This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
lib/warnings/utf8: Add missing tests
[perl5.git] / utf8.h
CommitLineData
a0ed51b3
LW
1/* utf8.h
2 *
f3cb6f94
KW
3 * This file contains definitions for use with the UTF-8 encoding. It
4 * actually also works with the variant UTF-8 encoding called UTF-EBCDIC, and
5 * hides almost all of the differences between these from the caller. In other
6 * words, someone should #include this file, and if the code is being compiled
7 * on an EBCDIC platform, things should mostly just work.
8 *
2eee27d7
SS
9 * Copyright (C) 2000, 2001, 2002, 2005, 2006, 2007, 2009,
10 * 2010, 2011 by Larry Wall and others
a0ed51b3
LW
11 *
12 * You may distribute under the terms of either the GNU General Public
13 * License or the Artistic License, as specified in the README file.
14 *
15 */
16
57f0e7e2
KW
17#ifndef H_UTF8 /* Guard against recursive inclusion */
18#define H_UTF8 1
19
39e02b42 20/* Use UTF-8 as the default script encoding?
1e54db1a 21 * Turning this on will break scripts having non-UTF-8 binary
39e02b42
JH
22 * data (such as Latin-1) in string literals. */
23#ifdef USE_UTF8_SCRIPTS
24# define USE_UTF8_IN_NAMES (!IN_BYTES)
25#else
26# define USE_UTF8_IN_NAMES (PL_hints & HINT_UTF8)
27#endif
28
3cd96634
KW
29#include "regcharclass.h"
30#include "unicode_constants.h"
31
051a06d4
KW
32/* For to_utf8_fold_flags, q.v. */
33#define FOLD_FLAGS_LOCALE 0x1
34#define FOLD_FLAGS_FULL 0x2
a0270393 35#define FOLD_FLAGS_NOMIX_ASCII 0x4
051a06d4 36
83199d38
KW
37/* For _core_swash_init(), internal core use only */
38#define _CORE_SWASH_INIT_USER_DEFINED_PROPERTY 0x1
5d3d13d1 39#define _CORE_SWASH_INIT_RETURN_IF_UNDEF 0x2
87367d5f 40#define _CORE_SWASH_INIT_ACCEPT_INVLIST 0x4
83199d38 41
7bbfa158
KW
42/*
43=head1 Unicode Support
44
45=for apidoc is_ascii_string
46
47This is a misleadingly-named synonym for L</is_invariant_string>.
48On ASCII-ish platforms, the name isn't misleading: the ASCII-range characters
49are exactly the UTF-8 invariants. But EBCDIC machines have more invariants
50than just the ASCII characters, so C<is_invariant_string> is preferred.
51
52=cut
53*/
54#define is_ascii_string(s, len) is_invariant_string(s, len)
55
de69f3af
KW
56#define uvchr_to_utf8(a,b) uvchr_to_utf8_flags(a,b,0)
57#define uvchr_to_utf8_flags(d,uv,flags) \
58 uvoffuni_to_utf8_flags(d,NATIVE_TO_UNI(uv),flags)
59#define utf8_to_uvchr_buf(s, e, lenp) \
842991ae 60 utf8n_to_uvchr(s, (U8*)(e) - (U8*)(s), lenp, \
de69f3af
KW
61 ckWARN_d(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY)
62
a0270393 63#define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
445bf929
KW
64#define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
65#define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0)
66#define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0)
67#define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0)
36bb2ab6 68
fd7cb289 69/* Source backward compatibility. */
fd7cb289
RGS
70#define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
71
eda9cac1
KW
72#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
73 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
baa60164 74#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
cea315b6 75#define FOLDEQ_LOCALE (1 << 1)
18f762c3
KW
76#define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
77#define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
a33c29bc 78
e6226b18
KW
79#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
80 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
81
1d72bdf6
NIS
82#ifdef EBCDIC
83/* The equivalent of these macros but implementing UTF-EBCDIC
84 are in the following header file:
85 */
86
87#include "utfebcdic.h"
fd7cb289 88
d06134e5 89#else /* ! EBCDIC */
73c4f7a1
GS
90START_EXTERN_C
91
a0ed51b3 92#ifdef DOINIT
6f06b55f 93EXTCONST unsigned char PL_utf8skip[] = {
b2635aa8
KW
94/* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
95/* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
96/* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
97/* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
98/* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
99/* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
