This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
utf8.c: Add comment
[perl5.git] / utf8.h
CommitLineData
a0ed51b3
LW
1/* utf8.h
2 *
83706693 3 * Copyright (C) 2000, 2001, 2002, 2005, 2006, 2007, 2009 by Larry Wall and others
a0ed51b3
LW
4 *
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
7 *
8 */
9
39e02b42 10/* Use UTF-8 as the default script encoding?
1e54db1a 11 * Turning this on will break scripts having non-UTF-8 binary
39e02b42
JH
12 * data (such as Latin-1) in string literals. */
13#ifdef USE_UTF8_SCRIPTS
14# define USE_UTF8_IN_NAMES (!IN_BYTES)
15#else
16# define USE_UTF8_IN_NAMES (PL_hints & HINT_UTF8)
17#endif
18
36bb2ab6
KW
19#define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, 1)
20#define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, 1)
21
fd7cb289
RGS
22/* Source backward compatibility. */
23#define uvuni_to_utf8(d, uv) uvuni_to_utf8_flags(d, uv, 0)
24#define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
25
eda9cac1
KW
26#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
27 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
a33c29bc 28#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
5e64d0fa 29#define FOLDEQ_UTF8_LOCALE (1 << 1)
18f762c3
KW
30#define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
31#define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
a33c29bc 32
e6226b18
KW
33/*
34=for apidoc ibcmp_utf8
35
36This is a synonym for (! foldEQ_utf8())
37
38=cut
39*/
40#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
41 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
42
1d72bdf6
NIS
43#ifdef EBCDIC
44/* The equivalent of these macros but implementing UTF-EBCDIC
45 are in the following header file:
46 */
47
48#include "utfebcdic.h"
fd7cb289 49
d06134e5 50#else /* ! EBCDIC */
73c4f7a1
GS
51START_EXTERN_C
52
a0ed51b3 53#ifdef DOINIT
6f06b55f 54EXTCONST unsigned char PL_utf8skip[] = {
a0ed51b3
LW
551,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
561,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
571,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
581,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
591,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus */
601,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus */
612,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* scripts */
3c77ea2b 623,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* cjk etc. */
7eb168e2
KW
637,13, /* Perl extended (not official UTF-8). Up to 72bit allowed (64-bit +
64 reserved). */
a0ed51b3
LW
65};
66#else
6f06b55f 67EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
68#endif
69
73c4f7a1 70END_EXTERN_C
7e2040f0 71
1d72bdf6
NIS
72/* Native character to iso-8859-1 */
73#define NATIVE_TO_ASCII(ch) (ch)
74#define ASCII_TO_NATIVE(ch) (ch)
75/* Transform after encoding */
76#define NATIVE_TO_UTF(ch) (ch)
77#define UTF_TO_NATIVE(ch) (ch)
78/* Transforms in wide UV chars */
79#define UNI_TO_NATIVE(ch) (ch)
80#define NATIVE_TO_UNI(ch) (ch)
81/* Transforms in invariant space */
82#define NATIVE_TO_NEED(enc,ch) (ch)
83#define ASCII_TO_NEED(enc,ch) (ch)
d7578b48 84
d06134e5 85/* As there are no translations, avoid the function wrapper */
1754c1a1
NC
86#define utf8n_to_uvchr utf8n_to_uvuni
87#define uvchr_to_utf8 uvuni_to_utf8
2b9d42f0 88
877d9f0d 89/*
9041c2e3 90
8c007b5a 91 The following table is from Unicode 3.2.
877d9f0d
JH
92
93 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
94
375122d7 95 U+0000..U+007F 00..7F
e1b711da 96 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 97 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 98 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 99 U+D000..U+D7FF ED 80..9F 80..BF
37e2e78e 100 U+D800..U+DFFF +++++++ utf16 surrogates, not legal utf8 +++++++
375122d7 101 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 102 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH
103 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
104 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
105
e1b711da 106Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
107caused by legal UTF-8 avoiding non-shortest encodings: it is technically
108possible to UTF-8-encode a single code point in different ways, but that is
109explicitly forbidden, and the shortest possible encoding should always be used
110(and that is what Perl does).
