| 1 | /* utf8.h |
| 2 | * |
| 3 | * Copyright (C) 2000, 2001, 2002, 2005, 2006, 2007, 2009, |
| 4 | * 2010, 2011 by Larry Wall and others |
| 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 | |
| 11 | /* Use UTF-8 as the default script encoding? |
| 12 | * Turning this on will break scripts having non-UTF-8 binary |
| 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 | |
| 20 | /* For to_utf8_fold_flags, q.v. */ |
| 21 | #define FOLD_FLAGS_LOCALE 0x1 |
| 22 | #define FOLD_FLAGS_FULL 0x2 |
| 23 | #define FOLD_FLAGS_NOMIX_ASCII 0x4 |
| 24 | |
| 25 | /* For _core_swash_init(), internal core use only */ |
| 26 | #define _CORE_SWASH_INIT_USER_DEFINED_PROPERTY 0x1 |
| 27 | #define _CORE_SWASH_INIT_RETURN_IF_UNDEF 0x2 |
| 28 | #define _CORE_SWASH_INIT_ACCEPT_INVLIST 0x4 |
| 29 | |
| 30 | #define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL) |
| 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) |
| 36 | |
| 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 | |
| 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) |
| 43 | #define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0) |
| 44 | #define FOLDEQ_UTF8_LOCALE (1 << 1) |
| 45 | #define FOLDEQ_S1_ALREADY_FOLDED (1 << 2) |
| 46 | #define FOLDEQ_S2_ALREADY_FOLDED (1 << 3) |
| 47 | |
| 48 | /* |
| 49 | =for apidoc ibcmp_utf8 |
| 50 | |
| 51 | This 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 | |
| 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" |
| 64 | |
| 65 | #else /* ! EBCDIC */ |
| 66 | START_EXTERN_C |
| 67 | |
| 68 | #ifdef DOINIT |
| 69 | EXTCONST unsigned char PL_utf8skip[] = { |
| 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). */ |
| 89 | }; |
| 90 | #else |
| 91 | EXTCONST unsigned char PL_utf8skip[]; |
| 92 | #endif |
| 93 | |
| 94 | END_EXTERN_C |
| 95 | |
| 96 | /* Native character to iso-8859-1 */ |
| 97 | #define NATIVE_TO_ASCII(ch) (ch) |
| 98 | #define ASCII_TO_NATIVE(ch) (ch) |
| 99 | /* Transform after encoding */ |
| 100 | #define NATIVE_TO_UTF(ch) (ch) |
| 101 | #define UTF_TO_NATIVE(ch) (ch) |
| 102 | /* Transforms in wide UV chars */ |
| 103 | #define UNI_TO_NATIVE(ch) (ch) |
| 104 | #define NATIVE_TO_UNI(ch) (ch) |
| 105 | /* Transforms in invariant space */ |
| 106 | #define NATIVE_TO_NEED(enc,ch) (ch) |
| 107 | #define ASCII_TO_NEED(enc,ch) (ch) |
| 108 | |
| 109 | /* As there are no translations, avoid the function wrapper */ |
| 110 | #define utf8n_to_uvchr utf8n_to_uvuni |
| 111 | #define valid_utf8_to_uvchr valid_utf8_to_uvuni |
| 112 | #define uvchr_to_utf8 uvuni_to_utf8 |
| 113 | |
| 114 | /* |
| 115 | |
| 116 | The following table is from Unicode 3.2. |
| 117 | |
| 118 | Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte |
| 119 | |
| 120 | U+0000..U+007F 00..7F |
| 121 | U+0080..U+07FF * C2..DF 80..BF |
| 122 | U+0800..U+0FFF E0 * A0..BF 80..BF |
| 123 | U+1000..U+CFFF E1..EC 80..BF 80..BF |
| 124 | U+D000..U+D7FF ED 80..9F 80..BF |
| 125 | U+D800..U+DFFF +++++++ utf16 surrogates, not legal utf8 +++++++ |
| 126 | U+E000..U+FFFF EE..EF 80..BF 80..BF |
| 127 | U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF |
| 128 | U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF |
| 129 | U+100000..U+10FFFF F4 80..8F 80..BF 80..BF |
| 130 | Below are non-Unicode code points |
| 131 | U+110000..U+13FFFF F4 90..BF 80..BF 80..BF |
| 132 | U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF |
| 133 | U+200000: F8.. * 88..BF 80..BF 80..BF 80..BF |
| 134 | |
