This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
https://perl5.git.perl.org / perl5.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
utf8.c: Shorten hash key for speed
[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
83199d38 28
a0270393 29#define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
051a06d4
KW		 30#define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, \
31 FOLD_FLAGS_FULL, NULL)
32#define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0, NULL)
33#define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0, NULL)
34#define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0, NULL)
36bb2ab6 35
fd7cb289
RGS		 36/* Source backward compatibility. */
37#define uvuni_to_utf8(d, uv) uvuni_to_utf8_flags(d, uv, 0)
38#define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
39
eda9cac1
KW		 40#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
41 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
a33c29bc 42#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
5e64d0fa 43#define FOLDEQ_UTF8_LOCALE (1 << 1)
18f762c3
KW		 44#define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
45#define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
a33c29bc 46
e6226b18
KW		 47/*
48=for apidoc ibcmp_utf8
49
50This is a synonym for (! foldEQ_utf8())
51
52=cut
53*/
54#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
55 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
56
1d72bdf6
NIS		 57#ifdef EBCDIC
58/* The equivalent of these macros but implementing UTF-EBCDIC
59 are in the following header file:
60 */
61
62#include "utfebcdic.h"
fd7cb289 63
d06134e5 64#else /* ! EBCDIC */
73c4f7a1
GS		 65START_EXTERN_C
66
a0ed51b3 67#ifdef DOINIT
6f06b55f 68EXTCONST unsigned char PL_utf8skip[] = {
b2635aa8
KW		 69/* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
70/* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
71/* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
72/* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
73/* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
74/* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
75/* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
76/* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
77/* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
78/* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
79/* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
80/* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
81/* 0xC0 */ 2,2, /* overlong */
82/* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
83/* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */
84/* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */
85/* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */
86/* 0xFE */ 7,13, /* Perl extended (never was official UTF-8). Up to 72bit
87 allowed (64-bit + reserved). */
a0ed51b3
LW		 88};
89#else
6f06b55f 90EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW		 91#endif
92
73c4f7a1 93END_EXTERN_C
7e2040f0 94
1d72bdf6
NIS		 95/* Native character to iso-8859-1 */
96#define NATIVE_TO_ASCII(ch) (ch)
97#define ASCII_TO_NATIVE(ch) (ch)
98/* Transform after encoding */
99#define NATIVE_TO_UTF(ch) (ch)
100#define UTF_TO_NATIVE(ch) (ch)
101/* Transforms in wide UV chars */
102#define UNI_TO_NATIVE(ch) (ch)
103#define NATIVE_TO_UNI(ch) (ch)
104/* Transforms in invariant space */
105#define NATIVE_TO_NEED(enc,ch) (ch)
106#define ASCII_TO_NEED(enc,ch) (ch)
d7578b48 107
d06134e5 108/* As there are no translations, avoid the function wrapper */
1754c1a1 109#define utf8n_to_uvchr utf8n_to_uvuni
6dd9dce9 110#define valid_utf8_to_uvchr valid_utf8_to_uvuni
1754c1a1 111#define uvchr_to_utf8 uvuni_to_utf8
2b9d42f0 112
877d9f0d 113/*
9041c2e3 114
8c007b5a 115 The following table is from Unicode 3.2.
877d9f0d
JH		 116
117 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
118
375122d7 119 U+0000..U+007F 00..7F
e1b711da 120 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 121 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 122 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 123 U+D000..U+D7FF ED 80..9F 80..BF
37e2e78e 124 U+D800..U+DFFF +++++++ utf16 surrogates, not legal utf8 +++++++
375122d7 125 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 126 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH		 127 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
128 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
b2635aa8
KW		 129 Below are non-Unicode code points
130 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF
131 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF
132 U+200000: F8.. * 88..BF 80..BF 80..BF 80..BF
877d9f0d 133
e1b711da 134Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW		 135caused by legal UTF-8 avoiding non-shortest encodings: it is technically
136possible to UTF-8-encode a single code point in different ways, but that is
137explicitly forbidden, and the shortest possible encoding should always be used
138(and that is what Perl does).
