diff options
| author | Mitja Felicijan <mitja.felicijan@gmail.com> | 2026-01-21 20:22:09 +0100 |
|---|---|---|
| committer | Mitja Felicijan <mitja.felicijan@gmail.com> | 2026-01-21 20:22:09 +0100 |
| commit | 5a8dbc6347b3541e84fe669b22c17ad3b715e258 (patch) | |
| tree | b148c450939688caaaeb4adac6f2faa1eaffe649 /vendor/github.com/mitjafelicijan/go-tree-sitter/utf16.h | |
| download | qwe-editor-5a8dbc6347b3541e84fe669b22c17ad3b715e258.tar.gz | |
Engage!
Diffstat (limited to 'vendor/github.com/mitjafelicijan/go-tree-sitter/utf16.h')
| -rw-r--r-- | vendor/github.com/mitjafelicijan/go-tree-sitter/utf16.h | 733 |
1 files changed, 733 insertions, 0 deletions
diff --git a/vendor/github.com/mitjafelicijan/go-tree-sitter/utf16.h b/vendor/github.com/mitjafelicijan/go-tree-sitter/utf16.h new file mode 100644 index 0000000..9c31330 --- /dev/null +++ b/vendor/github.com/mitjafelicijan/go-tree-sitter/utf16.h | |||
| @@ -0,0 +1,733 @@ | |||
| 1 | // © 2016 and later: Unicode, Inc. and others. | ||
| 2 | // License & terms of use: http://www.unicode.org/copyright.html | ||
| 3 | /* | ||
| 4 | ******************************************************************************* | ||
| 5 | * | ||
| 6 | * Copyright (C) 1999-2012, International Business Machines | ||
| 7 | * Corporation and others. All Rights Reserved. | ||
| 8 | * | ||
| 9 | ******************************************************************************* | ||
| 10 | * file name: utf16.h | ||
| 11 | * encoding: UTF-8 | ||
| 12 | * tab size: 8 (not used) | ||
| 13 | * indentation:4 | ||
| 14 | * | ||
| 15 | * created on: 1999sep09 | ||
| 16 | * created by: Markus W. Scherer | ||
| 17 | */ | ||
| 18 | |||
| 19 | /** | ||
| 20 | * \file | ||
| 21 | * \brief C API: 16-bit Unicode handling macros | ||
| 22 | * | ||
| 23 | * This file defines macros to deal with 16-bit Unicode (UTF-16) code units and strings. | ||
| 24 | * | ||
| 25 | * For more information see utf.h and the ICU User Guide Strings chapter | ||
| 26 | * (http://userguide.icu-project.org/strings). | ||
| 27 | * | ||
| 28 | * <em>Usage:</em> | ||
| 29 | * ICU coding guidelines for if() statements should be followed when using these macros. | ||
| 30 | * Compound statements (curly braces {}) must be used for if-else-while... | ||
| 31 | * bodies and all macro statements should be terminated with semicolon. | ||
| 32 | */ | ||
| 33 | |||
| 34 | #ifndef __UTF16_H__ | ||
| 35 | #define __UTF16_H__ | ||
| 36 | |||
| 37 | #include "umachine.h" | ||
| 38 | #ifndef __UTF_H__ | ||
| 39 | # include "utf.h" | ||
| 40 | #endif | ||
| 41 | |||
| 42 | /* single-code point definitions -------------------------------------------- */ | ||
| 43 | |||
| 44 | /** | ||
| 45 | * Does this code unit alone encode a code point (BMP, not a surrogate)? | ||
| 46 | * @param c 16-bit code unit | ||
| 47 | * @return TRUE or FALSE | ||
| 48 | * @stable ICU 2.4 | ||
| 49 | */ | ||
| 50 | #define U16_IS_SINGLE(c) !U_IS_SURROGATE(c) | ||
| 51 | |||
| 52 | /** | ||
| 53 | * Is this code unit a lead surrogate (U+d800..U+dbff)? | ||
| 54 | * @param c 16-bit code unit | ||
| 55 | * @return TRUE or FALSE | ||
| 56 | * @stable ICU 2.4 | ||
| 57 | */ | ||
| 58 | #define U16_IS_LEAD(c) (((c)&0xfffffc00)==0xd800) | ||
| 59 | |||
| 60 | /** | ||
| 61 | * Is this code unit a trail surrogate (U+dc00..U+dfff)? | ||
| 62 | * @param c 16-bit code unit | ||
| 63 | * @return TRUE or FALSE | ||
| 64 | * @stable ICU 2.4 | ||
| 65 | */ | ||
| 66 | #define U16_IS_TRAIL(c) (((c)&0xfffffc00)==0xdc00) | ||
| 67 | |||
| 68 | /** | ||
| 69 | * Is this code unit a surrogate (U+d800..U+dfff)? | ||
| 70 | * @param c 16-bit code unit | ||
| 71 | * @return TRUE or FALSE | ||
| 72 | * @stable ICU 2.4 | ||
| 73 | */ | ||
| 74 | #define U16_IS_SURROGATE(c) U_IS_SURROGATE(c) | ||
| 75 | |||
| 76 | /** | ||
| 77 | * Assuming c is a surrogate code point (U16_IS_SURROGATE(c)), | ||
| 78 | * is it a lead surrogate? | ||
| 79 | * @param c 16-bit code unit | ||
| 80 | * @return TRUE or FALSE | ||
| 81 | * @stable ICU 2.4 | ||
| 82 | */ | ||
| 83 | #define U16_IS_SURROGATE_LEAD(c) (((c)&0x400)==0) | ||
| 84 | |||
| 85 | /** | ||