| 135 | Note the gaps before several of the byte entries above marked by '*'. These are |
| 136 | caused by legal UTF-8 avoiding non-shortest encodings: it is technically |
| 137 | possible to UTF-8-encode a single code point in different ways, but that is |
| 138 | explicitly forbidden, and the shortest possible encoding should always be used |
| 139 | (and that is what Perl does). |
| 140 | |
| 141 | */ |
| 142 | |
| 143 | /* |
| 144 | Another way to look at it, as bits: |
| 145 | |
| 146 | Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte |
| 147 | |
| 148 | 0aaa aaaa 0aaa aaaa |
| 149 | 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa |
| 150 | cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa |
| 151 | 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa |
| 152 | |
| 153 | As you can see, the continuation bytes all begin with C<10>, and the |
| 154 | leading bits of the start byte tell how many bytes there are in the |
| 155 | encoded character. |
| 156 | |
| 157 | Perl's extended UTF-8 means we can have start bytes up to FF. |
| 158 | |
| 159 | */ |
| 160 | |
| 161 | #define UNI_IS_INVARIANT(c) (((UV)c) < 0x80) |
| 162 | #define UTF8_IS_START(c) (((U8)c) >= 0xc2) |
| 163 | #define UTF8_IS_CONTINUATION(c) (((U8)c) >= 0x80 && (((U8)c) <= 0xbf)) |
| 164 | #define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80) |
| 165 | |
| 166 | /* Masking with 0xfe allows low bit to be 0 or 1; thus this matches 0xc[23] */ |
| 167 | #define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)c & 0xfe) == 0xc2) |
| 168 | |
| 169 | #define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len)))) |
| 170 | |
| 171 | /* Masks out the initial one bits in a start byte, leaving the real data ones. |
| 172 | * Doesn't work on an invariant byte */ |
| 173 | #define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2))) |
| 174 | |
| 175 | #define UTF_CONTINUATION_MARK 0x80 |
| 176 | #define UTF_ACCUMULATION_SHIFT 6 |
| 177 | |
| 178 | /* 2**UTF_ACCUMULATION_SHIFT - 1 */ |
| 179 | #define UTF_CONTINUATION_MASK ((U8)0x3f) |
| 180 | |
| 181 | /* This sets the UTF_CONTINUATION_MASK in the upper bits of a word. If a value |
| 182 | * is anded with it, and the result is non-zero, then using the original value |
| 183 | * in UTF8_ACCUMULATE will overflow, shifting bits off the left */ |
| 184 | #define UTF_ACCUMULATION_OVERFLOW_MASK \ |
| 185 | (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) - UTF_ACCUMULATION_SHIFT)) |
| 186 | |
| 187 | #ifdef HAS_QUAD |
| 188 | #define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \ |
| 189 | (uv) < 0x800 ? 2 : \ |
| 190 | (uv) < 0x10000 ? 3 : \ |
| 191 | (uv) < 0x200000 ? 4 : \ |
| 192 | (uv) < 0x4000000 ? 5 : \ |
| 193 | (uv) < 0x80000000 ? 6 : \ |
| 194 | (uv) < UTF8_QUAD_MAX ? 7 : 13 ) |
| 195 | #else |
| 196 | /* No, I'm not even going to *TRY* putting #ifdef inside a #define */ |
| 197 | #define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \ |
| 198 | (uv) < 0x800 ? 2 : \ |
| 199 | (uv) < 0x10000 ? 3 : \ |
| 200 | (uv) < 0x200000 ? 4 : \ |
| 201 | (uv) < 0x4000000 ? 5 : \ |
| 202 | (uv) < 0x80000000 ? 6 : 7 ) |
| 203 | #endif |
| 204 | |
| 205 | #endif /* EBCDIC vs ASCII */ |
| 206 | |
| 207 | /* Rest of these are attributes of Unicode and perl's internals rather than the |
| 208 | * encoding, or happen to be the same in both ASCII and EBCDIC (at least at |
| 209 | * this level; the macros that some of these call may have different |
| 210 | * definitions in the two encodings */ |
| 211 | |