8c007b5a 139
877d9f0d
JH		 140 */
141
8c007b5a
JH		 142/*
143 Another way to look at it, as bits:
144
b2635aa8 145 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
8c007b5a 146
b2635aa8
KW		 147 0aaa aaaa 0aaa aaaa
148 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa
149 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa
150 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa
8c007b5a
JH		 151
152As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 153leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH		 154encoded character.
155
65ab9279
TC		 156Perl's extended UTF-8 means we can have start bytes up to FF.
157
8c007b5a
JH		 158*/
159
c4d5f83a 160#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
0447e8df 161#define UTF8_IS_START(c) (((U8)c) >= 0xc2)
c512ce4f
JH		 162#define UTF8_IS_CONTINUATION(c) (((U8)c) >= 0x80 && (((U8)c) <= 0xbf))
163#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
0ae1fa71
CH		 164
165/* Masking with 0xfe allows low bit to be 0 or 1; thus this matches 0xc[23] */
166#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)c & 0xfe) == 0xc2)
8850bf83 167
22901f30 168#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len))))
b2635aa8
KW		 169
170/* Masks out the initial one bits in a start byte, leaving the real data ones.
171 * Doesn't work on an invariant byte */
22901f30 172#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6
NIS		 173
174#define UTF_CONTINUATION_MARK 0x80
175#define UTF_ACCUMULATION_SHIFT 6
b2635aa8
KW		 176
177/* 2**UTF_ACCUMULATION_SHIFT - 1 */
1d72bdf6 178#define UTF_CONTINUATION_MASK ((U8)0x3f)
c512ce4f 179
eb83ed87
KW		 180/* This sets the UTF_CONTINUATION_MASK in the upper bits of a word. If a value
181 * is anded with it, and the result is non-zero, then using the original value
182 * in UTF8_ACCUMULATE will overflow, shifting bits off the left */
183#define UTF_ACCUMULATION_OVERFLOW_MASK \
184 (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) - UTF_ACCUMULATION_SHIFT))
185
1d68d6cd 186#ifdef HAS_QUAD
5bbb0b5a 187#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC		 188 (uv) < 0x800 ? 2 : \
189 (uv) < 0x10000 ? 3 : \
190 (uv) < 0x200000 ? 4 : \
191 (uv) < 0x4000000 ? 5 : \
192 (uv) < 0x80000000 ? 6 : \
9041c2e3 193 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC		 194#else
195/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5bbb0b5a 196#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC		 197 (uv) < 0x800 ? 2 : \
198 (uv) < 0x10000 ? 3 : \
199 (uv) < 0x200000 ? 4 : \
200 (uv) < 0x4000000 ? 5 : \
201 (uv) < 0x80000000 ? 6 : 7 )
202#endif
203
d06134e5
KW		 204#endif /* EBCDIC vs ASCII */
205
206/* Rest of these are attributes of Unicode and perl's internals rather than the
207 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
208 * this level; the macros that some of these call may have different
209 * definitions in the two encodings */
210
211#define NATIVE8_TO_UNI(ch) NATIVE_TO_ASCII(ch) /* a clearer synonym */
212
213#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) | (((U8)new) & UTF_CONTINUATION_MASK))
214
2950f2a7
KW		 215/* Convert a two (not one) byte utf8 character to a unicode code point value.