| 86 | * Assuming c is a surrogate code point (U16_IS_SURROGATE(c)), | ||
| 87 | * is it a trail surrogate? | ||
| 88 | * @param c 16-bit code unit | ||
| 89 | * @return TRUE or FALSE | ||
| 90 | * @stable ICU 4.2 | ||
| 91 | */ | ||
| 92 | #define U16_IS_SURROGATE_TRAIL(c) (((c)&0x400)!=0) | ||
| 93 | |||
| 94 | /** | ||
| 95 | * Helper constant for U16_GET_SUPPLEMENTARY. | ||
| 96 | * @internal | ||
| 97 | */ | ||
| 98 | #define U16_SURROGATE_OFFSET ((0xd800<<10UL)+0xdc00-0x10000) | ||
| 99 | |||
| 100 | /** | ||
| 101 | * Get a supplementary code point value (U+10000..U+10ffff) | ||
| 102 | * from its lead and trail surrogates. | ||
| 103 | * The result is undefined if the input values are not | ||
| 104 | * lead and trail surrogates. | ||
| 105 | * | ||
| 106 | * @param lead lead surrogate (U+d800..U+dbff) | ||
| 107 | * @param trail trail surrogate (U+dc00..U+dfff) | ||
| 108 | * @return supplementary code point (U+10000..U+10ffff) | ||
| 109 | * @stable ICU 2.4 | ||
| 110 | */ | ||
| 111 | #define U16_GET_SUPPLEMENTARY(lead, trail) \ | ||
| 112 | (((UChar32)(lead)<<10UL)+(UChar32)(trail)-U16_SURROGATE_OFFSET) | ||
| 113 | |||
| 114 | |||
| 115 | /** | ||
| 116 | * Get the lead surrogate (0xd800..0xdbff) for a | ||
| 117 | * supplementary code point (0x10000..0x10ffff). | ||
| 118 | * @param supplementary 32-bit code point (U+10000..U+10ffff) | ||
| 119 | * @return lead surrogate (U+d800..U+dbff) for supplementary | ||
| 120 | * @stable ICU 2.4 | ||
| 121 | */ | ||
| 122 | #define U16_LEAD(supplementary) (UChar)(((supplementary)>>10)+0xd7c0) | ||
| 123 | |||
| 124 | /** | ||
| 125 | * Get the trail surrogate (0xdc00..0xdfff) for a | ||
| 126 | * supplementary code point (0x10000..0x10ffff). | ||
| 127 | * @param supplementary 32-bit code point (U+10000..U+10ffff) | ||
| 128 | * @return trail surrogate (U+dc00..U+dfff) for supplementary | ||
| 129 | * @stable ICU 2.4 | ||
| 130 | */ | ||
| 131 | #define U16_TRAIL(supplementary) (UChar)(((supplementary)&0x3ff)|0xdc00) | ||
| 132 | |||
| 133 | /** | ||
| 134 | * How many 16-bit code units are used to encode this Unicode code point? (1 or 2) | ||
| 135 | * The result is not defined if c is not a Unicode code point (U+0000..U+10ffff). | ||
| 136 | * @param c 32-bit code point | ||
| 137 | * @return 1 or 2 | ||
| 138 | * @stable ICU 2.4 | ||
| 139 | */ | ||
| 140 | #define U16_LENGTH(c) ((uint32_t)(c)<=0xffff ? 1 : 2) | ||
| 141 | |||
| 142 | /** | ||
| 143 | * The maximum number of 16-bit code units per Unicode code point (U+0000..U+10ffff). | ||
| 144 | * @return 2 | ||
| 145 | * @stable ICU 2.4 | ||
| 146 | */ | ||
| 147 | #define U16_MAX_LENGTH 2 | ||
| 148 | |||
| 149 | /** | ||
| 150 | * Get a code point from a string at a random-access offset, | ||
| 151 | * without changing the offset. | ||
| 152 | * "Unsafe" macro, assumes well-formed UTF-16. | ||
| 153 | * | ||
| 154 | * The offset may point to either the lead or trail surrogate unit | ||
| 155 | * for a supplementary code point, in which case the macro will read | ||
| 156 | * the adjacent matching surrogate as well. | ||
| 157 | * The result is undefined if the offset points to a single, unpaired surrogate. | ||
| 158 | * Iteration through a string is more efficient with U16_NEXT_UNSAFE or U16_NEXT. | ||
| 159 | * | ||
| 160 | * @param s const UChar * string | ||
| 161 | * @param i string offset | ||
| 162 | * @param c output UChar32 variable | ||
| 163 | * @see U16_GET | ||
| 164 | * @stable ICU 2.4 | ||
| 165 | */ | ||
| 166 | #define U16_GET_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 167 | (c)=(s)[i]; \ | ||
| 168 | if(U16_IS_SURROGATE(c)) { \ | ||
| 169 | if(U16_IS_SURROGATE_LEAD(c)) { \ | ||
| 170 | (c)=U16_GET_SUPPLEMENTARY((c), (s)[(i)+1]); \ | ||
| 171 | } else { \ | ||
| 172 | (c)=U16_GET_SUPPLEMENTARY((s)[(i)-1], (c)); \ | ||
| 173 | } \ | ||
| 174 | } \ | ||
| 175 | } UPRV_BLOCK_MACRO_END | ||
| 176 | |||
| 177 | /** | ||
| 178 | * Get a code point from a string at a random-access offset, | ||
| 179 | * without changing the offset. | ||
| 180 | * "Safe" macro, handles unpaired surrogates and checks for string boundaries. | ||