| 212 | #define NATIVE8_TO_UNI(ch) NATIVE_TO_ASCII(ch) /* a clearer synonym */ |
| 213 | |
| 214 | #define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) | (((U8)new) & UTF_CONTINUATION_MASK)) |
| 215 | |
| 216 | /* Convert a two (not one) byte utf8 character to a unicode code point value. |
| 217 | * Needs just one iteration of accumulate. Should not be used unless it is |
| 218 | * known that the two bytes are legal: 1) two-byte start, and 2) continuation. |
| 219 | * Note that the result can be larger than 255 if the input character is not |
| 220 | * downgradable */ |
| 221 | #define TWO_BYTE_UTF8_TO_UNI(HI, LO) \ |
| 222 | UTF8_ACCUMULATE((NATIVE_TO_UTF(HI) & UTF_START_MASK(2)), \ |
| 223 | NATIVE_TO_UTF(LO)) |
| 224 | |
| 225 | #define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)] |
| 226 | |
| 227 | #define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c)) |
| 228 | #define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE8_TO_UNI(c)) |
| 229 | |
| 230 | #define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */ |
| 231 | |
| 232 | /* The macros in the next sets are used to generate the two utf8 or utfebcdic |
| 233 | * bytes from an ordinal that is known to fit into two bytes; it must be less |
| 234 | * than 0x3FF to work across both encodings. */ |
| 235 | /* Nocast allows these to be used in the case label of a switch statement */ |
| 236 | #define UTF8_TWO_BYTE_HI_nocast(c) UTF_TO_NATIVE(((c) >> UTF_ACCUMULATION_SHIFT) | (0xFF & UTF_START_MARK(2))) |
| 237 | #define UTF8_TWO_BYTE_LO_nocast(c) UTF_TO_NATIVE(((c) & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK) |
| 238 | |
| 239 | #define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c))) |
| 240 | #define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c))) |
| 241 | |
| 242 | /* This name is used when the source is a single byte */ |
| 243 | #define UTF8_EIGHT_BIT_HI(c) UTF8_TWO_BYTE_HI((U8)(c)) |
| 244 | #define UTF8_EIGHT_BIT_LO(c) UTF8_TWO_BYTE_LO((U8)(c)) |
| 245 | |
| 246 | /* |
| 247 | * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem |
| 248 | * from an earlier version of these macros in which they didn't call the |
| 249 | * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the |
| 250 | * beginning of a utf8 character. Now that foo_utf8() determines that itself, |
| 251 | * no need to do it again here |
| 252 | */ |
| 253 | #define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \ |
| 254 | ? isIDFIRST(*(p)) \ |
| 255 | : isIDFIRST_utf8((const U8*)p)) |
| 256 | #define isALNUM_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \ |
| 257 | ? isALNUM(*(p)) \ |
| 258 | : isALNUM_utf8((const U8*)p)) |
| 259 | |
| 260 | #define isIDFIRST_lazy(p) isIDFIRST_lazy_if(p,1) |
| 261 | #define isALNUM_lazy(p) isALNUM_lazy_if(p,1) |
| 262 | |
| 263 | #define UTF8_MAXBYTES 13 |
| 264 | /* How wide can a single UTF-8 encoded character become in bytes. |
| 265 | * NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 |
| 266 | * since UTF-8 is an encoding of Unicode and given Unicode's current |
| 267 | * upper limit only four bytes is possible. Perl thinks of UTF-8 |
| 268 | * as a way to encode non-negative integers in a binary format. */ |
| 269 | #define UTF8_MAXLEN UTF8_MAXBYTES |
| 270 | |
| 271 | /* The maximum number of UTF-8 bytes a single Unicode character can |
| 272 | * uppercase/lowercase/fold into; this number depends on the Unicode |
| 273 | * version. An example of maximal expansion is the U+03B0 which |