100/* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
101/* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
102/* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
103/* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
104/* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
105/* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
106/* 0xC0 */ 2,2, /* overlong */
107/* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
108/* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */
109/* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */
110/* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */
111/* 0xFE */ 7,13, /* Perl extended (never was official UTF-8). Up to 72bit
112 allowed (64-bit + reserved). */
a0ed51b3
LW
113};
114#else
6f06b55f 115EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
116#endif
117
73c4f7a1 118END_EXTERN_C
7e2040f0 119
59a449d5
KW
120/* Native character to/from iso-8859-1. Are the identity functions on ASCII
121 * platforms */
122#define NATIVE_TO_LATIN1(ch) (ch)
123#define LATIN1_TO_NATIVE(ch) (ch)
124
125/* I8 is an intermediate version of UTF-8 used only in UTF-EBCDIC. We thus
126 * consider it to be identical to UTF-8 on ASCII platforms. Strictly speaking
127 * UTF-8 and UTF-EBCDIC are two different things, but we often conflate them
128 * because they are 8-bit encodings that serve the same purpose in Perl, and
129 * rarely do we need to distinguish them. The term "NATIVE_UTF8" applies to
130 * whichever one is applicable on the current platform */
131#define NATIVE_UTF8_TO_I8(ch) (ch)
132#define I8_TO_NATIVE_UTF8(ch) (ch)
133
1d72bdf6
NIS
134/* Transforms in wide UV chars */
135#define UNI_TO_NATIVE(ch) (ch)
136#define NATIVE_TO_UNI(ch) (ch)
d7578b48 137
877d9f0d 138/*
9041c2e3 139
8c007b5a 140 The following table is from Unicode 3.2.
877d9f0d
JH
141
142 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
143
375122d7 144 U+0000..U+007F 00..7F
e1b711da 145 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 146 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 147 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 148 U+D000..U+D7FF ED 80..9F 80..BF
537124e4 149 U+D800..U+DFFF ED A0..BF 80..BF (surrogates)
375122d7 150 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 151 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH
152 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
153 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
b2635aa8
KW
154 Below are non-Unicode code points
155 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF
156 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF
537124e4 157 U+200000..: F8.. * 88..BF 80..BF 80..BF 80..BF
877d9f0d 158
e1b711da 159Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
160caused by legal UTF-8 avoiding non-shortest encodings: it is technically
161possible to UTF-8-encode a single code point in different ways, but that is
162explicitly forbidden, and the shortest possible encoding should always be used
15824458 163(and that is what Perl does). The non-shortest ones are called 'overlongs'.
8c007b5a 164
877d9f0d
JH
165 */
166
8c007b5a
JH
167/*
168 Another way to look at it, as bits:
169
b2635aa8 170 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
8c007b5a 171
b2635aa8
KW
172 0aaa aaaa 0aaa aaaa
173 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa
174 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa
175 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa
8c007b5a
JH
176
177As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 178leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
179encoded character.
180
65ab9279
TC
181Perl's extended UTF-8 means we can have start bytes up to FF.
182
8c007b5a
JH
183*/
184
15824458
KW
185/* Is the representation of the Unicode code point 'c' the same regardless of
186 * being encoded in UTF-8 or not? */
c4d5f83a 187#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
15824458
KW
188
189/* Is the UTF8-encoded byte 'c' part of a variant sequence in UTF-8? This is
190 * the inverse of UTF8_IS_INVARIANT */
191#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
192
193/* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence?