| 181 | * | ||
| 182 | * The offset may point to either the lead or trail surrogate unit | ||
| 183 | * for a supplementary code point, in which case the macro will read | ||
| 184 | * the adjacent matching surrogate as well. | ||
| 185 | * | ||
| 186 | * The length can be negative for a NUL-terminated string. | ||
| 187 | * | ||
| 188 | * If the offset points to a single, unpaired surrogate, then | ||
| 189 | * c is set to that unpaired surrogate. | ||
| 190 | * Iteration through a string is more efficient with U16_NEXT_UNSAFE or U16_NEXT. | ||
| 191 | * | ||
| 192 | * @param s const UChar * string | ||
| 193 | * @param start starting string offset (usually 0) | ||
| 194 | * @param i string offset, must be start<=i<length | ||
| 195 | * @param length string length | ||
| 196 | * @param c output UChar32 variable | ||
| 197 | * @see U16_GET_UNSAFE | ||
| 198 | * @stable ICU 2.4 | ||
| 199 | */ | ||
| 200 | #define U16_GET(s, start, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 201 | (c)=(s)[i]; \ | ||
| 202 | if(U16_IS_SURROGATE(c)) { \ | ||
| 203 | uint16_t __c2; \ | ||
| 204 | if(U16_IS_SURROGATE_LEAD(c)) { \ | ||
| 205 | if((i)+1!=(length) && U16_IS_TRAIL(__c2=(s)[(i)+1])) { \ | ||
| 206 | (c)=U16_GET_SUPPLEMENTARY((c), __c2); \ | ||
| 207 | } \ | ||
| 208 | } else { \ | ||
| 209 | if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \ | ||
| 210 | (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \ | ||
| 211 | } \ | ||
| 212 | } \ | ||
| 213 | } \ | ||
| 214 | } UPRV_BLOCK_MACRO_END | ||
| 215 | |||
| 216 | /** | ||
| 217 | * Get a code point from a string at a random-access offset, | ||
| 218 | * without changing the offset. | ||
| 219 | * "Safe" macro, handles unpaired surrogates and checks for string boundaries. | ||
| 220 | * | ||
| 221 | * The offset may point to either the lead or trail surrogate unit | ||
| 222 | * for a supplementary code point, in which case the macro will read | ||
| 223 | * the adjacent matching surrogate as well. | ||
| 224 | * | ||
| 225 | * The length can be negative for a NUL-terminated string. | ||
| 226 | * | ||
| 227 | * If the offset points to a single, unpaired surrogate, then | ||
| 228 | * c is set to U+FFFD. | ||
| 229 | * Iteration through a string is more efficient with U16_NEXT_UNSAFE or U16_NEXT_OR_FFFD. | ||
| 230 | * | ||
| 231 | * @param s const UChar * string | ||
| 232 | * @param start starting string offset (usually 0) | ||
| 233 | * @param i string offset, must be start<=i<length | ||
| 234 | * @param length string length | ||
| 235 | * @param c output UChar32 variable | ||
| 236 | * @see U16_GET_UNSAFE | ||
| 237 | * @stable ICU 60 | ||
| 238 | */ | ||
| 239 | #define U16_GET_OR_FFFD(s, start, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 240 | (c)=(s)[i]; \ | ||
| 241 | if(U16_IS_SURROGATE(c)) { \ | ||
| 242 | uint16_t __c2; \ | ||
| 243 | if(U16_IS_SURROGATE_LEAD(c)) { \ | ||
| 244 | if((i)+1!=(length) && U16_IS_TRAIL(__c2=(s)[(i)+1])) { \ | ||
| 245 | (c)=U16_GET_SUPPLEMENTARY((c), __c2); \ | ||
| 246 | } else { \ | ||
| 247 | (c)=0xfffd; \ | ||
| 248 | } \ | ||
| 249 | } else { \ | ||
| 250 | if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \ | ||
| 251 | (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \ | ||
| 252 | } else { \ | ||
| 253 | (c)=0xfffd; \ | ||
| 254 | } \ | ||
| 255 | } \ | ||
| 256 | } \ | ||
| 257 | } UPRV_BLOCK_MACRO_END | ||
| 258 | |||
| 259 | /* definitions with forward iteration --------------------------------------- */ | ||
| 260 | |||
| 261 | /** | ||
| 262 | * Get a code point from a string at a code point boundary offset, | ||
| 263 | * and advance the offset to the next code point boundary. | ||
| 264 | * (Post-incrementing forward iteration.) | ||
| 265 | * "Unsafe" macro, assumes well-formed UTF-16. | ||
| 266 | * | ||
| 267 | * The offset may point to the lead surrogate unit | ||
| 268 | * for a supplementary code point, in which case the macro will read | ||
| 269 | * the following trail surrogate as well. | ||
| 270 | * If the offset points to a trail surrogate, then that itself | ||
| 271 | * will be returned as the code point. | ||
