[1] | 1 | /** @file
|
---|
[8245] | 2 | * IPRT - Unicode Code Points.
|
---|
[1] | 3 | */
|
---|
| 4 |
|
---|
| 5 | /*
|
---|
[76553] | 6 | * Copyright (C) 2006-2019 Oracle Corporation
|
---|
[1] | 7 | *
|
---|
| 8 | * This file is part of VirtualBox Open Source Edition (OSE), as
|
---|
| 9 | * available from http://www.virtualbox.org. This file is free software;
|
---|
| 10 | * you can redistribute it and/or modify it under the terms of the GNU
|
---|
[5999] | 11 | * General Public License (GPL) as published by the Free Software
|
---|
| 12 | * Foundation, in version 2 as it comes in the "COPYING" file of the
|
---|
| 13 | * VirtualBox OSE distribution. VirtualBox OSE is distributed in the
|
---|
| 14 | * hope that it will be useful, but WITHOUT ANY WARRANTY of any kind.
|
---|
| 15 | *
|
---|
| 16 | * The contents of this file may alternatively be used under the terms
|
---|
| 17 | * of the Common Development and Distribution License Version 1.0
|
---|
| 18 | * (CDDL) only, as it comes in the "COPYING.CDDL" file of the
|
---|
| 19 | * VirtualBox OSE distribution, in which case the provisions of the
|
---|
| 20 | * CDDL are applicable instead of those of the GPL.
|
---|
| 21 | *
|
---|
| 22 | * You may elect to license modified versions of this file under the
|
---|
| 23 | * terms and conditions of either the GPL or the CDDL or both.
|
---|
[1] | 24 | */
|
---|
| 25 |
|
---|
[76557] | 26 | #ifndef IPRT_INCLUDED_uni_h
|
---|
| 27 | #define IPRT_INCLUDED_uni_h
|
---|
[76507] | 28 | #ifndef RT_WITHOUT_PRAGMA_ONCE
|
---|
| 29 | # pragma once
|
---|
| 30 | #endif
|
---|
[1] | 31 |
|
---|
| 32 | /** @defgroup grp_rt_uni RTUniCp - Unicode Code Points
|
---|
| 33 | * @ingroup grp_rt
|
---|
| 34 | * @{
|
---|
| 35 | */
|
---|
| 36 |
|
---|
| 37 | /** @def RTUNI_USE_WCTYPE
|
---|
| 38 | * Define RTUNI_USE_WCTYPE to not use the IPRT unicode data but the
|
---|
| 39 | * data which the C runtime library provides. */
|
---|
[7170] | 40 | #ifdef DOXYGEN_RUNNING
|
---|
[1] | 41 | # define RTUNI_USE_WCTYPE
|
---|
| 42 | #endif
|
---|
| 43 |
|
---|
| 44 | #include <iprt/types.h>
|
---|
| 45 | #ifdef RTUNI_USE_WCTYPE
|
---|
| 46 | # include <wctype.h>
|
---|
| 47 | #endif
|
---|
| 48 |
|
---|
[20374] | 49 | RT_C_DECLS_BEGIN
|
---|
[1] | 50 |
|
---|
| 51 |
|
---|
[46011] | 52 | #ifndef RTUNI_USE_WCTYPE
|
---|
[1] | 53 |
|
---|
| 54 | /**
|
---|
| 55 | * A unicode flags range.
|
---|
| 56 | * @internal
|
---|
| 57 | */
|
---|
| 58 | typedef struct RTUNIFLAGSRANGE
|
---|
| 59 | {
|
---|
| 60 | /** The first code point of the range. */
|
---|
| 61 | RTUNICP BeginCP;
|
---|
| 62 | /** The last + 1 code point of the range. */
|
---|
| 63 | RTUNICP EndCP;
|
---|
| 64 | /** Pointer to the array of case folded code points. */
|
---|
| 65 | const uint8_t *pafFlags;
|
---|
| 66 | } RTUNIFLAGSRANGE;
|
---|
| 67 | /** Pointer to a flags range.