194 * This doesn't catch invariants (they are single-byte). It also excludes the
195 * illegal overlong sequences that begin with C0 and C1. */
0447e8df 196#define UTF8_IS_START(c) (((U8)c) >= 0xc2)
15824458
KW
197
198/* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the
199 * first byte thereof? */
e021c6e6 200#define UTF8_IS_CONTINUATION(c) ((((U8)c) & 0xC0) == 0x80)
0ae1fa71 201
15824458
KW
202/* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use
203 * UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to
204 * be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus
205 * this matches 0xc[23]. */
559c7f10 206#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)(c) & 0xfe) == 0xc2)
4ab10950 207
15824458
KW
208/* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that
209 * represent a code point > 255? */
7b4252f4 210#define UTF8_IS_ABOVE_LATIN1(c) ((U8)(c) >= 0xc4)
8850bf83 211
15824458
KW
212/* This defines the 1-bits that are to be in the first byte of a multi-byte
213 * UTF-8 encoded character that give the number of bytes that comprise the
f710bf0c 214 * character. 'len' is the number of bytes in the multi-byte sequence. */
ee372ee9 215#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFF & (0xFE << (7-(len)))))
b2635aa8
KW
216
217/* Masks out the initial one bits in a start byte, leaving the real data ones.
f710bf0c
KW
218 * Doesn't work on an invariant byte. 'len' is the number of bytes in the
219 * multi-byte sequence that comprises the character. */
22901f30 220#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6 221
15824458
KW
222/* This defines the bits that are to be in the continuation bytes of a multi-byte
223 * UTF-8 encoded character that indicate it is a continuation byte. */
1d72bdf6 224#define UTF_CONTINUATION_MARK 0x80
15824458
KW
225
226/* This is the number of low-order bits a continuation byte in a UTF-8 encoded
227 * sequence contributes to the specification of the code point. In the bit
228 * maps above, you see that the first 2 bits are a constant '10', leaving 6 of
229 * real information */
1d72bdf6 230#define UTF_ACCUMULATION_SHIFT 6
b2635aa8
KW
231
232/* 2**UTF_ACCUMULATION_SHIFT - 1 */
1d72bdf6 233#define UTF_CONTINUATION_MASK ((U8)0x3f)
c512ce4f 234
15824458
KW
235/* If a value is anded with this, and the result is non-zero, then using the
236 * original value in UTF8_ACCUMULATE will overflow, shifting bits off the left
237 * */
eb83ed87 238#define UTF_ACCUMULATION_OVERFLOW_MASK \
bb88be5f
KW
239 (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) \
240 - UTF_ACCUMULATION_SHIFT))
eb83ed87 241
6588300d
NC
242#if UVSIZE >= 8
243# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
244
5aaebcb3
KW
245/* Input is a true Unicode (not-native) code point */
246#define OFFUNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
247 (uv) < 0x800 ? 2 : \
248 (uv) < 0x10000 ? 3 : \
249 (uv) < 0x200000 ? 4 : \
250 (uv) < 0x4000000 ? 5 : \
251 (uv) < 0x80000000 ? 6 : \
9041c2e3 252 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC
253#else
254/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5aaebcb3 255#define OFFUNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
256 (uv) < 0x800 ? 2 : \
257 (uv) < 0x10000 ? 3 : \
258 (uv) < 0x200000 ? 4 : \
259 (uv) < 0x4000000 ? 5 : \
260 (uv) < 0x80000000 ? 6 : 7 )
261#endif
262
03c76984
KW