| 274 | * uppercases to U+03C5 U+0308 U+0301. The Unicode databases that |
| 275 | * tell these things are UnicodeData.txt, CaseFolding.txt, and |
| 276 | * SpecialCasing.txt. The value is 6 for strict Unicode characters, but it has |
| 277 | * to be as big as Perl allows for a single character */ |
| 278 | #define UTF8_MAXBYTES_CASE UTF8_MAXBYTES |
| 279 | |
| 280 | /* A Unicode character can fold to up to 3 characters */ |
| 281 | #define UTF8_MAX_FOLD_CHAR_EXPAND 3 |
| 282 | |
| 283 | #define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES) |
| 284 | #define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES) |
| 285 | #define IN_UNI_8_BIT \ |
| 286 | (CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT|HINT_LOCALE_NOT_CHARS) \ |
| 287 | && ! IN_LOCALE_RUNTIME && ! IN_BYTES) |
| 288 | |
| 289 | |
| 290 | #define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */ |
| 291 | |
| 292 | /* Allow first byte to be a continuation byte */ |
| 293 | #define UTF8_ALLOW_CONTINUATION 0x0002 |
| 294 | |
| 295 | /* Allow second... bytes to be non-continuation bytes */ |
| 296 | #define UTF8_ALLOW_NON_CONTINUATION 0x0004 |
| 297 | |
| 298 | /* expecting more bytes than were available in the string */ |
| 299 | #define UTF8_ALLOW_SHORT 0x0008 |
| 300 | |
| 301 | /* Overlong sequence; i.e., the code point can be specified in fewer bytes. */ |
| 302 | #define UTF8_ALLOW_LONG 0x0010 |
| 303 | |
| 304 | #define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */ |
| 305 | #define UTF8_WARN_SURROGATE 0x0040 |
| 306 | |
| 307 | #define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */ |
| 308 | #define UTF8_WARN_NONCHAR 0x0100 /* code points */ |
| 309 | |
| 310 | #define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */ |
| 311 | #define UTF8_WARN_SUPER 0x0400 /* points above the legal max */ |
| 312 | |
| 313 | /* Code points which never were part of the original UTF-8 standard, the first |
| 314 | * byte of which is a FE or FF on ASCII platforms. */ |
| 315 | #define UTF8_DISALLOW_FE_FF 0x0800 |
| 316 | #define UTF8_WARN_FE_FF 0x1000 |
| 317 | |
| 318 | #define UTF8_CHECK_ONLY 0x2000 |
| 319 | |
| 320 | /* For backwards source compatibility. They do nothing, as the default now |
| 321 | * includes what they used to mean. The first one's meaning was to allow the |
| 322 | * just the single non-character 0xFFFF */ |
| 323 | #define UTF8_ALLOW_FFFF 0 |
| 324 | #define UTF8_ALLOW_SURROGATE 0 |
| 325 | |
| 326 | #define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF) |
| 327 | #define UTF8_WARN_ILLEGAL_INTERCHANGE \ |
| 328 | (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF) |
| 329 | #define UTF8_ALLOW_ANY \ |
| 330 | (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE)) |
| 331 | #define UTF8_ALLOW_ANYUV \ |
| 332 | (UTF8_ALLOW_EMPTY \ |
| 333 | & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE)) |
| 334 | #define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \ |
| 335 | UTF8_ALLOW_ANYUV) |
| 336 | |
| 337 | /* Surrogates, non-character code points and above-Unicode code points are |
| 338 | * problematic in some contexts. This allows code that needs to check for |
| 339 | * those to to quickly exclude the vast majority of code points it will |
| 340 | * encounter */ |
| 341 | #ifdef EBCDIC |
| 342 | # define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE UTF_TO_NATIVE(0xF1) |
| 343 | #else |
| 344 | # define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE 0xED |
| 345 | #endif |
| 346 | |