| 272 | * The result is undefined if the offset points to a single, unpaired lead surrogate. | ||
| 273 | * | ||
| 274 | * @param s const UChar * string | ||
| 275 | * @param i string offset | ||
| 276 | * @param c output UChar32 variable | ||
| 277 | * @see U16_NEXT | ||
| 278 | * @stable ICU 2.4 | ||
| 279 | */ | ||
| 280 | #define U16_NEXT_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 281 | (c)=(s)[(i)++]; \ | ||
| 282 | if(U16_IS_LEAD(c)) { \ | ||
| 283 | (c)=U16_GET_SUPPLEMENTARY((c), (s)[(i)++]); \ | ||
| 284 | } \ | ||
| 285 | } UPRV_BLOCK_MACRO_END | ||
| 286 | |||
| 287 | /** | ||
| 288 | * Get a code point from a string at a code point boundary offset, | ||
| 289 | * and advance the offset to the next code point boundary. | ||
| 290 | * (Post-incrementing forward iteration.) | ||
| 291 | * "Safe" macro, handles unpaired surrogates and checks for string boundaries. | ||
| 292 | * | ||
| 293 | * The length can be negative for a NUL-terminated string. | ||
| 294 | * | ||
| 295 | * The offset may point to the lead surrogate unit | ||
| 296 | * for a supplementary code point, in which case the macro will read | ||
| 297 | * the following trail surrogate as well. | ||
| 298 | * If the offset points to a trail surrogate or | ||
| 299 | * to a single, unpaired lead surrogate, then c is set to that unpaired surrogate. | ||
| 300 | * | ||
| 301 | * @param s const UChar * string | ||
| 302 | * @param i string offset, must be i<length | ||
| 303 | * @param length string length | ||
| 304 | * @param c output UChar32 variable | ||
| 305 | * @see U16_NEXT_UNSAFE | ||
| 306 | * @stable ICU 2.4 | ||
| 307 | */ | ||
| 308 | #define U16_NEXT(s, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 309 | (c)=(s)[(i)++]; \ | ||
| 310 | if(U16_IS_LEAD(c)) { \ | ||
| 311 | uint16_t __c2; \ | ||
| 312 | if((i)!=(length) && U16_IS_TRAIL(__c2=(s)[(i)])) { \ | ||
| 313 | ++(i); \ | ||
| 314 | (c)=U16_GET_SUPPLEMENTARY((c), __c2); \ | ||
| 315 | } \ | ||
| 316 | } \ | ||
| 317 | } UPRV_BLOCK_MACRO_END | ||
| 318 | |||
| 319 | /** | ||
| 320 | * Get a code point from a string at a code point boundary offset, | ||
| 321 | * and advance the offset to the next code point boundary. | ||
| 322 | * (Post-incrementing forward iteration.) | ||
| 323 | * "Safe" macro, handles unpaired surrogates and checks for string boundaries. | ||
| 324 | * | ||
| 325 | * The length can be negative for a NUL-terminated string. | ||
| 326 | * | ||
| 327 | * The offset may point to the lead surrogate unit | ||
| 328 | * for a supplementary code point, in which case the macro will read | ||
| 329 | * the following trail surrogate as well. | ||
| 330 | * If the offset points to a trail surrogate or | ||
| 331 | * to a single, unpaired lead surrogate, then c is set to U+FFFD. | ||
| 332 | * | ||
| 333 | * @param s const UChar * string | ||
| 334 | * @param i string offset, must be i<length | ||
| 335 | * @param length string length | ||
| 336 | * @param c output UChar32 variable | ||
| 337 | * @see U16_NEXT_UNSAFE | ||
| 338 | * @stable ICU 60 | ||
| 339 | */ | ||
| 340 | #define U16_NEXT_OR_FFFD(s, i, length, c) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 341 | (c)=(s)[(i)++]; \ | ||
| 342 | if(U16_IS_SURROGATE(c)) { \ | ||
| 343 | uint16_t __c2; \ | ||
| 344 | if(U16_IS_SURROGATE_LEAD(c) && (i)!=(length) && U16_IS_TRAIL(__c2=(s)[(i)])) { \ | ||
| 345 | ++(i); \ | ||
| 346 | (c)=U16_GET_SUPPLEMENTARY((c), __c2); \ | ||
| 347 | } else { \ | ||
| 348 | (c)=0xfffd; \ | ||
| 349 | } \ | ||
| 350 | } \ | ||
| 351 | } UPRV_BLOCK_MACRO_END | ||
| 352 | |||
| 353 | /** | ||
| 354 | * Append a code point to a string, overwriting 1 or 2 code units. | ||
| 355 | * The offset points to the current end of the string contents | ||
| 356 | * and is advanced (post-increment). | ||
| 357 | * "Unsafe" macro, assumes a valid code point and sufficient space in the string. | ||
| 358 | * Otherwise, the result is undefined. | ||
| 359 | * | ||
| 360 | * @param s const UChar * string buffer | ||
| 361 | * @param i string offset | ||
| 362 | * @param c code point to append | ||
| 363 | * @see U16_APPEND | ||