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 since UTF-8
265 * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be
266 * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode
267 * non-negative integers in a binary format, even those above Unicode */
268#define UTF8_MAXBYTES 13
269
270/* The maximum number of UTF-8 bytes a single Unicode character can
271 * uppercase/lowercase/fold into. Unicode guarantees that the maximum
272 * expansion is 3 characters. On ASCIIish platforms, the highest Unicode
273 * character occupies 4 bytes, therefore this number would be 12, but this is
274 * smaller than the maximum width a single above-Unicode character can occupy,
275 * so use that instead */
276#if UTF8_MAXBYTES < 12
277#error UTF8_MAXBYTES must be at least 12
278#endif
279
0ed2b00b
KW
280/* ^? is defined to be DEL on ASCII systems. See the definition of toCTRL()
281 * for more */
282#define QUESTION_MARK_CTRL DEL_NATIVE
283
843a4590
KW
284#define MAX_UTF8_TWO_BYTE 0x7FF
285
03c76984
KW
286#define UTF8_MAXBYTES_CASE UTF8_MAXBYTES
287
d06134e5
KW
288#endif /* EBCDIC vs ASCII */
289
290/* Rest of these are attributes of Unicode and perl's internals rather than the
291 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
292 * this level; the macros that some of these call may have different
293 * definitions in the two encodings */
294
59a449d5
KW
295/* In domain restricted to ASCII, these may make more sense to the reader than
296 * the ones with Latin1 in the name */
297#define NATIVE_TO_ASCII(ch) NATIVE_TO_LATIN1(ch)
298#define ASCII_TO_NATIVE(ch) LATIN1_TO_NATIVE(ch)
299
300/* More or less misleadingly-named defines, retained for back compat */
301#define NATIVE_TO_UTF(ch) NATIVE_UTF8_TO_I8(ch)
302#define NATIVE_TO_I8(ch) NATIVE_UTF8_TO_I8(ch)
303#define UTF_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
304#define I8_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
305#define NATIVE8_TO_UNI(ch) NATIVE_TO_LATIN1(ch)
d06134e5 306
537124e4
KW
307/* Adds a UTF8 continuation byte 'new' of information to a running total code
308 * point 'old' of all the continuation bytes so far. This is designed to be
155d2738
KW
309 * used in a loop to convert from UTF-8 to the code point represented. Note
310 * that this is asymmetric on EBCDIC platforms, in that the 'new' parameter is
311 * the UTF-EBCDIC byte, whereas the 'old' parameter is a Unicode (not EBCDIC)
312 * code point in process of being generated */
313#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) \
314 | ((NATIVE_UTF8_TO_I8((U8)new)) \
315 & UTF_CONTINUATION_MASK))
d06134e5 316
4ab10950
KW
317/* This works in the face of malformed UTF-8. */
318#define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) (UTF8_IS_DOWNGRADEABLE_START(*s) \
319 && ( (e) - (s) > 1) \
320 && UTF8_IS_CONTINUATION(*((s)+1)))
321
5aaebcb3
KW
322/* Number of bytes a code point occupies in UTF-8. */
323#define NATIVE_SKIP(uv) OFFUNISKIP(NATIVE_TO_UNI(uv))
324
325/* Most code which says UNISKIP is really thinking in terms of native code
326 * points (0-255) plus all those beyond. This is an imprecise term, but having
327 * it means existing code continues to work. For precision, use NATIVE_SKIP
328 * and OFFUNISKIP */
329#define UNISKIP(uv) NATIVE_SKIP(uv)
330
94bb8c36 331/* Convert a two (not one) byte utf8 character to a native code point value.