8c007b5a 111
877d9f0d
JH
112 */
113
8c007b5a
JH
114/*
115 Another way to look at it, as bits:
116
117 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
118
119 0aaaaaaa 0aaaaaaa
120 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
121 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
122 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
123
124As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 125leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
126encoded character.
127
65ab9279
TC
128Perl's extended UTF-8 means we can have start bytes up to FF.
129
8c007b5a
JH
130*/
131
c4d5f83a 132#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
d06134e5
KW
133/* Note that C0 and C1 are invalid in legal UTF8, so the lower bound of the
134 * below might ought to be C2 */
65ab9279 135#define UTF8_IS_START(c) (((U8)c) >= 0xc0)
c512ce4f
JH
136#define UTF8_IS_CONTINUATION(c) (((U8)c) >= 0x80 && (((U8)c) <= 0xbf))
137#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
db42d148 138#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)c & 0xfc) == 0xc0)
8850bf83 139
22901f30
RGS
140#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len))))
141#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6
NIS
142
143#define UTF_CONTINUATION_MARK 0x80
144#define UTF_ACCUMULATION_SHIFT 6
145#define UTF_CONTINUATION_MASK ((U8)0x3f)
c512ce4f 146
1d68d6cd 147#ifdef HAS_QUAD
5bbb0b5a 148#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
149 (uv) < 0x800 ? 2 : \
150 (uv) < 0x10000 ? 3 : \
151 (uv) < 0x200000 ? 4 : \
152 (uv) < 0x4000000 ? 5 : \
153 (uv) < 0x80000000 ? 6 : \
9041c2e3 154 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC
155#else
156/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5bbb0b5a 157#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
158 (uv) < 0x800 ? 2 : \
159 (uv) < 0x10000 ? 3 : \
160 (uv) < 0x200000 ? 4 : \
161 (uv) < 0x4000000 ? 5 : \
162 (uv) < 0x80000000 ? 6 : 7 )
163#endif
164
d06134e5
KW
165#endif /* EBCDIC vs ASCII */
166
167/* Rest of these are attributes of Unicode and perl's internals rather than the
168 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
169 * this level; the macros that some of these call may have different
170 * definitions in the two encodings */
171
172#define NATIVE8_TO_UNI(ch) NATIVE_TO_ASCII(ch) /* a clearer synonym */
173
174#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) | (((U8)new) & UTF_CONTINUATION_MASK))
175
2950f2a7
KW
176/* Convert a two (not one) byte utf8 character to a unicode code point value.
177 * Needs just one iteration of accumulate. Should not be used unless it is
178 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
179 * Note that the result can be larger than 255 if the input character is not
180 * downgradable */
181#define TWO_BYTE_UTF8_TO_UNI(HI, LO) \
182 UTF8_ACCUMULATE((NATIVE_TO_UTF(HI) & UTF_START_MASK(2)), \
183 NATIVE_TO_UTF(LO))
184
d06134e5
KW
185#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
186
187#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c))
188#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE8_TO_UNI(c))
189
190#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
191
192/* The macros in the next sets are used to generate the two utf8 or utfebcdic
193 * bytes from an ordinal that is known to fit into two bytes; it must be less
194 * than 0x3FF to work across both encodings. */
195/* Nocast allows these to be used in the case label of a switch statement */