| 347 | /* ASCII EBCDIC I8 |
| 348 | * U+D7FF: \xED\x9F\xBF \xF1\xB5\xBF\xBF last before surrogates |
| 349 | * U+D800: \xED\xA0\x80 \xF1\xB6\xA0\xA0 1st surrogate |
| 350 | * U+DFFF: \xED\xBF\xBF \xF1\xB7\xBF\xBF final surrogate |
| 351 | * U+E000: \xEE\x80\x80 \xF1\xB8\xA0\xA0 next after surrogates |
| 352 | */ |
| 353 | #ifdef EBCDIC /* Both versions assume well-formed UTF8 */ |
| 354 | # define UTF8_IS_SURROGATE(s) (*(s) == UTF_TO_NATIVE(0xF1) \ |
| 355 | && ((*((s) +1) == UTF_TO_NATIVE(0xB6)) \ |
| 356 | || *((s) + 1) == UTF_TO_NATIVE(0xB7))) |
| 357 | /* <send> points to one beyond the end of the string that starts at <s> */ |
| 358 | # define UTF8_IS_REPLACEMENT(s, send) (*(s) == UTF_TO_NATIVE(0xEF) \ |
| 359 | && (send - s) >= 4 \ |
| 360 | && *((s) + 1) == UTF_TO_NATIVE(0xBF) \ |
| 361 | && *((s) + 2) == UTF_TO_NATIVE(0xBF) \ |
| 362 | && *((s) + 3) == UTF_TO_NATIVE(0xBD) |
| 363 | #else |
| 364 | # define UTF8_IS_SURROGATE(s) (*(s) == 0xED && *((s) + 1) >= 0xA0) |
| 365 | # define UTF8_IS_REPLACEMENT(s, send) (*(s) == 0xEF \ |
| 366 | && (send - s) >= 3 \ |
| 367 | && *((s) + 1) == 0xBF \ |
| 368 | && *((s) + 2) == 0xBD) |
| 369 | #endif |
| 370 | |
| 371 | /* ASCII EBCDIC I8 |
| 372 | * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode |
| 373 | * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0 |
| 374 | * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1 |
| 375 | */ |
| 376 | #ifdef EBCDIC /* Both versions assume well-formed UTF8 */ |
| 377 | # define UTF8_IS_SUPER(s) (*(s) >= UTF_TO_NATIVE(0xF9) \ |
| 378 | && (*(s) > UTF_TO_NATIVE(0xF9) || (*((s) + 1) >= UTF_TO_NATIVE(0xA2)))) |
| 379 | #else |
| 380 | # define UTF8_IS_SUPER(s) (*(s) >= 0xF4 \ |
| 381 | && (*(s) > 0xF4 || (*((s) + 1) >= 0x90))) |
| 382 | #endif |
| 383 | |
| 384 | /* ASCII EBCDIC I8 |
| 385 | * U+FDCF: \xEF\xB7\x8F \xF1\xBF\xAE\xAF last before non-char block |
| 386 | * U+FDD0: \xEF\xB7\x90 \xF1\xBF\xAE\xB0 first non-char in block |
| 387 | * U+FDEF: \xEF\xB7\xAF \xF1\xBF\xAF\xAF last non-char in block |
| 388 | * U+FDF0: \xEF\xB7\xB0 \xF1\xBF\xAF\xB0 first after non-char block |
| 389 | * U+FFFF: \xEF\xBF\xBF \xF1\xBF\xBF\xBF |
| 390 | * U+1FFFF: \xF0\x9F\xBF\xBF \xF3\xBF\xBF\xBF |
| 391 | * U+2FFFF: \xF0\xAF\xBF\xBF \xF5\xBF\xBF\xBF |
| 392 | * U+3FFFF: \xF0\xBF\xBF\xBF \xF7\xBF\xBF\xBF |
| 393 | * U+4FFFF: \xF1\x8F\xBF\xBF \xF8\xA9\xBF\xBF\xBF |
| 394 | * U+5FFFF: \xF1\x9F\xBF\xBF \xF8\xAB\xBF\xBF\xBF |
| 395 | * U+6FFFF: \xF1\xAF\xBF\xBF \xF8\xAD\xBF\xBF\xBF |
| 396 | * U+7FFFF: \xF1\xBF\xBF\xBF \xF8\xAF\xBF\xBF\xBF |
| 397 | * U+8FFFF: \xF2\x8F\xBF\xBF \xF8\xB1\xBF\xBF\xBF |
| 398 | * U+9FFFF: \xF2\x9F\xBF\xBF \xF8\xB3\xBF\xBF\xBF |
| 399 | * U+AFFFF: \xF2\xAF\xBF\xBF \xF8\xB5\xBF\xBF\xBF |
| 400 | * U+BFFFF: \xF2\xBF\xBF\xBF \xF8\xB7\xBF\xBF\xBF |
| 401 | * U+CFFFF: \xF3\x8F\xBF\xBF \xF8\xB9\xBF\xBF\xBF |
| 402 | * U+DFFFF: \xF3\x9F\xBF\xBF \xF8\xBB\xBF\xBF\xBF |
| 403 | * U+EFFFF: \xF3\xAF\xBF\xBF \xF8\xBD\xBF\xBF\xBF |
| 404 | * U+FFFFF: \xF3\xBF\xBF\xBF \xF8\xBF\xBF\xBF\xBF |
| 405 | * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF |
| 406 | */ |
| 407 | #define UTF8_IS_NONCHAR_(s) ( \ |
| 408 | *(s) >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \ |
| 409 | && ! UTF8_IS_SUPER(s) \ |
| 410 | && UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_FIRST_PROBLEMATIC(s) \ |
| 411 | |
| 412 | #ifdef EBCDIC /* Both versions assume well-formed UTF8 */ |
| 413 | # define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \ |
| 414 | ((*(s) == UTF_TO_NATIVE(0xF1) \ |
| 415 | && (*((s) + 1) == UTF_TO_NATIVE(0xBF) \ |
| 416 | && ((*((s) + 2) == UTF_TO_NATIVE(0xAE) \ |
| 417 | && *((s) + 3) >= UTF_TO_NATIVE(0xB0)) \ |
| 418 | || (*((s) + 2) == UTF_TO_NATIVE(0xAF) \ |
| 419 | && *((s) + 3) <= UTF_TO_NATIVE(0xAF))))) \ |
| 420 | || (UTF8SKIP(*(s)) > 3 \ |
| 421 | /* (These were all derived by inspection and experimentation with an */ \ |
| 422 | /* editor) The next line checks the next to final byte in the char */ \ |
| 423 | && *((s) + UTF8SKIP(*(s)) - 2) == UTF_TO_NATIVE(0xBF) \ |
| 424 | && *((s) + UTF8SKIP(*(s)) - 3) == UTF_TO_NATIVE(0xBF) \ |
| 425 | && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 4)) & 0x81) == 0x81 \ |
| 426 | && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 1)) & 0xBE) == 0XBE)) |
| 427 | #else |
| 428 | # define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \ |
| 429 | ((*(s) == 0xEF \ |
| 430 | && ((*((s) + 1) == 0xB7 && (*((s) + 2) >= 0x90 && (*((s) + 2) <= 0xAF)))\ |
| 431 | /* Gets U+FFF[EF] */ \ |
| 432 | || (*((s) + 1) == 0xBF && ((*((s) + 2) & 0xBE) == 0xBE)))) \ |
| 433 | || ((*((s) + 2) == 0xBF \ |
| 434 | && (*((s) + 3) & 0xBE) == 0xBE \ |
| 435 | /* Excludes things like U+10FFE = \xF0\x90\xBF\xBE */ \ |
| 436 | && (*((s) + 1) & 0x8F) == 0x8F))) |
| 437 | #endif |
| 438 | |
| 439 | #define UNICODE_SURROGATE_FIRST 0xD800 |
| 440 | #define UNICODE_SURROGATE_LAST 0xDFFF |
| 441 | #define UNICODE_REPLACEMENT 0xFFFD |
| 442 | #define UNICODE_BYTE_ORDER_MARK 0xFEFF |
| 443 | |
| 444 | /* Though our UTF-8 encoding can go beyond this, |
| 445 | * let's be conservative and do as Unicode says. */ |
| 446 | #define PERL_UNICODE_MAX 0x10FFFF |
| 447 | |
| 448 | #define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */ |
| 449 | #define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */ |
| 450 | #define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */ |
| 451 | #define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */ |
| 452 | #define UNICODE_DISALLOW_SURROGATE 0x0010 |
| 453 | #define UNICODE_DISALLOW_NONCHAR 0x0020 |
| 454 | #define UNICODE_DISALLOW_SUPER 0x0040 |
| 455 | #define UNICODE_DISALLOW_FE_FF 0x0080 |
| 456 | #define UNICODE_WARN_ILLEGAL_INTERCHANGE \ |
| 457 | (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER) |
| 458 | #define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \ |
| 459 | (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER) |
| 460 | |
| 461 | /* For backward source compatibility, as are now the default */ |
| 462 | #define UNICODE_ALLOW_SURROGATE 0 |
| 463 | #define UNICODE_ALLOW_SUPER 0 |
| 464 | #define UNICODE_ALLOW_ANY 0 |
| 465 | |
| 466 | #define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \ |
| 467 | (c) <= UNICODE_SURROGATE_LAST) |
| 468 | #define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT) |
| 469 | #define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK) |
| 470 | #define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \ |
| 471 | /* The other noncharacters end in FFFE or FFFF, which \ |
| 472 | * the mask below catches both of, but beyond the last \ |
| 473 | * official unicode code point, they aren't \ |
| 474 | * noncharacters, since those aren't Unicode \ |
| 475 | * characters at all */ \ |
| 476 | || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c))) |
| 477 | #define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX) |
| 478 | #define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF) |
| 479 | |