2950f2a7
KW
332 * Needs just one iteration of accumulate. Should not be used unless it is
333 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
334 * Note that the result can be larger than 255 if the input character is not
335 * downgradable */
94bb8c36
KW
336#define TWO_BYTE_UTF8_TO_NATIVE(HI, LO) \
337 UNI_TO_NATIVE(UTF8_ACCUMULATE((NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), \
155d2738 338 (LO)))
94bb8c36
KW
339
340/* Should never be used, and be deprecated */
341#define TWO_BYTE_UTF8_TO_UNI(HI, LO) NATIVE_TO_UNI(TWO_BYTE_UTF8_TO_NATIVE(HI, LO))
2950f2a7 342
15824458
KW
343/* How many bytes in the UTF-8 encoded character whose first (perhaps only)
344 * byte is pointed to by 's' */
d06134e5
KW
345#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
346
15824458
KW
347/* Is the byte 'c' the same character when encoded in UTF-8 as when not. This
348 * works on both UTF-8 encoded strings and non-encoded, as it returns TRUE in
349 * each for the exact same set of bit patterns. (And it works on any byte in a
350 * UTF-8 encoded string) */
bc3632a8 351#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_UTF8_TO_I8(c))
15824458 352
6f2d5cbc
KW
353/* Like the above, but its name implies a non-UTF8 input, and is implemented
354 * differently (for no particular reason) */
355#define NATIVE_BYTE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_LATIN1(c))
356
357/* Like the above, but accepts any UV as input */
358#define UVCHR_IS_INVARIANT(uv) UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))
d06134e5
KW
359
360#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
361
48ccf5e1
KW
362/* The macros in the next 4 sets are used to generate the two utf8 or utfebcdic
363 * bytes from an ordinal that is known to fit into exactly two (not one) bytes;
364 * it must be less than 0x3FF to work across both encodings. */
365
366/* These two are helper macros for the other three sets, and should not be used
367 * directly anywhere else. 'translate_function' is either NATIVE_TO_LATIN1
368 * (which works for code points up to 0xFF) or NATIVE_TO_UNI which works for any
369 * code point */
370#define __BASE_TWO_BYTE_HI(c, translate_function) \
371 I8_TO_NATIVE_UTF8((translate_function(c) >> UTF_ACCUMULATION_SHIFT) \
372 | UTF_START_MARK(2))
373#define __BASE_TWO_BYTE_LO(c, translate_function) \
374 I8_TO_NATIVE_UTF8((translate_function(c) & UTF_CONTINUATION_MASK) \
375 | UTF_CONTINUATION_MARK)
376
48ccf5e1
KW
377/* The next two macros should not be used. They were designed to be usable as
378 * the case label of a switch statement, but this doesn't work for EBCDIC. Use
9d0d3a03 379 * regen/unicode_constants.pl instead */
48ccf5e1
KW
380#define UTF8_TWO_BYTE_HI_nocast(c) __BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI)
381#define UTF8_TWO_BYTE_LO_nocast(c) __BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI)
382
383/* The next two macros are used when the source should be a single byte
384 * character; checked for under DEBUGGING */
385#define UTF8_EIGHT_BIT_HI(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
386 ((U8) __BASE_TWO_BYTE_HI(c, NATIVE_TO_LATIN1)))
387#define UTF8_EIGHT_BIT_LO(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
388 ((U8) __BASE_TWO_BYTE_LO(c, NATIVE_TO_LATIN1)))
389
390/* These final two macros in the series are used when the source can be any
391 * code point whose UTF-8 is known to occupy 2 bytes; they are less efficient
392 * than the EIGHT_BIT versions on EBCDIC platforms. We use the logical '~'
393 * operator instead of "<=" to avoid getting compiler warnings.
