
170
171/* Masking with 0xfe allows low bit to be 0 or 1; thus this matches 0xc[23] */
172#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)c & 0xfe) == 0xc2)
8850bf83 173
22901f30 174#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len))))
b2635aa8
KW
175
176/* Masks out the initial one bits in a start byte, leaving the real data ones.
177 * Doesn't work on an invariant byte */
22901f30 178#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6
NIS
179
180#define UTF_CONTINUATION_MARK 0x80
181#define UTF_ACCUMULATION_SHIFT 6
b2635aa8
KW
182
183/* 2**UTF_ACCUMULATION_SHIFT - 1 */
1d72bdf6 184#define UTF_CONTINUATION_MASK ((U8)0x3f)
c512ce4f 185
eb83ed87
KW
186/* This sets the UTF_CONTINUATION_MASK in the upper bits of a word. If a value
187 * is anded with it, and the result is non-zero, then using the original value
188 * in UTF8_ACCUMULATE will overflow, shifting bits off the left */
189#define UTF_ACCUMULATION_OVERFLOW_MASK \
bb88be5f
KW
190 (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) \
191 - UTF_ACCUMULATION_SHIFT))
eb83ed87 192
1d68d6cd 193#ifdef HAS_QUAD
5bbb0b5a 194#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
195 (uv) < 0x800 ? 2 : \
196 (uv) < 0x10000 ? 3 : \
197 (uv) < 0x200000 ? 4 : \
198 (uv) < 0x4000000 ? 5 : \
199 (uv) < 0x80000000 ? 6 : \
9041c2e3 200 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC
201#else
202/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5bbb0b5a 203#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
204 (uv) < 0x800 ? 2 : \
205 (uv) < 0x10000 ? 3 : \
206 (uv) < 0x200000 ? 4 : \
207 (uv) < 0x4000000 ? 5 : \
208 (uv) < 0x80000000 ? 6 : 7 )
209#endif
210
d06134e5
KW
211#endif /* EBCDIC vs ASCII */
212
213/* Rest of these are attributes of Unicode and perl's internals rather than the
214 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
215 * this level; the macros that some of these call may have different
216 * definitions in the two encodings */
217
218#define NATIVE8_TO_UNI(ch) NATIVE_TO_ASCII(ch) /* a clearer synonym */
219
bb88be5f
KW
220#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) \
221 | (((U8)new) & UTF_CONTINUATION_MASK))
d06134e5 222
2950f2a7
KW
223/* Convert a two (not one) byte utf8 character to a unicode code point value.
224 * Needs just one iteration of accumulate. Should not be used unless it is
225 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
226 * Note that the result can be larger than 255 if the input character is not
227 * downgradable */
228#define TWO_BYTE_UTF8_TO_UNI(HI, LO) \
229 UTF8_ACCUMULATE((NATIVE_TO_UTF(HI) & UTF_START_MASK(2)), \
230 NATIVE_TO_UTF(LO))
231
d06134e5
KW
232#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
233
234#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c))
235#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE8_TO_UNI(c))
236
237#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
238
239/* The macros in the next sets are used to generate the two utf8 or utfebcdic
240 * bytes from an ordinal that is known to fit into two bytes; it must be less
241 * than 0x3FF to work across both encodings. */
242/* Nocast allows these to be used in the case label of a switch statement */
428921e5
KW
243#define UTF8_TWO_BYTE_HI_nocast(c) NATIVE_TO_I8(((c) \
244 >> UTF_ACCUMULATION_SHIFT) | (0xFF & UTF_START_MARK(2)))
245#define UTF8_TWO_BYTE_LO_nocast(c) NATIVE_TO_I8(((c) & UTF_CONTINUATION_MASK) \
246 | UTF_CONTINUATION_MARK)
d06134e5
KW
247
248#define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c)))
249#define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c)))
250
251/* This name is used when the source is a single byte */
252#define UTF8_EIGHT_BIT_HI(c) UTF8_TWO_BYTE_HI((U8)(c))
253#define UTF8_EIGHT_BIT_LO(c) UTF8_TWO_BYTE_LO((U8)(c))
254
7e2040f0 255/*
e3036cf4 256 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
20df05f4
KW
257 * from an earlier version of these macros in which they didn't call the
258 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
259 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
260 * no need to do it again here
7e2040f0 261 */
e3036cf4
KW
262#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \
263 ? isIDFIRST(*(p)) \
264 : isIDFIRST_utf8((const U8*)p))
265#define isALNUM_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \
266 ? isALNUM(*(p)) \
267 : isALNUM_utf8((const U8*)p))
1d72bdf6 268
7e2040f0
GS
269#define isIDFIRST_lazy(p) isIDFIRST_lazy_if(p,1)
270#define isALNUM_lazy(p) isALNUM_lazy_if(p,1)
3bd709b1 271
89ebb4a3
JH
272#define UTF8_MAXBYTES 13
273/* How wide can a single UTF-8 encoded character become in bytes.