| 480 | #ifdef HAS_QUAD |
| 481 | # define UTF8_QUAD_MAX UINT64_C(0x1000000000) |
| 482 | #endif |
| 483 | |
| 484 | #define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3 |
| 485 | #define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2 |
| 486 | #define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3 |
| 487 | #define GREEK_SMALL_LETTER_MU 0x03BC |
| 488 | #define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */ |
| 489 | #define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */ |
| 490 | #define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E |
| 491 | #define LATIN_SMALL_LETTER_LONG_S 0x017F |
| 492 | #define KELVIN_SIGN 0x212A |
| 493 | #define ANGSTROM_SIGN 0x212B |
| 494 | |
| 495 | #define UNI_DISPLAY_ISPRINT 0x0001 |
| 496 | #define UNI_DISPLAY_BACKSLASH 0x0002 |
| 497 | #define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH) |
| 498 | #define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH) |
| 499 | |
| 500 | #ifndef EBCDIC |
| 501 | # define LATIN_SMALL_LETTER_SHARP_S 0x00DF |
| 502 | # define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS 0x00FF |
| 503 | # define MICRO_SIGN 0x00B5 |
| 504 | # define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE 0x00C5 |
| 505 | # define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE 0x00E5 |
| 506 | #endif |
| 507 | |
| 508 | #define ANYOF_FOLD_SHARP_S(node, input, end) \ |
| 509 | (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \ |
| 510 | (ANYOF_NONBITMAP(node)) && \ |
| 511 | (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \ |
| 512 | ((end) > (input) + 1) && \ |
| 513 | toLOWER((input)[0]) == 's' && \ |
| 514 | toLOWER((input)[1]) == 's') |
| 515 | #define SHARP_S_SKIP 2 |
| 516 | |
| 517 | #ifndef EBCDIC |
| 518 | /* If you want to exclude surrogates, and beyond legal Unicode, see the blame |
| 519 | * log for earlier versions which gave details for these */ |
| 520 | # define IS_UTF8_CHAR_1(p) \ |
| 521 | ((p)[0] <= 0x7F) |
| 522 | # define IS_UTF8_CHAR_2(p) \ |
| 523 | ((p)[0] >= 0xC2 && (p)[0] <= 0xDF && \ |
| 524 | (p)[1] >= 0x80 && (p)[1] <= 0xBF) |
| 525 | # define IS_UTF8_CHAR_3a(p) \ |
| 526 | ((p)[0] == 0xE0 && \ |
| 527 | (p)[1] >= 0xA0 && (p)[1] <= 0xBF && \ |
| 528 | (p)[2] >= 0x80 && (p)[2] <= 0xBF) |
| 529 | # define IS_UTF8_CHAR_3b(p) \ |
| 530 | ((p)[0] >= 0xE1 && (p)[0] <= 0xEF && \ |
| 531 | (p)[1] >= 0x80 && (p)[1] <= 0xBF && \ |
| 532 | (p)[2] >= 0x80 && (p)[2] <= 0xBF) |
| 533 | # define IS_UTF8_CHAR_4a(p) \ |
| 534 | ((p)[0] == 0xF0 && \ |
| 535 | (p)[1] >= 0x90 && (p)[1] <= 0xBF && \ |
| 536 | (p)[2] >= 0x80 && (p)[2] <= 0xBF && \ |
| 537 | (p)[3] >= 0x80 && (p)[3] <= 0xBF) |
| 538 | /* The 0xF7 allows us to go to 0x1fffff (0x200000 would |
| 539 | * require five bytes). Not doing any further code points |
| 540 | * since that is not needed (and that would not be strict |
| 541 | * UTF-8, anyway). The "slow path" in Perl_is_utf8_char() |
| 542 | * will take care of the "extended UTF-8". */ |
| 543 | # define IS_UTF8_CHAR_4b(p) \ |
| 544 | ((p)[0] >= 0xF1 && (p)[0] <= 0xF7 && \ |
| 545 | (p)[1] >= 0x80 && (p)[1] <= 0xBF && \ |
| 546 | (p)[2] >= 0x80 && (p)[2] <= 0xBF && \ |
| 547 | (p)[3] >= 0x80 && (p)[3] <= 0xBF) |
| 548 | |
| 549 | # define IS_UTF8_CHAR_3(p) \ |
| 550 | (IS_UTF8_CHAR_3a(p) || \ |
| 551 | IS_UTF8_CHAR_3b(p)) |
| 552 | # define IS_UTF8_CHAR_4(p) \ |
| 553 | (IS_UTF8_CHAR_4a(p) || \ |
| 554 | IS_UTF8_CHAR_4b(p)) |
| 555 | |
| 556 | /* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it |