394 * MAX_PORTABLE_UTF8_TWO_BYTE should be exactly all one bits in the lower few
395 * places, so the ~ works */
396#define UTF8_TWO_BYTE_HI(c) \
397 (__ASSERT_((sizeof(c) == 1) \
398 || !(((WIDEST_UTYPE)(c)) & ~MAX_PORTABLE_UTF8_TWO_BYTE)) \
d9759938 399 ((U8) __BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI)))
48ccf5e1
KW
400#define UTF8_TWO_BYTE_LO(c) \
401 (__ASSERT_((sizeof(c) == 1) \
402 || !(((WIDEST_UTYPE)(c)) & ~MAX_PORTABLE_UTF8_TWO_BYTE)) \
d9759938 403 ((U8) __BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI)))
d06134e5 404
e7214ce8
KW
405/* This is illegal in any well-formed UTF-8 in both EBCDIC and ASCII
406 * as it is only in overlongs. */
407#define ILLEGAL_UTF8_BYTE I8_TO_NATIVE_UTF8(0xC1)
408
7e2040f0 409/*
e3036cf4 410 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
20df05f4
KW
411 * from an earlier version of these macros in which they didn't call the
412 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
413 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
414 * no need to do it again here
7e2040f0 415 */
e3036cf4
KW
416#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \
417 ? isIDFIRST(*(p)) \
418 : isIDFIRST_utf8((const U8*)p))
32636478
KW
419#define isWORDCHAR_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \
420 ? isWORDCHAR(*(p)) \
421 : isWORDCHAR_utf8((const U8*)p))
422#define isALNUM_lazy_if(p,UTF) isWORDCHAR_lazy_if(p,UTF)
1d72bdf6 423
89ebb4a3
JH
424#define UTF8_MAXLEN UTF8_MAXBYTES
425
8cb75cc8
KW
426/* A Unicode character can fold to up to 3 characters */
427#define UTF8_MAX_FOLD_CHAR_EXPAND 3
428
a98fe34d 429#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 430#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
66cbab2c 431#define IN_UNI_8_BIT \
d6ded950
KW
432 (((CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT)) \
433 || (CopHINTS_get(PL_curcop) & HINT_LOCALE_PARTIAL \
434 /* -1 below is for :not_characters */ \
435 && _is_in_locale_category(FALSE, -1))) \
436 && ! IN_BYTES)
b36bf33f 437
1d72bdf6 438
c76687c5
KW
439#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
440
441/* Allow first byte to be a continuation byte */
1d72bdf6 442#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
443
444/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 445#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
446
447/* expecting more bytes than were available in the string */
448#define UTF8_ALLOW_SHORT 0x0008
449
450/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
451#define UTF8_ALLOW_LONG 0x0010
452
453#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
454#define UTF8_WARN_SURROGATE 0x0040
455
456#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
457#define UTF8_WARN_NONCHAR 0x0100 /* code points */
458
459#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
460#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
461
462/* Code points which never were part of the original UTF-8 standard, the first
ea5ced44
KW
463 * byte of which is a FE or FF on ASCII platforms. If the first byte is FF, it
464 * will overflow a 32-bit word. If the first byte is FE, it will overflow a
465 * signed 32-bit word. */
949cf498
KW
466#define UTF8_DISALLOW_FE_FF 0x0800
467#define UTF8_WARN_FE_FF 0x1000
468
469#define UTF8_CHECK_ONLY 0x2000
470
471/* For backwards source compatibility. They do nothing, as the default now
472 * includes what they used to mean. The first one's meaning was to allow the
473 * just the single non-character 0xFFFF */
474#define UTF8_ALLOW_FFFF 0
475#define UTF8_ALLOW_SURROGATE 0
476
33d9abfb 477#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
949cf498 478#define UTF8_WARN_ILLEGAL_INTERCHANGE \
33d9abfb 479 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
949cf498
KW
480#define UTF8_ALLOW_ANY \
481 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
482#define UTF8_ALLOW_ANYUV \
483 (UTF8_ALLOW_EMPTY \
484 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
485#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
486 UTF8_ALLOW_ANYUV)
1d72bdf6 487
7131f24d
KW
488/* Surrogates, non-character code points and above-Unicode code points are
489 * problematic in some contexts. This allows code that needs to check for
490 * those to to quickly exclude the vast majority of code points it will
491 * encounter */
b96a92fb
KW
492#define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
493 FIRST_SURROGATE_UTF8_FIRST_BYTE
7131f24d 494
b96a92fb
KW
495#define UTF8_IS_SURROGATE(s) cBOOL(is_SURROGATE_utf8(s))
496#define UTF8_IS_REPLACEMENT(s, send) cBOOL(is_REPLACEMENT_utf8_safe(s,send))
7131f24d
KW
497
498/* ASCII EBCDIC I8
499 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
500 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
501 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
a1776718
KW
502 *
503 * BE AWARE that this test doesn't rule out malformed code points, in