216 * Needs just one iteration of accumulate. Should not be used unless it is
217 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
218 * Note that the result can be larger than 255 if the input character is not
219 * downgradable */
220#define TWO_BYTE_UTF8_TO_UNI(HI, LO) \
221 UTF8_ACCUMULATE((NATIVE_TO_UTF(HI) & UTF_START_MASK(2)), \
222 NATIVE_TO_UTF(LO))
223
d06134e5
KW		 224#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
225
226#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c))
227#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE8_TO_UNI(c))
228
229#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
230
231/* The macros in the next sets are used to generate the two utf8 or utfebcdic
232 * bytes from an ordinal that is known to fit into two bytes; it must be less
233 * than 0x3FF to work across both encodings. */
234/* Nocast allows these to be used in the case label of a switch statement */
0072151e 235#define UTF8_TWO_BYTE_HI_nocast(c) UTF_TO_NATIVE(((c) >> UTF_ACCUMULATION_SHIFT) | (0xFF & UTF_START_MARK(2)))
14c28784 236#define UTF8_TWO_BYTE_LO_nocast(c) UTF_TO_NATIVE(((c) & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK)
d06134e5
KW		 237
238#define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c)))
239#define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c)))
240
241/* This name is used when the source is a single byte */
242#define UTF8_EIGHT_BIT_HI(c) UTF8_TWO_BYTE_HI((U8)(c))
243#define UTF8_EIGHT_BIT_LO(c) UTF8_TWO_BYTE_LO((U8)(c))
244
7e2040f0 245/*
e3036cf4 246 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
20df05f4
KW		 247 * from an earlier version of these macros in which they didn't call the
248 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
249 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
250 * no need to do it again here
7e2040f0 251 */
e3036cf4
KW		 252#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \
253 ? isIDFIRST(*(p)) \
254 : isIDFIRST_utf8((const U8*)p))
255#define isALNUM_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \
256 ? isALNUM(*(p)) \
257 : isALNUM_utf8((const U8*)p))
1d72bdf6 258
7e2040f0
GS		 259#define isIDFIRST_lazy(p) isIDFIRST_lazy_if(p,1)
260#define isALNUM_lazy(p) isALNUM_lazy_if(p,1)
3bd709b1 261
89ebb4a3
JH		 262#define UTF8_MAXBYTES 13
263/* How wide can a single UTF-8 encoded character become in bytes.
264 * NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8
265 * since UTF-8 is an encoding of Unicode and given Unicode's current
266 * upper limit only four bytes is possible. Perl thinks of UTF-8
267 * as a way to encode non-negative integers in a binary format. */
268#define UTF8_MAXLEN UTF8_MAXBYTES
269
89ebb4a3
JH		 270/* The maximum number of UTF-8 bytes a single Unicode character can
271 * uppercase/lowercase/fold into; this number depends on the Unicode
272 * version. An example of maximal expansion is the U+03B0 which
273 * uppercases to U+03C5 U+0308 U+0301. The Unicode databases that
d06134e5 274 * tell these things are UnicodeData.txt, CaseFolding.txt, and
88d45ddf
KW		 275 * SpecialCasing.txt. The value is 6 for strict Unicode characters, but it has
276 * to be as big as Perl allows for a single character */
277#define UTF8_MAXBYTES_CASE UTF8_MAXBYTES
3bd709b1 278
8cb75cc8
KW		 279/* A Unicode character can fold to up to 3 characters */
280#define UTF8_MAX_FOLD_CHAR_EXPAND 3
281
a98fe34d 282#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 283#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
66cbab2c 284#define IN_UNI_8_BIT \
b36bf33f
KW		 285 (CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT|HINT_LOCALE_NOT_CHARS) \
286 && ! IN_LOCALE_RUNTIME && ! IN_BYTES)
287
1d72bdf6 288
c76687c5
KW		 289#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
290
291/* Allow first byte to be a continuation byte */
1d72bdf6 292#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW		 293
294/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 295#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW		 296