0072151e 196#define UTF8_TWO_BYTE_HI_nocast(c) UTF_TO_NATIVE(((c) >> UTF_ACCUMULATION_SHIFT) | (0xFF & UTF_START_MARK(2)))
14c28784 197#define UTF8_TWO_BYTE_LO_nocast(c) UTF_TO_NATIVE(((c) & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK)
d06134e5
KW
198
199#define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c)))
200#define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c)))
201
202/* This name is used when the source is a single byte */
203#define UTF8_EIGHT_BIT_HI(c) UTF8_TWO_BYTE_HI((U8)(c))
204#define UTF8_EIGHT_BIT_LO(c) UTF8_TWO_BYTE_LO((U8)(c))
205
7e2040f0 206/*
e3036cf4 207 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
20df05f4
KW
208 * from an earlier version of these macros in which they didn't call the
209 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
210 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
211 * no need to do it again here
7e2040f0 212 */
e3036cf4
KW
213#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \
214 ? isIDFIRST(*(p)) \
215 : isIDFIRST_utf8((const U8*)p))
216#define isALNUM_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \
217 ? isALNUM(*(p)) \
218 : isALNUM_utf8((const U8*)p))
1d72bdf6 219
7e2040f0
GS
220#define isIDFIRST_lazy(p) isIDFIRST_lazy_if(p,1)
221#define isALNUM_lazy(p) isALNUM_lazy_if(p,1)
3bd709b1 222
89ebb4a3
JH
223#define UTF8_MAXBYTES 13
224/* How wide can a single UTF-8 encoded character become in bytes.
225 * NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8
226 * since UTF-8 is an encoding of Unicode and given Unicode's current
227 * upper limit only four bytes is possible. Perl thinks of UTF-8
228 * as a way to encode non-negative integers in a binary format. */
229#define UTF8_MAXLEN UTF8_MAXBYTES
230
89ebb4a3
JH
231/* The maximum number of UTF-8 bytes a single Unicode character can
232 * uppercase/lowercase/fold into; this number depends on the Unicode
233 * version. An example of maximal expansion is the U+03B0 which
234 * uppercases to U+03C5 U+0308 U+0301. The Unicode databases that
d06134e5 235 * tell these things are UnicodeData.txt, CaseFolding.txt, and
88d45ddf
KW
236 * SpecialCasing.txt. The value is 6 for strict Unicode characters, but it has
237 * to be as big as Perl allows for a single character */
238#define UTF8_MAXBYTES_CASE UTF8_MAXBYTES
3bd709b1 239
8cb75cc8
KW
240/* A Unicode character can fold to up to 3 characters */
241#define UTF8_MAX_FOLD_CHAR_EXPAND 3
242
a98fe34d 243#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 244#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
1863b879 245#define IN_UNI_8_BIT ( (CopHINTS_get(PL_curcop) & HINT_UNI_8_BIT) \
61fc5122 246 && ! IN_LOCALE_RUNTIME && ! IN_BYTES)
1d72bdf6 247
c76687c5
KW
248#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
249
250/* Allow first byte to be a continuation byte */
1d72bdf6 251#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
252
253/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 254#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
255
256/* expecting more bytes than were available in the string */
257#define UTF8_ALLOW_SHORT 0x0008
258
259/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
260#define UTF8_ALLOW_LONG 0x0010
261
262#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
263#define UTF8_WARN_SURROGATE 0x0040
264
265#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
266#define UTF8_WARN_NONCHAR 0x0100 /* code points */
267
268#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