| 557 | * (1) allows UTF-8 encoded UTF-16 surrogates |
| 558 | * (2) it allows code points past U+10FFFF. |
| 559 | * The Perl_is_utf8_char() full "slow" code will handle the Perl |
| 560 | * "extended UTF-8". */ |
| 561 | # define IS_UTF8_CHAR(p, n) \ |
| 562 | ((n) == 1 ? IS_UTF8_CHAR_1(p) : \ |
| 563 | (n) == 2 ? IS_UTF8_CHAR_2(p) : \ |
| 564 | (n) == 3 ? IS_UTF8_CHAR_3(p) : \ |
| 565 | (n) == 4 ? IS_UTF8_CHAR_4(p) : 0) |
| 566 | |
| 567 | # define IS_UTF8_CHAR_FAST(n) ((n) <= 4) |
| 568 | |
| 569 | #else /* EBCDIC */ |
| 570 | |
| 571 | /* This is an attempt to port IS_UTF8_CHAR to EBCDIC based on eyeballing. |
| 572 | * untested. If want to exclude surrogates and above-Unicode, see the |
| 573 | * definitions for UTF8_IS_SURROGATE and UTF8_IS_SUPER */ |
| 574 | # define IS_UTF8_CHAR_1(p) \ |
| 575 | (NATIVE_TO_ASCII((p)[0]) <= 0x9F) |
| 576 | # define IS_UTF8_CHAR_2(p) \ |
| 577 | (NATIVE_TO_I8((p)[0]) >= 0xC5 && NATIVE_TO_I8((p)[0]) <= 0xDF && \ |
| 578 | NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF) |
| 579 | # define IS_UTF8_CHAR_3(p) \ |
| 580 | (NATIVE_TO_I8((p)[0]) == 0xE1 && NATIVE_TO_I8((p)[1]) <= 0xEF && \ |
| 581 | NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \ |
| 582 | NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF) |
| 583 | # define IS_UTF8_CHAR_4a(p) \ |
| 584 | (NATIVE_TO_I8((p)[0]) == 0xF0 && \ |
| 585 | NATIVE_TO_I8((p)[1]) >= 0xB0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \ |
| 586 | NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \ |
| 587 | NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF) |
| 588 | # define IS_UTF8_CHAR_4b(p) \ |
| 589 | (NATIVE_TO_I8((p)[0]) >= 0xF1 && NATIVE_TO_I8((p)[0]) <= 0xF7 && \ |
| 590 | NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \ |
| 591 | NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \ |
| 592 | NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF) |
| 593 | # define IS_UTF8_CHAR_5a(p) \ |
| 594 | (NATIVE_TO_I8((p)[0]) == 0xF8 && \ |
| 595 | NATIVE_TO_I8((p)[1]) >= 0xA8 && NATIVE_TO_I8((p)[1]) <= 0xBF && \ |
| 596 | NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \ |
| 597 | NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \ |
| 598 | NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF) |
| 599 | # define IS_UTF8_CHAR_5b(p) \ |
| 600 | (NATIVE_TO_I8((p)[0]) >= 0xF9 && NATIVE_TO_I8((p)[1]) <= 0xFB && \ |
| 601 | NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \ |
| 602 | NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \ |
| 603 | NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \ |
| 604 | NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF) |
| 605 | |
| 606 | # define IS_UTF8_CHAR_4(p) \ |
| 607 | (IS_UTF8_CHAR_4a(p) || \ |
| 608 | IS_UTF8_CHAR_4b(p)) |
| 609 | # define IS_UTF8_CHAR_5(p) \ |
| 610 | (IS_UTF8_CHAR_5a(p) || \ |
| 611 | IS_UTF8_CHAR_5b(p)) |
| 612 | # define IS_UTF8_CHAR(p, n) \ |
| 613 | ((n) == 1 ? IS_UTF8_CHAR_1(p) : \ |
| 614 | (n) == 2 ? IS_UTF8_CHAR_2(p) : \ |
| 615 | (n) == 3 ? IS_UTF8_CHAR_3(p) : \ |
| 616 | (n) == 4 ? IS_UTF8_CHAR_4(p) : \ |
| 617 | (n) == 5 ? IS_UTF8_CHAR_5(p) : 0) |
| 618 | |
| 619 | # define IS_UTF8_CHAR_FAST(n) ((n) <= 5) |
| 620 | |
| 621 | #endif /* IS_UTF8_CHAR() for UTF-8 */ |
| 622 | |
| 623 | /* |
| 624 | * Local variables: |
| 625 | * c-indentation-style: bsd |
| 626 | * c-basic-offset: 4 |
| 627 | * indent-tabs-mode: nil |
| 628 | * End: |
| 629 | * |
| 630 | * ex: set ts=8 sts=4 sw=4 et: |
| 631 | */ |