297/* expecting more bytes than were available in the string */
298#define UTF8_ALLOW_SHORT 0x0008
299
300/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
301#define UTF8_ALLOW_LONG 0x0010
302
303#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
304#define UTF8_WARN_SURROGATE 0x0040
305
306#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
307#define UTF8_WARN_NONCHAR 0x0100 /* code points */
308
309#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
310#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
311
312/* Code points which never were part of the original UTF-8 standard, the first
313 * byte of which is a FE or FF on ASCII platforms. */
314#define UTF8_DISALLOW_FE_FF 0x0800
315#define UTF8_WARN_FE_FF 0x1000
316
317#define UTF8_CHECK_ONLY 0x2000
318
319/* For backwards source compatibility. They do nothing, as the default now
320 * includes what they used to mean. The first one's meaning was to allow the
321 * just the single non-character 0xFFFF */
322#define UTF8_ALLOW_FFFF 0
323#define UTF8_ALLOW_SURROGATE 0
324
33d9abfb 325#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
949cf498 326#define UTF8_WARN_ILLEGAL_INTERCHANGE \
33d9abfb 327 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
949cf498
KW		 328#define UTF8_ALLOW_ANY \
329 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
330#define UTF8_ALLOW_ANYUV \
331 (UTF8_ALLOW_EMPTY \
332 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
TS		 333#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
334 UTF8_ALLOW_ANYUV)
1d72bdf6 335
7131f24d
KW		 336/* Surrogates, non-character code points and above-Unicode code points are
337 * problematic in some contexts. This allows code that needs to check for
338 * those to to quickly exclude the vast majority of code points it will
339 * encounter */
340#ifdef EBCDIC
341# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE UTF_TO_NATIVE(0xF1)
342#else
343# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE 0xED
344#endif
345
346/* ASCII EBCDIC I8
347 * U+D7FF: \xED\x9F\xBF \xF1\xB5\xBF\xBF last before surrogates
348 * U+D800: \xED\xA0\x80 \xF1\xB6\xA0\xA0 1st surrogate
349 * U+DFFF: \xED\xBF\xBF \xF1\xB7\xBF\xBF final surrogate
350 * U+E000: \xEE\x80\x80 \xF1\xB8\xA0\xA0 next after surrogates
351 */
352#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
b2635aa8
KW		 353# define UTF8_IS_SURROGATE(s) (*(s) == UTF_TO_NATIVE(0xF1) \
354 && ((*((s) +1) == UTF_TO_NATIVE(0xB6)) \
355 || *((s) + 1) == UTF_TO_NATIVE(0xB7)))
28936164
KW		 356 /* <send> points to one beyond the end of the string that starts at <s> */
357# define UTF8_IS_REPLACEMENT(s, send) (*(s) == UTF_TO_NATIVE(0xEF) \
358 && (send - s) >= 4 \
359 && *((s) + 1) == UTF_TO_NATIVE(0xBF) \
360 && *((s) + 2) == UTF_TO_NATIVE(0xBF) \
361 && *((s) + 3) == UTF_TO_NATIVE(0xBD)
7131f24d
KW		 362#else
363# define UTF8_IS_SURROGATE(s) (*(s) == 0xED && *((s) + 1) >= 0xA0)
28936164
KW		 364# define UTF8_IS_REPLACEMENT(s, send) (*(s) == 0xEF \
365 && (send - s) >= 3 \
366 && *((s) + 1) == 0xBF \
367 && *((s) + 2) == 0xBD)
7131f24d
KW		 368#endif
369
370/* ASCII EBCDIC I8
371 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
372 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
373 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
374 */
375#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
376# define UTF8_IS_SUPER(s) (*(s) >= UTF_TO_NATIVE(0xF9) \
377 && (*(s) > UTF_TO_NATIVE(0xF9) || (*((s) + 1) >= UTF_TO_NATIVE(0xA2))))
378#else
379# define UTF8_IS_SUPER(s) (*(s) >= 0xF4 \
380 && (*(s) > 0xF4 || (*((s) + 1) >= 0x90)))
381#endif
382
383/* ASCII EBCDIC I8
384 * U+FDCF: \xEF\xB7\x8F \xF1\xBF\xAE\xAF last before non-char block
385 * U+FDD0: \xEF\xB7\x90 \xF1\xBF\xAE\xB0 first non-char in block
386 * U+FDEF: \xEF\xB7\xAF \xF1\xBF\xAF\xAF last non-char in block
387 * U+FDF0: \xEF\xB7\xB0 \xF1\xBF\xAF\xB0 first after non-char block