| 364 | * @stable ICU 2.4 | ||
| 365 | */ | ||
| 366 | #define U16_APPEND_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 367 | if((uint32_t)(c)<=0xffff) { \ | ||
| 368 | (s)[(i)++]=(uint16_t)(c); \ | ||
| 369 | } else { \ | ||
| 370 | (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \ | ||
| 371 | (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \ | ||
| 372 | } \ | ||
| 373 | } UPRV_BLOCK_MACRO_END | ||
| 374 | |||
| 375 | /** | ||
| 376 | * Append a code point to a string, overwriting 1 or 2 code units. | ||
| 377 | * The offset points to the current end of the string contents | ||
| 378 | * and is advanced (post-increment). | ||
| 379 | * "Safe" macro, checks for a valid code point. | ||
| 380 | * If a surrogate pair is written, checks for sufficient space in the string. | ||
| 381 | * If the code point is not valid or a trail surrogate does not fit, | ||
| 382 | * then isError is set to TRUE. | ||
| 383 | * | ||
| 384 | * @param s const UChar * string buffer | ||
| 385 | * @param i string offset, must be i<capacity | ||
| 386 | * @param capacity size of the string buffer | ||
| 387 | * @param c code point to append | ||
| 388 | * @param isError output UBool set to TRUE if an error occurs, otherwise not modified | ||
| 389 | * @see U16_APPEND_UNSAFE | ||
| 390 | * @stable ICU 2.4 | ||
| 391 | */ | ||
| 392 | #define U16_APPEND(s, i, capacity, c, isError) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 393 | if((uint32_t)(c)<=0xffff) { \ | ||
| 394 | (s)[(i)++]=(uint16_t)(c); \ | ||
| 395 | } else if((uint32_t)(c)<=0x10ffff && (i)+1<(capacity)) { \ | ||
| 396 | (s)[(i)++]=(uint16_t)(((c)>>10)+0xd7c0); \ | ||
| 397 | (s)[(i)++]=(uint16_t)(((c)&0x3ff)|0xdc00); \ | ||
| 398 | } else /* c>0x10ffff or not enough space */ { \ | ||
| 399 | (isError)=TRUE; \ | ||
| 400 | } \ | ||
| 401 | } UPRV_BLOCK_MACRO_END | ||
| 402 | |||
| 403 | /** | ||
| 404 | * Advance the string offset from one code point boundary to the next. | ||
| 405 | * (Post-incrementing iteration.) | ||
| 406 | * "Unsafe" macro, assumes well-formed UTF-16. | ||
| 407 | * | ||
| 408 | * @param s const UChar * string | ||
| 409 | * @param i string offset | ||
| 410 | * @see U16_FWD_1 | ||
| 411 | * @stable ICU 2.4 | ||
| 412 | */ | ||
| 413 | #define U16_FWD_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 414 | if(U16_IS_LEAD((s)[(i)++])) { \ | ||
| 415 | ++(i); \ | ||
| 416 | } \ | ||
| 417 | } UPRV_BLOCK_MACRO_END | ||
| 418 | |||
| 419 | /** | ||
| 420 | * Advance the string offset from one code point boundary to the next. | ||
| 421 | * (Post-incrementing iteration.) | ||
| 422 | * "Safe" macro, handles unpaired surrogates and checks for string boundaries. | ||
| 423 | * | ||
| 424 | * The length can be negative for a NUL-terminated string. | ||
| 425 | * | ||
| 426 | * @param s const UChar * string | ||
| 427 | * @param i string offset, must be i<length | ||
| 428 | * @param length string length | ||
| 429 | * @see U16_FWD_1_UNSAFE | ||
| 430 | * @stable ICU 2.4 | ||
| 431 | */ | ||
| 432 | #define U16_FWD_1(s, i, length) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 433 | if(U16_IS_LEAD((s)[(i)++]) && (i)!=(length) && U16_IS_TRAIL((s)[i])) { \ | ||
| 434 | ++(i); \ | ||
| 435 | } \ | ||
| 436 | } UPRV_BLOCK_MACRO_END | ||
| 437 | |||
| 438 | /** | ||
| 439 | * Advance the string offset from one code point boundary to the n-th next one, | ||
| 440 | * i.e., move forward by n code points. | ||
| 441 | * (Post-incrementing iteration.) | ||
| 442 | * "Unsafe" macro, assumes well-formed UTF-16. | ||
| 443 | * | ||
| 444 | * @param s const UChar * string | ||
| 445 | * @param i string offset | ||
| 446 | * @param n number of code points to skip | ||
| 447 | * @see U16_FWD_N | ||
| 448 | * @stable ICU 2.4 | ||
| 449 | */ | ||
| 450 | #define U16_FWD_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 451 | int32_t __N=(n); \ | ||
| 452 | while(__N>0) { \ | ||
| 453 | U16_FWD_1_UNSAFE(s, i); \ | ||
| 454 | --__N; \ | ||
| 455 | } \ | ||
| 456 | } UPRV_BLOCK_MACRO_END | ||
| 457 | |||
| 458 | /** | ||
| 459 | * Advance the string offset from one code point boundary to the n-th next one, | ||