269#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
270
271/* Code points which never were part of the original UTF-8 standard, the first
272 * byte of which is a FE or FF on ASCII platforms. */
273#define UTF8_DISALLOW_FE_FF 0x0800
274#define UTF8_WARN_FE_FF 0x1000
275
276#define UTF8_CHECK_ONLY 0x2000
277
278/* For backwards source compatibility. They do nothing, as the default now
279 * includes what they used to mean. The first one's meaning was to allow the
280 * just the single non-character 0xFFFF */
281#define UTF8_ALLOW_FFFF 0
282#define UTF8_ALLOW_SURROGATE 0
283
33d9abfb 284#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
949cf498 285#define UTF8_WARN_ILLEGAL_INTERCHANGE \
33d9abfb 286 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
949cf498
KW
287#define UTF8_ALLOW_ANY \
288 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
289#define UTF8_ALLOW_ANYUV \
290 (UTF8_ALLOW_EMPTY \
291 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
292#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
293 UTF8_ALLOW_ANYUV)
1d72bdf6 294
7131f24d
KW
295/* Surrogates, non-character code points and above-Unicode code points are
296 * problematic in some contexts. This allows code that needs to check for
297 * those to to quickly exclude the vast majority of code points it will
298 * encounter */
299#ifdef EBCDIC
300# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE UTF_TO_NATIVE(0xF1)
301#else
302# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE 0xED
303#endif
304
305/* ASCII EBCDIC I8
306 * U+D7FF: \xED\x9F\xBF \xF1\xB5\xBF\xBF last before surrogates
307 * U+D800: \xED\xA0\x80 \xF1\xB6\xA0\xA0 1st surrogate
308 * U+DFFF: \xED\xBF\xBF \xF1\xB7\xBF\xBF final surrogate
309 * U+E000: \xEE\x80\x80 \xF1\xB8\xA0\xA0 next after surrogates
310 */
311#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
312# define UTF8_IS_SURROGATE(s) (*(s) == UTF_TO_NATIVE(0xF1) \
880d3d38 313 && ((*((s) +1) == UTF_TO_NATIVE(0xB6)) || *((s) + 1) == UTF_TO_NATIVE(0xB7)))
7131f24d
KW
314#else
315# define UTF8_IS_SURROGATE(s) (*(s) == 0xED && *((s) + 1) >= 0xA0)
316#endif
317
318/* ASCII EBCDIC I8
319 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
320 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
321 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
322 */
323#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
324# define UTF8_IS_SUPER(s) (*(s) >= UTF_TO_NATIVE(0xF9) \
325 && (*(s) > UTF_TO_NATIVE(0xF9) || (*((s) + 1) >= UTF_TO_NATIVE(0xA2))))
326#else
327# define UTF8_IS_SUPER(s) (*(s) >= 0xF4 \
328 && (*(s) > 0xF4 || (*((s) + 1) >= 0x90)))
329#endif
330
331/* ASCII EBCDIC I8
332 * U+FDCF: \xEF\xB7\x8F \xF1\xBF\xAE\xAF last before non-char block
333 * U+FDD0: \xEF\xB7\x90 \xF1\xBF\xAE\xB0 first non-char in block
334 * U+FDEF: \xEF\xB7\xAF \xF1\xBF\xAF\xAF last non-char in block
335 * U+FDF0: \xEF\xB7\xB0 \xF1\xBF\xAF\xB0 first after non-char block
336 * U+FFFF: \xEF\xBF\xBF \xF1\xBF\xBF\xBF
337 * U+1FFFF: \xF0\x9F\xBF\xBF \xF3\xBF\xBF\xBF
338 * U+2FFFF: \xF0\xAF\xBF\xBF \xF5\xBF\xBF\xBF
339 * U+3FFFF: \xF0\xBF\xBF\xBF \xF7\xBF\xBF\xBF
340 * U+4FFFF: \xF1\x8F\xBF\xBF \xF8\xA9\xBF\xBF\xBF
341 * U+5FFFF: \xF1\x9F\xBF\xBF \xF8\xAB\xBF\xBF\xBF
342 * U+6FFFF: \xF1\xAF\xBF\xBF \xF8\xAD\xBF\xBF\xBF
343 * U+7FFFF: \xF1\xBF\xBF\xBF \xF8\xAF\xBF\xBF\xBF
344 * U+8FFFF: \xF2\x8F\xBF\xBF \xF8\xB1\xBF\xBF\xBF
345 * U+9FFFF: \xF2\x9F\xBF\xBF \xF8\xB3\xBF\xBF\xBF
346 * U+AFFFF: \xF2\xAF\xBF\xBF \xF8\xB5\xBF\xBF\xBF