388 * U+FFFF: \xEF\xBF\xBF \xF1\xBF\xBF\xBF
389 * U+1FFFF: \xF0\x9F\xBF\xBF \xF3\xBF\xBF\xBF
390 * U+2FFFF: \xF0\xAF\xBF\xBF \xF5\xBF\xBF\xBF
391 * U+3FFFF: \xF0\xBF\xBF\xBF \xF7\xBF\xBF\xBF
392 * U+4FFFF: \xF1\x8F\xBF\xBF \xF8\xA9\xBF\xBF\xBF
393 * U+5FFFF: \xF1\x9F\xBF\xBF \xF8\xAB\xBF\xBF\xBF
394 * U+6FFFF: \xF1\xAF\xBF\xBF \xF8\xAD\xBF\xBF\xBF
395 * U+7FFFF: \xF1\xBF\xBF\xBF \xF8\xAF\xBF\xBF\xBF
396 * U+8FFFF: \xF2\x8F\xBF\xBF \xF8\xB1\xBF\xBF\xBF
397 * U+9FFFF: \xF2\x9F\xBF\xBF \xF8\xB3\xBF\xBF\xBF
398 * U+AFFFF: \xF2\xAF\xBF\xBF \xF8\xB5\xBF\xBF\xBF
399 * U+BFFFF: \xF2\xBF\xBF\xBF \xF8\xB7\xBF\xBF\xBF
400 * U+CFFFF: \xF3\x8F\xBF\xBF \xF8\xB9\xBF\xBF\xBF
401 * U+DFFFF: \xF3\x9F\xBF\xBF \xF8\xBB\xBF\xBF\xBF
402 * U+EFFFF: \xF3\xAF\xBF\xBF \xF8\xBD\xBF\xBF\xBF
403 * U+FFFFF: \xF3\xBF\xBF\xBF \xF8\xBF\xBF\xBF\xBF
404 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF
405 */
406#define UTF8_IS_NONCHAR_(s) ( \
407 *(s) >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
408 && ! UTF8_IS_SUPER(s) \
409 && UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_FIRST_PROBLEMATIC(s) \
410
411#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
412# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
413 ((*(s) == UTF_TO_NATIVE(0xF1) \
414 && (*((s) + 1) == UTF_TO_NATIVE(0xBF) \
415 && ((*((s) + 2) == UTF_TO_NATIVE(0xAE) \
416 && *((s) + 3) >= UTF_TO_NATIVE(0xB0)) \
417 || (*((s) + 2) == UTF_TO_NATIVE(0xAF) \
418 && *((s) + 3) <= UTF_TO_NATIVE(0xAF))))) \
419 || (UTF8SKIP(*(s)) > 3 \
420 /* (These were all derived by inspection and experimentation with an */ \
421 /* editor) The next line checks the next to final byte in the char */ \
422 && *((s) + UTF8SKIP(*(s)) - 2) == UTF_TO_NATIVE(0xBF) \
423 && *((s) + UTF8SKIP(*(s)) - 3) == UTF_TO_NATIVE(0xBF) \
424 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 4)) & 0x81) == 0x81 \
425 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 1)) & 0xBE) == 0XBE))
426#else
427# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
428 ((*(s) == 0xEF \
429 && ((*((s) + 1) == 0xB7 && (*((s) + 2) >= 0x90 && (*((s) + 2) <= 0xAF)))\
430 /* Gets U+FFF[EF] */ \
431 || (*((s) + 1) == 0xBF && ((*((s) + 2) & 0xBE) == 0xBE)))) \
432 || ((*((s) + 2) == 0xBF \
433 && (*((s) + 3) & 0xBE) == 0xBE \
434 /* Excludes things like U+10FFE = \xF0\x90\xBF\xBE */ \
435 && (*((s) + 1) & 0x8F) == 0x8F)))
436#endif
437
c867b360
JH		 438#define UNICODE_SURROGATE_FIRST 0xD800
439#define UNICODE_SURROGATE_LAST 0xDFFF
440#define UNICODE_REPLACEMENT 0xFFFD
441#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 442
b851fbc1 443/* Though our UTF-8 encoding can go beyond this,
c76687c5 444 * let's be conservative and do as Unicode says. */
b851fbc1
JH		 445#define PERL_UNICODE_MAX 0x10FFFF
446
949cf498
KW		 447#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
448#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
449#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
450#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
451#define UNICODE_DISALLOW_SURROGATE 0x0010
452#define UNICODE_DISALLOW_NONCHAR 0x0020
453#define UNICODE_DISALLOW_SUPER 0x0040
454#define UNICODE_DISALLOW_FE_FF 0x0080
455#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
456 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
457#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
458 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
459
460/* For backward source compatibility, as are now the default */
461#define UNICODE_ALLOW_SURROGATE 0
462#define UNICODE_ALLOW_SUPER 0
463#define UNICODE_ALLOW_ANY 0
b851fbc1 464
1d72bdf6
NIS		 465#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
466 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 467#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 468#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW		 469#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
470 /* The other noncharacters end in FFFE or FFFF, which \
471 * the mask below catches both of, but beyond the last \