| 460 | * i.e., move forward by n code points. | ||
| 461 | * (Post-incrementing iteration.) | ||
| 462 | * "Safe" macro, handles unpaired surrogates and checks for string boundaries. | ||
| 463 | * | ||
| 464 | * The length can be negative for a NUL-terminated string. | ||
| 465 | * | ||
| 466 | * @param s const UChar * string | ||
| 467 | * @param i int32_t string offset, must be i<length | ||
| 468 | * @param length int32_t string length | ||
| 469 | * @param n number of code points to skip | ||
| 470 | * @see U16_FWD_N_UNSAFE | ||
| 471 | * @stable ICU 2.4 | ||
| 472 | */ | ||
| 473 | #define U16_FWD_N(s, i, length, n) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 474 | int32_t __N=(n); \ | ||
| 475 | while(__N>0 && ((i)<(length) || ((length)<0 && (s)[i]!=0))) { \ | ||
| 476 | U16_FWD_1(s, i, length); \ | ||
| 477 | --__N; \ | ||
| 478 | } \ | ||
| 479 | } UPRV_BLOCK_MACRO_END | ||
| 480 | |||
| 481 | /** | ||
| 482 | * Adjust a random-access offset to a code point boundary | ||
| 483 | * at the start of a code point. | ||
| 484 | * If the offset points to the trail surrogate of a surrogate pair, | ||
| 485 | * then the offset is decremented. | ||
| 486 | * Otherwise, it is not modified. | ||
| 487 | * "Unsafe" macro, assumes well-formed UTF-16. | ||
| 488 | * | ||
| 489 | * @param s const UChar * string | ||
| 490 | * @param i string offset | ||
| 491 | * @see U16_SET_CP_START | ||
| 492 | * @stable ICU 2.4 | ||
| 493 | */ | ||
| 494 | #define U16_SET_CP_START_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 495 | if(U16_IS_TRAIL((s)[i])) { \ | ||
| 496 | --(i); \ | ||
| 497 | } \ | ||
| 498 | } UPRV_BLOCK_MACRO_END | ||
| 499 | |||
| 500 | /** | ||
| 501 | * Adjust a random-access offset to a code point boundary | ||
| 502 | * at the start of a code point. | ||
| 503 | * If the offset points to the trail surrogate of a surrogate pair, | ||
| 504 | * then the offset is decremented. | ||
| 505 | * Otherwise, it is not modified. | ||
| 506 | * "Safe" macro, handles unpaired surrogates and checks for string boundaries. | ||
| 507 | * | ||
| 508 | * @param s const UChar * string | ||
| 509 | * @param start starting string offset (usually 0) | ||
| 510 | * @param i string offset, must be start<=i | ||
| 511 | * @see U16_SET_CP_START_UNSAFE | ||
| 512 | * @stable ICU 2.4 | ||
| 513 | */ | ||
| 514 | #define U16_SET_CP_START(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 515 | if(U16_IS_TRAIL((s)[i]) && (i)>(start) && U16_IS_LEAD((s)[(i)-1])) { \ | ||
| 516 | --(i); \ | ||
| 517 | } \ | ||
| 518 | } UPRV_BLOCK_MACRO_END | ||
| 519 | |||
| 520 | /* definitions with backward iteration -------------------------------------- */ | ||
| 521 | |||
| 522 | /** | ||
| 523 | * Move the string offset from one code point boundary to the previous one | ||
| 524 | * and get the code point between them. | ||
| 525 | * (Pre-decrementing backward iteration.) | ||
| 526 | * "Unsafe" macro, assumes well-formed UTF-16. | ||
| 527 | * | ||
| 528 | * The input offset may be the same as the string length. | ||
| 529 | * If the offset is behind a trail surrogate unit | ||
| 530 | * for a supplementary code point, then the macro will read | ||
| 531 | * the preceding lead surrogate as well. | ||
| 532 | * If the offset is behind a lead surrogate, then that itself | ||
| 533 | * will be returned as the code point. | ||
| 534 | * The result is undefined if the offset is behind a single, unpaired trail surrogate. | ||
| 535 | * | ||
| 536 | * @param s const UChar * string | ||
| 537 | * @param i string offset | ||
| 538 | * @param c output UChar32 variable | ||
| 539 | * @see U16_PREV | ||
| 540 | * @stable ICU 2.4 | ||
| 541 | */ | ||
| 542 | #define U16_PREV_UNSAFE(s, i, c) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 543 | (c)=(s)[--(i)]; \ | ||
| 544 | if(U16_IS_TRAIL(c)) { \ | ||
| 545 | (c)=U16_GET_SUPPLEMENTARY((s)[--(i)], (c)); \ | ||
| 546 | } \ | ||
| 547 | } UPRV_BLOCK_MACRO_END | ||
| 548 | |||
| 549 | /** | ||
| 550 | * Move the string offset from one code point boundary to the previous one | ||