504 * particular overlongs */
7131f24d 505#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
bc3632a8
KW
506# define UTF8_IS_SUPER(s) (NATIVE_UTF8_TO_I8(* (U8*) (s)) >= 0xF9 \
507 && (NATIVE_UTF8_TO_I8(* (U8*) (s)) > 0xF9 \
a3481822 508 || (NATIVE_UTF8_TO_I8(* ((U8*) (s) + 1)) >= 0xA2)))
7131f24d 509#else
0404c92a
KW
510# define UTF8_IS_SUPER(s) (*(U8*) (s) >= 0xF4 \
511 && (*(U8*) (s) > 0xF4 || (*((U8*) (s) + 1) >= 0x90)))
7131f24d
KW
512#endif
513
b96a92fb
KW
514/* These are now machine generated, and the 'given' clause is no longer
515 * applicable */
516#define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
517 cBOOL(is_NONCHAR_utf8(s))
518#define UTF8_IS_NONCHAR_(s) \
519 UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s)
7131f24d 520
c867b360
JH
521#define UNICODE_SURROGATE_FIRST 0xD800
522#define UNICODE_SURROGATE_LAST 0xDFFF
523#define UNICODE_REPLACEMENT 0xFFFD
524#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 525
b851fbc1 526/* Though our UTF-8 encoding can go beyond this,
c76687c5 527 * let's be conservative and do as Unicode says. */
b851fbc1
JH
528#define PERL_UNICODE_MAX 0x10FFFF
529
949cf498
KW
530#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
531#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
532#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
533#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
534#define UNICODE_DISALLOW_SURROGATE 0x0010
535#define UNICODE_DISALLOW_NONCHAR 0x0020
536#define UNICODE_DISALLOW_SUPER 0x0040
537#define UNICODE_DISALLOW_FE_FF 0x0080
bb88be5f
KW
538#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
539 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
540#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
541 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
949cf498
KW
542
543/* For backward source compatibility, as are now the default */
544#define UNICODE_ALLOW_SURROGATE 0
545#define UNICODE_ALLOW_SUPER 0
546#define UNICODE_ALLOW_ANY 0
b851fbc1 547
1d72bdf6
NIS
548#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
549 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 550#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 551#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW
552#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
553 /* The other noncharacters end in FFFE or FFFF, which \
554 * the mask below catches both of, but beyond the last \
555 * official unicode code point, they aren't \
556 * noncharacters, since those aren't Unicode \
557 * characters at all */ \
558 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
559#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
560#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
1d72bdf6 561
ec34087a
KW
562#define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE
563#define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \
564 LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE
565#define MICRO_SIGN MICRO_SIGN_NATIVE
566#define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \
567 LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE
568#define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \
569 LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE
09091399
JH
570#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
571#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
572#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 573#define GREEK_SMALL_LETTER_MU 0x03BC
9e682c18
KW
574#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case
575 of MICRON */
576#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
577#define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
578#define LATIN_SMALL_LETTER_LONG_S 0x017F
a9f50d33
KW
579#define LATIN_SMALL_LIGATURE_LONG_S_T 0xFB05
580#define LATIN_SMALL_LIGATURE_ST 0xFB06
9e682c18
KW
581#define KELVIN_SIGN 0x212A
582#define ANGSTROM_SIGN 0x212B
09091399 583
9e55ce06 584#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
585#define UNI_DISPLAY_BACKSLASH 0x0002
586#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
587#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 588
5cd46e1f
KW
589#define ANYOF_FOLD_SHARP_S(node, input, end) \
590 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 591 (ANYOF_NONBITMAP(node)) && \
39065660 592 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f 593 ((end) > (input) + 1) && \
305b8651 594 isALPHA_FOLD_EQ((input)[0], 's'))
6302f837 595
ebc501f0 596#define SHARP_S_SKIP 2
3b0fc154 597
a4f7a67c
KW
598/* If you want to exclude surrogates, and beyond legal Unicode, see the blame
599 * log for earlier versions which gave details for these */
4d646140 600
6302f837
KW
601/* A helper macro for isUTF8_CHAR, so use that one, and not this one. This is
602 * retained solely for backwards compatibility and may be deprecated and
603 * removed in a future Perl version.