472 * official unicode code point, they aren't \
473 * noncharacters, since those aren't Unicode \
474 * characters at all */ \
475 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
476#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
477#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
1d72bdf6 478
f067b878
NA		 479#ifdef HAS_QUAD
480# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
481#endif
3bd709b1 482
09091399
JH		 483#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
484#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
485#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 486#define GREEK_SMALL_LETTER_MU 0x03BC
78a0d3cc
KW		 487#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */
488#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
97298f37 489#define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
8d64d87f
KW		 490#define LATIN_SMALL_LETTER_LONG_S 0x017F
491#define KELVIN_SIGN 0x212A
492#define ANGSTROM_SIGN 0x212B
09091399 493
9e55ce06 494#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH		 495#define UNI_DISPLAY_BACKSLASH 0x0002
496#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
497#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 498
5cd46e1f
KW		 499#ifndef EBCDIC
500# define LATIN_SMALL_LETTER_SHARP_S 0x00DF
501# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS 0x00FF
502# define MICRO_SIGN 0x00B5
f508a607
KW		 503# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE 0x00C5
504# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE 0x00E5
5cd46e1f
KW		 505#endif
506
507#define ANYOF_FOLD_SHARP_S(node, input, end) \
508 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 509 (ANYOF_NONBITMAP(node)) && \
39065660 510 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f
JH		 511 ((end) > (input) + 1) && \
512 toLOWER((input)[0]) == 's' && \
513 toLOWER((input)[1]) == 's')
ebc501f0 514#define SHARP_S_SKIP 2
3b0fc154 515
2f454f11 516#ifndef EBCDIC
a4f7a67c
KW		 517/* If you want to exclude surrogates, and beyond legal Unicode, see the blame
518 * log for earlier versions which gave details for these */
c78f6c49 519# define IS_UTF8_CHAR_1(p) \
3b0fc154 520 ((p)[0] <= 0x7F)
c78f6c49 521# define IS_UTF8_CHAR_2(p) \
3b0fc154
JH		 522 ((p)[0] >= 0xC2 && (p)[0] <= 0xDF && \
523 (p)[1] >= 0x80 && (p)[1] <= 0xBF)
c78f6c49 524# define IS_UTF8_CHAR_3a(p) \
3b0fc154
JH		 525 ((p)[0] == 0xE0 && \
526 (p)[1] >= 0xA0 && (p)[1] <= 0xBF && \
527 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49 528# define IS_UTF8_CHAR_3b(p) \
a4f7a67c 529 ((p)[0] >= 0xE1 && (p)[0] <= 0xEF && \
3b0fc154
JH		 530 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
531 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49 532# define IS_UTF8_CHAR_4a(p) \
3b0fc154
JH		 533 ((p)[0] == 0xF0 && \
534 (p)[1] >= 0x90 && (p)[1] <= 0xBF && \
535 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
536 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
a4f7a67c 537/* The 0xF7 allows us to go to 0x1fffff (0x200000 would
3b0fc154
JH		 538 * require five bytes). Not doing any further code points
539 * since that is not needed (and that would not be strict
540 * UTF-8, anyway). The "slow path" in Perl_is_utf8_char()
541 * will take care of the "extended UTF-8". */
a4f7a67c
KW		 542# define IS_UTF8_CHAR_4b(p) \
543 ((p)[0] >= 0xF1 && (p)[0] <= 0xF7 && \
3b0fc154
JH		 544 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
545 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
546 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
547
c78f6c49 548# define IS_UTF8_CHAR_3(p) \
3b0fc154 549 (IS_UTF8_CHAR_3a(p) || \
a4f7a67c 550 IS_UTF8_CHAR_3b(p))
c78f6c49 551# define IS_UTF8_CHAR_4(p) \
3b0fc154 552 (IS_UTF8_CHAR_4a(p) || \
a4f7a67c 553 IS_UTF8_CHAR_4b(p))
3b0fc154
JH		 554
555/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
556 * (1) allows UTF-8 encoded UTF-16 surrogates
557 * (2) it allows code points past U+10FFFF.