347 * U+BFFFF: \xF2\xBF\xBF\xBF \xF8\xB7\xBF\xBF\xBF
348 * U+CFFFF: \xF3\x8F\xBF\xBF \xF8\xB9\xBF\xBF\xBF
349 * U+DFFFF: \xF3\x9F\xBF\xBF \xF8\xBB\xBF\xBF\xBF
350 * U+EFFFF: \xF3\xAF\xBF\xBF \xF8\xBD\xBF\xBF\xBF
351 * U+FFFFF: \xF3\xBF\xBF\xBF \xF8\xBF\xBF\xBF\xBF
352 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF
353 */
354#define UTF8_IS_NONCHAR_(s) ( \
355 *(s) >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
356 && ! UTF8_IS_SUPER(s) \
357 && UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_FIRST_PROBLEMATIC(s) \
358
359#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
360# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
361 ((*(s) == UTF_TO_NATIVE(0xF1) \
362 && (*((s) + 1) == UTF_TO_NATIVE(0xBF) \
363 && ((*((s) + 2) == UTF_TO_NATIVE(0xAE) \
364 && *((s) + 3) >= UTF_TO_NATIVE(0xB0)) \
365 || (*((s) + 2) == UTF_TO_NATIVE(0xAF) \
366 && *((s) + 3) <= UTF_TO_NATIVE(0xAF))))) \
367 || (UTF8SKIP(*(s)) > 3 \
368 /* (These were all derived by inspection and experimentation with an */ \
369 /* editor) The next line checks the next to final byte in the char */ \
370 && *((s) + UTF8SKIP(*(s)) - 2) == UTF_TO_NATIVE(0xBF) \
371 && *((s) + UTF8SKIP(*(s)) - 3) == UTF_TO_NATIVE(0xBF) \
372 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 4)) & 0x81) == 0x81 \
373 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 1)) & 0xBE) == 0XBE))
374#else
375# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
376 ((*(s) == 0xEF \
377 && ((*((s) + 1) == 0xB7 && (*((s) + 2) >= 0x90 && (*((s) + 2) <= 0xAF)))\
378 /* Gets U+FFF[EF] */ \
379 || (*((s) + 1) == 0xBF && ((*((s) + 2) & 0xBE) == 0xBE)))) \
380 || ((*((s) + 2) == 0xBF \
381 && (*((s) + 3) & 0xBE) == 0xBE \
382 /* Excludes things like U+10FFE = \xF0\x90\xBF\xBE */ \
383 && (*((s) + 1) & 0x8F) == 0x8F)))
384#endif
385
c867b360
JH
386#define UNICODE_SURROGATE_FIRST 0xD800
387#define UNICODE_SURROGATE_LAST 0xDFFF
388#define UNICODE_REPLACEMENT 0xFFFD
389#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 390
b851fbc1 391/* Though our UTF-8 encoding can go beyond this,
c76687c5 392 * let's be conservative and do as Unicode says. */
b851fbc1
JH
393#define PERL_UNICODE_MAX 0x10FFFF
394
949cf498
KW
395#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
396#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
397#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
398#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
399#define UNICODE_DISALLOW_SURROGATE 0x0010
400#define UNICODE_DISALLOW_NONCHAR 0x0020
401#define UNICODE_DISALLOW_SUPER 0x0040
402#define UNICODE_DISALLOW_FE_FF 0x0080
403#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
404 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
405#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
406 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
407
408/* For backward source compatibility, as are now the default */
409#define UNICODE_ALLOW_SURROGATE 0
410#define UNICODE_ALLOW_SUPER 0
411#define UNICODE_ALLOW_ANY 0
b851fbc1 412
1d72bdf6
NIS
413#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
414 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 415#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 416#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW
417#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
418 /* The other noncharacters end in FFFE or FFFF, which \
419 * the mask below catches both of, but beyond the last \
420 * official unicode code point, they aren't \