| 551 | * and get the code point between them. | ||
| 552 | * (Pre-decrementing backward iteration.) | ||
| 553 | * "Safe" macro, handles unpaired surrogates and checks for string boundaries. | ||
| 554 | * | ||
| 555 | * The input offset may be the same as the string length. | ||
| 556 | * If the offset is behind a trail surrogate unit | ||
| 557 | * for a supplementary code point, then the macro will read | ||
| 558 | * the preceding lead surrogate as well. | ||
| 559 | * If the offset is behind a lead surrogate or behind a single, unpaired | ||
| 560 | * trail surrogate, then c is set to that unpaired surrogate. | ||
| 561 | * | ||
| 562 | * @param s const UChar * string | ||
| 563 | * @param start starting string offset (usually 0) | ||
| 564 | * @param i string offset, must be start<i | ||
| 565 | * @param c output UChar32 variable | ||
| 566 | * @see U16_PREV_UNSAFE | ||
| 567 | * @stable ICU 2.4 | ||
| 568 | */ | ||
| 569 | #define U16_PREV(s, start, i, c) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 570 | (c)=(s)[--(i)]; \ | ||
| 571 | if(U16_IS_TRAIL(c)) { \ | ||
| 572 | uint16_t __c2; \ | ||
| 573 | if((i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \ | ||
| 574 | --(i); \ | ||
| 575 | (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \ | ||
| 576 | } \ | ||
| 577 | } \ | ||
| 578 | } UPRV_BLOCK_MACRO_END | ||
| 579 | |||
| 580 | /** | ||
| 581 | * Move the string offset from one code point boundary to the previous one | ||
| 582 | * and get the code point between them. | ||
| 583 | * (Pre-decrementing backward iteration.) | ||
| 584 | * "Safe" macro, handles unpaired surrogates and checks for string boundaries. | ||
| 585 | * | ||
| 586 | * The input offset may be the same as the string length. | ||
| 587 | * If the offset is behind a trail surrogate unit | ||
| 588 | * for a supplementary code point, then the macro will read | ||
| 589 | * the preceding lead surrogate as well. | ||
| 590 | * If the offset is behind a lead surrogate or behind a single, unpaired | ||
| 591 | * trail surrogate, then c is set to U+FFFD. | ||
| 592 | * | ||
| 593 | * @param s const UChar * string | ||
| 594 | * @param start starting string offset (usually 0) | ||
| 595 | * @param i string offset, must be start<i | ||
| 596 | * @param c output UChar32 variable | ||
| 597 | * @see U16_PREV_UNSAFE | ||
| 598 | * @stable ICU 60 | ||
| 599 | */ | ||
| 600 | #define U16_PREV_OR_FFFD(s, start, i, c) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 601 | (c)=(s)[--(i)]; \ | ||
| 602 | if(U16_IS_SURROGATE(c)) { \ | ||
| 603 | uint16_t __c2; \ | ||
| 604 | if(U16_IS_SURROGATE_TRAIL(c) && (i)>(start) && U16_IS_LEAD(__c2=(s)[(i)-1])) { \ | ||
| 605 | --(i); \ | ||
| 606 | (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \ | ||
| 607 | } else { \ | ||
| 608 | (c)=0xfffd; \ | ||
| 609 | } \ | ||
| 610 | } \ | ||
| 611 | } UPRV_BLOCK_MACRO_END | ||
| 612 | |||
| 613 | /** | ||
| 614 | * Move the string offset from one code point boundary to the previous one. | ||
| 615 | * (Pre-decrementing backward iteration.) | ||
| 616 | * The input offset may be the same as the string length. | ||
| 617 | * "Unsafe" macro, assumes well-formed UTF-16. | ||
| 618 | * | ||
| 619 | * @param s const UChar * string | ||
| 620 | * @param i string offset | ||
| 621 | * @see U16_BACK_1 | ||
| 622 | * @stable ICU 2.4 | ||
| 623 | */ | ||
| 624 | #define U16_BACK_1_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 625 | if(U16_IS_TRAIL((s)[--(i)])) { \ | ||
| 626 | --(i); \ | ||
| 627 | } \ | ||
| 628 | } UPRV_BLOCK_MACRO_END | ||
| 629 | |||
| 630 | /** | ||
| 631 | * Move the string offset from one code point boundary to the previous one. | ||
| 632 | * (Pre-decrementing backward iteration.) | ||
| 633 | * The input offset may be the same as the string length. | ||
| 634 | * "Safe" macro, handles unpaired surrogates and checks for string boundaries. | ||
| 635 | * | ||
| 636 | * @param s const UChar * string | ||
| 637 | * @param start starting string offset (usually 0) | ||
| 638 | * @param i string offset, must be start<i | ||
| 639 | * @see U16_BACK_1_UNSAFE | ||