558 * The Perl_is_utf8_char() full "slow" code will handle the Perl
559 * "extended UTF-8". */
c78f6c49 560# define IS_UTF8_CHAR(p, n) \
3b0fc154
JH		 561 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
562 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
563 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
564 (n) == 4 ? IS_UTF8_CHAR_4(p) : 0)
565
c78f6c49 566# define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
768c67ee 567
2f454f11
KW		 568#else /* EBCDIC */
569
570/* This is an attempt to port IS_UTF8_CHAR to EBCDIC based on eyeballing.
571 * untested. If want to exclude surrogates and above-Unicode, see the
572 * definitions for UTF8_IS_SURROGATE and UTF8_IS_SUPER */
573# define IS_UTF8_CHAR_1(p) \
574 (NATIVE_TO_ASCII((p)[0]) <= 0x9F)
575# define IS_UTF8_CHAR_2(p) \
576 (NATIVE_TO_I8((p)[0]) >= 0xC5 && NATIVE_TO_I8((p)[0]) <= 0xDF && \
577 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF)
578# define IS_UTF8_CHAR_3(p) \
579 (NATIVE_TO_I8((p)[0]) == 0xE1 && NATIVE_TO_I8((p)[1]) <= 0xEF && \
580 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
581 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF)
582# define IS_UTF8_CHAR_4a(p) \
583 (NATIVE_TO_I8((p)[0]) == 0xF0 && \
584 NATIVE_TO_I8((p)[1]) >= 0xB0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
585 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
586 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
587# define IS_UTF8_CHAR_4b(p) \
588 (NATIVE_TO_I8((p)[0]) >= 0xF1 && NATIVE_TO_I8((p)[0]) <= 0xF7 && \
589 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
590 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
591 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
592# define IS_UTF8_CHAR_5a(p) \
593 (NATIVE_TO_I8((p)[0]) == 0xF8 && \
594 NATIVE_TO_I8((p)[1]) >= 0xA8 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
595 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
596 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
597 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
598# define IS_UTF8_CHAR_5b(p) \
599 (NATIVE_TO_I8((p)[0]) >= 0xF9 && NATIVE_TO_I8((p)[1]) <= 0xFB && \
600 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
601 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
602 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
603 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
604
605# define IS_UTF8_CHAR_4(p) \
606 (IS_UTF8_CHAR_4a(p) || \
607 IS_UTF8_CHAR_4b(p))
608# define IS_UTF8_CHAR_5(p) \
609 (IS_UTF8_CHAR_5a(p) || \
610 IS_UTF8_CHAR_5b(p))
611# define IS_UTF8_CHAR(p, n) \
612 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
613 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
614 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
615 (n) == 4 ? IS_UTF8_CHAR_4(p) : \
616 (n) == 5 ? IS_UTF8_CHAR_5(p) : 0)
617
618# define IS_UTF8_CHAR_FAST(n) ((n) <= 5)
619
77263263 620#endif /* IS_UTF8_CHAR() for UTF-8 */
e9a8c099
MHM		 621
622/*
623 * Local variables:
624 * c-indentation-style: bsd
625 * c-basic-offset: 4
14d04a33 626 * indent-tabs-mode: nil
e9a8c099
MHM		 627 * End:
628 *
14d04a33 629 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 630 */
Unnamed repository; edit this file 'description' to name the repository.