421 * noncharacters, since those aren't Unicode \
422 * characters at all */ \
423 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
424#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
425#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
1d72bdf6 426
f067b878
NA
427#ifdef HAS_QUAD
428# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
429#endif
3bd709b1 430
09091399
JH
431#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
432#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
433#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 434#define GREEK_SMALL_LETTER_MU 0x03BC
78a0d3cc
KW
435#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */
436#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
97298f37 437#define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
09091399 438
9e55ce06 439#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
440#define UNI_DISPLAY_BACKSLASH 0x0002
441#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
442#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 443
5cd46e1f
KW
444#ifndef EBCDIC
445# define LATIN_SMALL_LETTER_SHARP_S 0x00DF
446# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS 0x00FF
447# define MICRO_SIGN 0x00B5
f508a607
KW
448# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE 0x00C5
449# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE 0x00E5
5cd46e1f
KW
450#endif
451
452#define ANYOF_FOLD_SHARP_S(node, input, end) \
453 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 454 (ANYOF_NONBITMAP(node)) && \
39065660 455 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f
JH
456 ((end) > (input) + 1) && \
457 toLOWER((input)[0]) == 's' && \
458 toLOWER((input)[1]) == 's')
ebc501f0 459#define SHARP_S_SKIP 2
3b0fc154 460
2f454f11 461#ifndef EBCDIC
c78f6c49 462# define IS_UTF8_CHAR_1(p) \
3b0fc154 463 ((p)[0] <= 0x7F)
c78f6c49 464# define IS_UTF8_CHAR_2(p) \
3b0fc154
JH
465 ((p)[0] >= 0xC2 && (p)[0] <= 0xDF && \
466 (p)[1] >= 0x80 && (p)[1] <= 0xBF)
c78f6c49 467# define IS_UTF8_CHAR_3a(p) \
3b0fc154
JH
468 ((p)[0] == 0xE0 && \
469 (p)[1] >= 0xA0 && (p)[1] <= 0xBF && \
470 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49 471# define IS_UTF8_CHAR_3b(p) \
3b0fc154
JH
472 ((p)[0] >= 0xE1 && (p)[0] <= 0xEC && \
473 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
474 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49 475# define IS_UTF8_CHAR_3c(p) \
3b0fc154
JH
476 ((p)[0] == 0xED && \
477 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
478 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49
KW
479 /* In IS_UTF8_CHAR_3c(p) one could use
480 * (p)[1] >= 0x80 && (p)[1] <= 0x9F
481 * if one wanted to exclude surrogates. */
482# define IS_UTF8_CHAR_3d(p) \
3b0fc154
JH
483 ((p)[0] >= 0xEE && (p)[0] <= 0xEF && \
484 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
485 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49 486# define IS_UTF8_CHAR_4a(p) \
3b0fc154
JH
487 ((p)[0] == 0xF0 && \
488 (p)[1] >= 0x90 && (p)[1] <= 0xBF && \
489 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
490 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
c78f6c49 491# define IS_UTF8_CHAR_4b(p) \
3b0fc154
JH
492 ((p)[0] >= 0xF1 && (p)[0] <= 0xF3 && \
493 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
494 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
495 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
496/* In IS_UTF8_CHAR_4c(p) one could use
497 * (p)[0] == 0xF4
498 * if one wanted to stop at the Unicode limit U+10FFFF.