274 * NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8
275 * since UTF-8 is an encoding of Unicode and given Unicode's current
276 * upper limit only four bytes is possible. Perl thinks of UTF-8
277 * as a way to encode non-negative integers in a binary format. */
278#define UTF8_MAXLEN UTF8_MAXBYTES
279
89ebb4a3
JH
280/* The maximum number of UTF-8 bytes a single Unicode character can
281 * uppercase/lowercase/fold into; this number depends on the Unicode
282 * version. An example of maximal expansion is the U+03B0 which
283 * uppercases to U+03C5 U+0308 U+0301. The Unicode databases that
d06134e5 284 * tell these things are UnicodeData.txt, CaseFolding.txt, and
88d45ddf
KW
285 * SpecialCasing.txt. The value is 6 for strict Unicode characters, but it has
286 * to be as big as Perl allows for a single character */
287#define UTF8_MAXBYTES_CASE UTF8_MAXBYTES
3bd709b1 288
8cb75cc8
KW
289/* A Unicode character can fold to up to 3 characters */
290#define UTF8_MAX_FOLD_CHAR_EXPAND 3
291
a98fe34d 292#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 293#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
66cbab2c 294#define IN_UNI_8_BIT \
b36bf33f
KW
295 (CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT|HINT_LOCALE_NOT_CHARS) \
296 && ! IN_LOCALE_RUNTIME && ! IN_BYTES)
297
1d72bdf6 298
c76687c5
KW
299#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
300
301/* Allow first byte to be a continuation byte */
1d72bdf6 302#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
303
304/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 305#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
306
307/* expecting more bytes than were available in the string */
308#define UTF8_ALLOW_SHORT 0x0008
309
310/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
311#define UTF8_ALLOW_LONG 0x0010
312
313#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
314#define UTF8_WARN_SURROGATE 0x0040
315
316#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
317#define UTF8_WARN_NONCHAR 0x0100 /* code points */
318
319#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
320#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
321
322/* Code points which never were part of the original UTF-8 standard, the first
323 * byte of which is a FE or FF on ASCII platforms. */
324#define UTF8_DISALLOW_FE_FF 0x0800
325#define UTF8_WARN_FE_FF 0x1000
326
327#define UTF8_CHECK_ONLY 0x2000
328
329/* For backwards source compatibility. They do nothing, as the default now
330 * includes what they used to mean. The first one's meaning was to allow the
331 * just the single non-character 0xFFFF */
332#define UTF8_ALLOW_FFFF 0
333#define UTF8_ALLOW_SURROGATE 0
334
33d9abfb 335#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
949cf498 336#define UTF8_WARN_ILLEGAL_INTERCHANGE \
33d9abfb 337 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
949cf498
KW
338#define UTF8_ALLOW_ANY \
339 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
340#define UTF8_ALLOW_ANYUV \
341 (UTF8_ALLOW_EMPTY \
342 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
343#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
344 UTF8_ALLOW_ANYUV)
1d72bdf6 345
7131f24d
KW
346/* Surrogates, non-character code points and above-Unicode code points are
347 * problematic in some contexts. This allows code that needs to check for
348 * those to to quickly exclude the vast majority of code points it will
349 * encounter */
b96a92fb
KW
350#define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
351 FIRST_SURROGATE_UTF8_FIRST_BYTE
7131f24d 352
b96a92fb
KW
353#define UTF8_IS_SURROGATE(s) cBOOL(is_SURROGATE_utf8(s))
354#define UTF8_IS_REPLACEMENT(s, send) cBOOL(is_REPLACEMENT_utf8_safe(s,send))
7131f24d
KW
355
356/* ASCII EBCDIC I8
357 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
358 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
359 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
360 */
361#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
428921e5
KW
362# define UTF8_IS_SUPER(s) (NATIVE_TO_I8(*(s)) >= 0xF9 \
363 && (NATIVE_TO_I8(*(s)) > 0xF9) || (NATIVE_TO_I8(*((s)) + 1 >= 0xA2)))
7131f24d
KW
364#else
365# define UTF8_IS_SUPER(s) (*(s) >= 0xF4 \
366 && (*(s) > 0xF4 || (*((s) + 1) >= 0x90)))
367#endif
368
b96a92fb
KW
369/* These are now machine generated, and the 'given' clause is no longer
370 * applicable */
371#define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
372 cBOOL(is_NONCHAR_utf8(s))
373#define UTF8_IS_NONCHAR_(s) \
374 UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s)
7131f24d 375
c867b360
JH
376#define UNICODE_SURROGATE_FIRST 0xD800
377#define UNICODE_SURROGATE_LAST 0xDFFF
378#define UNICODE_REPLACEMENT 0xFFFD
379#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 380
b851fbc1 381/* Though our UTF-8 encoding can go beyond this,
c76687c5 382 * let's be conservative and do as Unicode says. */
b851fbc1
JH
383#define PERL_UNICODE_MAX 0x10FFFF
384
949cf498
KW
385#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
386#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
387#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