604 *
605 * regen/regcharclass.pl generates is_UTF8_CHAR_utf8() macros for up to these
d9f92374
KW
606 * number of bytes. So this has to be coordinated with that file */
607#ifdef EBCDIC
608# define IS_UTF8_CHAR_FAST(n) ((n) <= 3)
609#else
610# define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
611#endif
612
4d646140 613#ifndef EBCDIC
6302f837
KW
614/* A helper macro for isUTF8_CHAR, so use that one instead of this. This was
615 * generated by regen/regcharclass.pl, and then moved here. The lines that
616 * generated it were then commented out. This was done solely because it takes
617 * on the order of 10 minutes to generate, and is never going to change, unless
618 * the generated code is improved.
619 *
620 * The EBCDIC versions have been cut to not cover all of legal Unicode, so
621 * don't take too long to generate, and there is a separate one for each code
622 * page, so they are in regcharclass.h instead of here */
39a0f513 623/*
5dca9278 624 UTF8_CHAR: Matches legal UTF-8 encoded characters from 2 through 4 bytes
39a0f513 625
5dca9278 626 0x80 - 0x1FFFFF
39a0f513 627*/
4d646140 628/*** GENERATED CODE ***/
5dca9278
KW
629#define is_UTF8_CHAR_utf8_no_length_checks(s) \
630( ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \
631 ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \
632: ( 0xE0 == ((U8*)s)[0] ) ? \
633 ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
634: ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) ? \
635 ( ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
636: ( 0xF0 == ((U8*)s)[0] ) ? \
637 ( ( ( ( 0x90 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xBF ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
638: ( ( ( ( 0xF1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xF7 ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )
4d646140 639#endif
3b0fc154 640
6302f837 641/*
5dca9278
KW
642
643=for apidoc Am|STRLEN|isUTF8_CHAR|const U8 *s|const U8 *e
644
645Returns the number of bytes beginning at C<s> which form a legal UTF-8 (or
646UTF-EBCDIC) encoded character, looking no further than C<e - s> bytes into
647C<s>. Returns 0 if the sequence starting at C<s> through C<e - 1> is not
648well-formed UTF-8
6302f837
KW
649
650Note that an INVARIANT character (i.e. ASCII on non-EBCDIC
5dca9278
KW
651machines) is a valid UTF-8 character.
652
653=cut
654*/
6302f837 655
dd9bc2b0 656#define isUTF8_CHAR(s, e) (UNLIKELY((e) <= (s)) \
6302f837
KW
657 ? 0 \
658 : (UTF8_IS_INVARIANT(*s)) \
659 ? 1 \
dd9bc2b0 660 : UNLIKELY(((e) - (s)) < UTF8SKIP(s)) \
6302f837 661 ? 0 \
dd9bc2b0 662 : LIKELY(IS_UTF8_CHAR_FAST(UTF8SKIP(s))) \
5dca9278 663 ? is_UTF8_CHAR_utf8_no_length_checks(s) \
6302f837
KW
664 : _is_utf8_char_slow(s, e))
665
3cedd9d9
KW
666#define is_utf8_char_buf(buf, buf_end) isUTF8_CHAR(buf, buf_end)
667
6302f837
KW
668/* Do not use; should be deprecated. Use isUTF8_CHAR() instead; this is
669 * retained solely for backwards compatibility */
670#define IS_UTF8_CHAR(p, n) (isUTF8_CHAR(p, (p) + (n)) == n)
e9a8c099 671
57f0e7e2
KW
672#endif /* H_UTF8 */
673
e9a8c099
MHM
674/*
675 * Local variables:
676 * c-indentation-style: bsd
677 * c-basic-offset: 4
14d04a33 678 * indent-tabs-mode: nil
e9a8c099
MHM
679 * End:
680 *
14d04a33 681 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 682 */