| 640 | * @stable ICU 2.4 | ||
| 641 | */ | ||
| 642 | #define U16_BACK_1(s, start, i) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 643 | if(U16_IS_TRAIL((s)[--(i)]) && (i)>(start) && U16_IS_LEAD((s)[(i)-1])) { \ | ||
| 644 | --(i); \ | ||
| 645 | } \ | ||
| 646 | } UPRV_BLOCK_MACRO_END | ||
| 647 | |||
| 648 | /** | ||
| 649 | * Move the string offset from one code point boundary to the n-th one before it, | ||
| 650 | * i.e., move backward by n code points. | ||
| 651 | * (Pre-decrementing backward iteration.) | ||
| 652 | * The input offset may be the same as the string length. | ||
| 653 | * "Unsafe" macro, assumes well-formed UTF-16. | ||
| 654 | * | ||
| 655 | * @param s const UChar * string | ||
| 656 | * @param i string offset | ||
| 657 | * @param n number of code points to skip | ||
| 658 | * @see U16_BACK_N | ||
| 659 | * @stable ICU 2.4 | ||
| 660 | */ | ||
| 661 | #define U16_BACK_N_UNSAFE(s, i, n) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 662 | int32_t __N=(n); \ | ||
| 663 | while(__N>0) { \ | ||
| 664 | U16_BACK_1_UNSAFE(s, i); \ | ||
| 665 | --__N; \ | ||
| 666 | } \ | ||
| 667 | } UPRV_BLOCK_MACRO_END | ||
| 668 | |||
| 669 | /** | ||
| 670 | * Move the string offset from one code point boundary to the n-th one before it, | ||
| 671 | * i.e., move backward by n code points. | ||
| 672 | * (Pre-decrementing backward iteration.) | ||
| 673 | * The input offset may be the same as the string length. | ||
| 674 | * "Safe" macro, handles unpaired surrogates and checks for string boundaries. | ||
| 675 | * | ||
| 676 | * @param s const UChar * string | ||
| 677 | * @param start start of string | ||
| 678 | * @param i string offset, must be start<i | ||
| 679 | * @param n number of code points to skip | ||
| 680 | * @see U16_BACK_N_UNSAFE | ||
| 681 | * @stable ICU 2.4 | ||
| 682 | */ | ||
| 683 | #define U16_BACK_N(s, start, i, n) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 684 | int32_t __N=(n); \ | ||
| 685 | while(__N>0 && (i)>(start)) { \ | ||
| 686 | U16_BACK_1(s, start, i); \ | ||
| 687 | --__N; \ | ||
| 688 | } \ | ||
| 689 | } UPRV_BLOCK_MACRO_END | ||
| 690 | |||
| 691 | /** | ||
| 692 | * Adjust a random-access offset to a code point boundary after a code point. | ||
| 693 | * If the offset is behind the lead surrogate of a surrogate pair, | ||
| 694 | * then the offset is incremented. | ||
| 695 | * Otherwise, it is not modified. | ||
| 696 | * The input offset may be the same as the string length. | ||
| 697 | * "Unsafe" macro, assumes well-formed UTF-16. | ||
| 698 | * | ||
| 699 | * @param s const UChar * string | ||
| 700 | * @param i string offset | ||
| 701 | * @see U16_SET_CP_LIMIT | ||
| 702 | * @stable ICU 2.4 | ||
| 703 | */ | ||
| 704 | #define U16_SET_CP_LIMIT_UNSAFE(s, i) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 705 | if(U16_IS_LEAD((s)[(i)-1])) { \ | ||
| 706 | ++(i); \ | ||
| 707 | } \ | ||
| 708 | } UPRV_BLOCK_MACRO_END | ||
| 709 | |||
| 710 | /** | ||
| 711 | * Adjust a random-access offset to a code point boundary after a code point. | ||
| 712 | * If the offset is behind the lead surrogate of a surrogate pair, | ||
| 713 | * then the offset is incremented. | ||
| 714 | * Otherwise, it is not modified. | ||
| 715 | * The input offset may be the same as the string length. | ||
| 716 | * "Safe" macro, handles unpaired surrogates and checks for string boundaries. | ||
| 717 | * | ||
| 718 | * The length can be negative for a NUL-terminated string. | ||
| 719 | * | ||
| 720 | * @param s const UChar * string | ||
| 721 | * @param start int32_t starting string offset (usually 0) | ||
| 722 | * @param i int32_t string offset, start<=i<=length | ||
| 723 | * @param length int32_t string length | ||
| 724 | * @see U16_SET_CP_LIMIT_UNSAFE | ||
| 725 | * @stable ICU 2.4 | ||
| 726 | */ | ||
| 727 | #define U16_SET_CP_LIMIT(s, start, i, length) UPRV_BLOCK_MACRO_BEGIN { \ | ||
| 728 | if((start)<(i) && ((i)<(length) || (length)<0) && U16_IS_LEAD((s)[(i)-1]) && U16_IS_TRAIL((s)[i])) { \ | ||
| 729 | ++(i); \ | ||
| 730 | } \ | ||
| 731 | } UPRV_BLOCK_MACRO_END | ||
| 732 | |||
| 733 | #endif | ||