388#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
389#define UNICODE_DISALLOW_SURROGATE 0x0010
390#define UNICODE_DISALLOW_NONCHAR 0x0020
391#define UNICODE_DISALLOW_SUPER 0x0040
392#define UNICODE_DISALLOW_FE_FF 0x0080
bb88be5f
KW
393#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
394 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
395#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
396 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
949cf498
KW
397
398/* For backward source compatibility, as are now the default */
399#define UNICODE_ALLOW_SURROGATE 0
400#define UNICODE_ALLOW_SUPER 0
401#define UNICODE_ALLOW_ANY 0
b851fbc1 402
1d72bdf6
NIS
403#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
404 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 405#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 406#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW
407#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
408 /* The other noncharacters end in FFFE or FFFF, which \
409 * the mask below catches both of, but beyond the last \
410 * official unicode code point, they aren't \
411 * noncharacters, since those aren't Unicode \
412 * characters at all */ \
413 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
414#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
415#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
1d72bdf6 416
f067b878
NA
417#ifdef HAS_QUAD
418# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
419#endif
3bd709b1 420
ec34087a
KW
421#define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE
422#define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \
423 LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE
424#define MICRO_SIGN MICRO_SIGN_NATIVE
425#define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \
426 LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE
427#define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \
428 LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE
09091399
JH
429#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
430#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
431#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 432#define GREEK_SMALL_LETTER_MU 0x03BC
78a0d3cc
KW
433#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */
434#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
97298f37 435#define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
8d64d87f
KW
436#define LATIN_SMALL_LETTER_LONG_S 0x017F
437#define KELVIN_SIGN 0x212A
438#define ANGSTROM_SIGN 0x212B
09091399 439
9e55ce06 440#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
441#define UNI_DISPLAY_BACKSLASH 0x0002
442#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
443#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 444
5cd46e1f
KW
445#define ANYOF_FOLD_SHARP_S(node, input, end) \
446 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 447 (ANYOF_NONBITMAP(node)) && \
39065660 448 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f
JH
449 ((end) > (input) + 1) && \
450 toLOWER((input)[0]) == 's' && \
451 toLOWER((input)[1]) == 's')
ebc501f0 452#define SHARP_S_SKIP 2
3b0fc154 453
2f454f11 454#ifndef EBCDIC
a4f7a67c
KW
455/* If you want to exclude surrogates, and beyond legal Unicode, see the blame
456 * log for earlier versions which gave details for these */
c78f6c49 457# define IS_UTF8_CHAR_1(p) \
3b0fc154 458 ((p)[0] <= 0x7F)
c78f6c49 459# define IS_UTF8_CHAR_2(p) \
3b0fc154
JH
460 ((p)[0] >= 0xC2 && (p)[0] <= 0xDF && \
461 (p)[1] >= 0x80 && (p)[1] <= 0xBF)
c78f6c49 462# define IS_UTF8_CHAR_3a(p) \
3b0fc154
JH
463 ((p)[0] == 0xE0 && \
464 (p)[1] >= 0xA0 && (p)[1] <= 0xBF && \
465 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49 466# define IS_UTF8_CHAR_3b(p) \
a4f7a67c 467 ((p)[0] >= 0xE1 && (p)[0] <= 0xEF && \
3b0fc154
JH
468 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
469 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49 470# define IS_UTF8_CHAR_4a(p) \
3b0fc154
JH
471 ((p)[0] == 0xF0 && \
472 (p)[1] >= 0x90 && (p)[1] <= 0xBF && \
473 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
474 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
a4f7a67c 475/* The 0xF7 allows us to go to 0x1fffff (0x200000 would
3b0fc154
JH
476 * require five bytes). Not doing any further code points
477 * since that is not needed (and that would not be strict
478 * UTF-8, anyway). The "slow path" in Perl_is_utf8_char()
479 * will take care of the "extended UTF-8". */
a4f7a67c
KW
480# define IS_UTF8_CHAR_4b(p) \
481 ((p)[0] >= 0xF1 && (p)[0] <= 0xF7 && \
3b0fc154
JH
482 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
483 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
484 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
485
c78f6c49 486# define IS_UTF8_CHAR_3(p) \
3b0fc154 487 (IS_UTF8_CHAR_3a(p) || \
a4f7a67c 488 IS_UTF8_CHAR_3b(p))
c78f6c49 489# define IS_UTF8_CHAR_4(p) \
3b0fc154 490 (IS_UTF8_CHAR_4a(p) || \
a4f7a67c 491 IS_UTF8_CHAR_4b(p))
3b0fc154
JH
492
493/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
494 * (1) allows UTF-8 encoded UTF-16 surrogates
495 * (2) it allows code points past U+10FFFF.