499 * The 0xF7 allows us to go to 0x1fffff (0x200000 would
500 * require five bytes). Not doing any further code points
501 * since that is not needed (and that would not be strict
502 * UTF-8, anyway). The "slow path" in Perl_is_utf8_char()
503 * will take care of the "extended UTF-8". */
c78f6c49 504# define IS_UTF8_CHAR_4c(p) \
a18d6e6e 505 ((p)[0] >= 0xF4 && (p)[0] <= 0xF7 && \
3b0fc154
JH
506 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
507 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
508 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
509
c78f6c49 510# define IS_UTF8_CHAR_3(p) \
3b0fc154
JH
511 (IS_UTF8_CHAR_3a(p) || \
512 IS_UTF8_CHAR_3b(p) || \
513 IS_UTF8_CHAR_3c(p) || \
514 IS_UTF8_CHAR_3d(p))
c78f6c49 515# define IS_UTF8_CHAR_4(p) \
3b0fc154
JH
516 (IS_UTF8_CHAR_4a(p) || \
517 IS_UTF8_CHAR_4b(p) || \
518 IS_UTF8_CHAR_4c(p))
519
520/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
521 * (1) allows UTF-8 encoded UTF-16 surrogates
522 * (2) it allows code points past U+10FFFF.
523 * The Perl_is_utf8_char() full "slow" code will handle the Perl
524 * "extended UTF-8". */
c78f6c49 525# define IS_UTF8_CHAR(p, n) \
3b0fc154
JH
526 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
527 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
528 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
529 (n) == 4 ? IS_UTF8_CHAR_4(p) : 0)
530
c78f6c49 531# define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
768c67ee 532
2f454f11
KW
533#else /* EBCDIC */
534
535/* This is an attempt to port IS_UTF8_CHAR to EBCDIC based on eyeballing.
536 * untested. If want to exclude surrogates and above-Unicode, see the
537 * definitions for UTF8_IS_SURROGATE and UTF8_IS_SUPER */
538# define IS_UTF8_CHAR_1(p) \
539 (NATIVE_TO_ASCII((p)[0]) <= 0x9F)
540# define IS_UTF8_CHAR_2(p) \
541 (NATIVE_TO_I8((p)[0]) >= 0xC5 && NATIVE_TO_I8((p)[0]) <= 0xDF && \
542 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF)
543# define IS_UTF8_CHAR_3(p) \
544 (NATIVE_TO_I8((p)[0]) == 0xE1 && NATIVE_TO_I8((p)[1]) <= 0xEF && \
545 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
546 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF)
547# define IS_UTF8_CHAR_4a(p) \
548 (NATIVE_TO_I8((p)[0]) == 0xF0 && \
549 NATIVE_TO_I8((p)[1]) >= 0xB0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
550 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
551 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
552# define IS_UTF8_CHAR_4b(p) \
553 (NATIVE_TO_I8((p)[0]) >= 0xF1 && NATIVE_TO_I8((p)[0]) <= 0xF7 && \
554 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
555 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
556 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
557# define IS_UTF8_CHAR_5a(p) \
558 (NATIVE_TO_I8((p)[0]) == 0xF8 && \
559 NATIVE_TO_I8((p)[1]) >= 0xA8 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
560 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
561 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
562 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
563# define IS_UTF8_CHAR_5b(p) \
564 (NATIVE_TO_I8((p)[0]) >= 0xF9 && NATIVE_TO_I8((p)[1]) <= 0xFB && \
565 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
566 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
567 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
568 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
569
570# define IS_UTF8_CHAR_4(p) \
571 (IS_UTF8_CHAR_4a(p) || \
572 IS_UTF8_CHAR_4b(p))
573# define IS_UTF8_CHAR_5(p) \
574 (IS_UTF8_CHAR_5a(p) || \
575 IS_UTF8_CHAR_5b(p))
576# define IS_UTF8_CHAR(p, n) \
577 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
578 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
579 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
580 (n) == 4 ? IS_UTF8_CHAR_4(p) : \
581 (n) == 5 ? IS_UTF8_CHAR_5(p) : 0)
582
583# define IS_UTF8_CHAR_FAST(n) ((n) <= 5)
584
77263263 585#endif /* IS_UTF8_CHAR() for UTF-8 */
e9a8c099
MHM
586
587/*
588 * Local variables:
589 * c-indentation-style: bsd
590 * c-basic-offset: 4
591 * indent-tabs-mode: t
592 * End:
593 *
594 * ex: set ts=8 sts=4 sw=4 noet:
595 */