496 * The Perl_is_utf8_char() full "slow" code will handle the Perl
497 * "extended UTF-8". */
c78f6c49 498# define IS_UTF8_CHAR(p, n) \
3b0fc154
JH
499 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
500 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
501 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
502 (n) == 4 ? IS_UTF8_CHAR_4(p) : 0)
503
c78f6c49 504# define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
768c67ee 505
2f454f11
KW
506#else /* EBCDIC */
507
508/* This is an attempt to port IS_UTF8_CHAR to EBCDIC based on eyeballing.
509 * untested. If want to exclude surrogates and above-Unicode, see the
510 * definitions for UTF8_IS_SURROGATE and UTF8_IS_SUPER */
511# define IS_UTF8_CHAR_1(p) \
512 (NATIVE_TO_ASCII((p)[0]) <= 0x9F)
513# define IS_UTF8_CHAR_2(p) \
514 (NATIVE_TO_I8((p)[0]) >= 0xC5 && NATIVE_TO_I8((p)[0]) <= 0xDF && \
515 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF)
516# define IS_UTF8_CHAR_3(p) \
517 (NATIVE_TO_I8((p)[0]) == 0xE1 && NATIVE_TO_I8((p)[1]) <= 0xEF && \
518 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
519 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF)
520# define IS_UTF8_CHAR_4a(p) \
521 (NATIVE_TO_I8((p)[0]) == 0xF0 && \
522 NATIVE_TO_I8((p)[1]) >= 0xB0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
523 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
524 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
525# define IS_UTF8_CHAR_4b(p) \
526 (NATIVE_TO_I8((p)[0]) >= 0xF1 && NATIVE_TO_I8((p)[0]) <= 0xF7 && \
527 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
528 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
529 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
530# define IS_UTF8_CHAR_5a(p) \
531 (NATIVE_TO_I8((p)[0]) == 0xF8 && \
532 NATIVE_TO_I8((p)[1]) >= 0xA8 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
533 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
534 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
535 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
536# define IS_UTF8_CHAR_5b(p) \
537 (NATIVE_TO_I8((p)[0]) >= 0xF9 && NATIVE_TO_I8((p)[1]) <= 0xFB && \
538 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
539 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
540 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
541 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
542
543# define IS_UTF8_CHAR_4(p) \
544 (IS_UTF8_CHAR_4a(p) || \
545 IS_UTF8_CHAR_4b(p))
546# define IS_UTF8_CHAR_5(p) \
547 (IS_UTF8_CHAR_5a(p) || \
548 IS_UTF8_CHAR_5b(p))
549# define IS_UTF8_CHAR(p, n) \
550 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
551 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
552 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
553 (n) == 4 ? IS_UTF8_CHAR_4(p) : \
554 (n) == 5 ? IS_UTF8_CHAR_5(p) : 0)
555
556# define IS_UTF8_CHAR_FAST(n) ((n) <= 5)
557
77263263 558#endif /* IS_UTF8_CHAR() for UTF-8 */
e9a8c099
MHM
559
560/*
561 * Local variables:
562 * c-indentation-style: bsd
563 * c-basic-offset: 4
14d04a33 564 * indent-tabs-mode: nil
e9a8c099
MHM
565 * End:
566 *
14d04a33 567 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 568 */