|
---|
| 68 | * @internal */
|
---|
| 69 | typedef RTUNIFLAGSRANGE *PRTUNIFLAGSRANGE;
|
---|
| 70 | /** Pointer to a const flags range.
|
---|
| 71 | * @internal */
|
---|
| 72 | typedef const RTUNIFLAGSRANGE *PCRTUNIFLAGSRANGE;
|
---|
| 73 |
|
---|
| 74 | /**
|
---|
| 75 | * A unicode case folded range.
|
---|
| 76 | * @internal
|
---|
| 77 | */
|
---|
| 78 | typedef struct RTUNICASERANGE
|
---|
| 79 | {
|
---|
| 80 | /** The first code point of the range. */
|
---|
| 81 | RTUNICP BeginCP;
|
---|
| 82 | /** The last + 1 code point of the range. */
|
---|
| 83 | RTUNICP EndCP;
|
---|
| 84 | /** Pointer to the array of case folded code points. */
|
---|
| 85 | PCRTUNICP paFoldedCPs;
|
---|
| 86 | } RTUNICASERANGE;
|
---|
| 87 | /** Pointer to a case folded range.
|
---|
| 88 | * @internal */
|
---|
| 89 | typedef RTUNICASERANGE *PRTUNICASERANGE;
|
---|
| 90 | /** Pointer to a const case folded range.
|
---|
| 91 | * @internal */
|
---|
| 92 | typedef const RTUNICASERANGE *PCRTUNICASERANGE;
|
---|
| 93 |
|
---|
| 94 | /** @name Unicode Code Point Flags.
|
---|
| 95 | * @internal
|
---|
| 96 | * @{ */
|
---|
[28878] | 97 | #define RTUNI_UPPER RT_BIT(0)
|
---|
| 98 | #define RTUNI_LOWER RT_BIT(1)
|
---|
| 99 | #define RTUNI_ALPHA RT_BIT(2)
|
---|
| 100 | #define RTUNI_XDIGIT RT_BIT(3)
|
---|
| 101 | #define RTUNI_DDIGIT RT_BIT(4)
|
---|
| 102 | #define RTUNI_WSPACE RT_BIT(5)
|
---|
[5605] | 103 | /*#define RTUNI_BSPACE RT_BIT(6) - later */
|
---|
[28878] | 104 | /** When set, the codepoint requires further checking wrt NFC and NFD
|
---|
| 105 | * normalization. I.e. set when either of QC_NFD and QC_NFC are not Y. */
|
---|
| 106 | #define RTUNI_QC_NFX RT_BIT(7)
|
---|
[1] | 107 | /** @} */
|
---|
| 108 |
|
---|
| 109 |
|
---|
| 110 | /**
|
---|
| 111 | * Array of flags ranges.
|
---|
| 112 | * @internal
|
---|
| 113 | */
|
---|
| 114 | extern RTDATADECL(const RTUNIFLAGSRANGE) g_aRTUniFlagsRanges[];
|
---|
| 115 |
|
---|
| 116 | /**
|
---|
| 117 | * Gets the flags for a unicode code point.
|
---|
| 118 | *
|
---|
| 119 | * @returns The flag mask. (RTUNI_*)
|
---|
| 120 | * @param CodePoint The unicode code point.
|
---|
| 121 | * @internal
|
---|
| 122 | */
|
---|
| 123 | DECLINLINE(RTUNICP) rtUniCpFlags(RTUNICP CodePoint)
|
---|
| 124 | {
|
---|
| 125 | PCRTUNIFLAGSRANGE pCur = &g_aRTUniFlagsRanges[0];
|
---|
| 126 | do
|
---|
| 127 | {
|
---|
| 128 | if (pCur->EndCP > CodePoint)
|
---|
| 129 | {
|
---|
| 130 | if (pCur->BeginCP <= CodePoint)
|
---|
[33560] | 131 | return pCur->pafFlags[CodePoint - pCur->BeginCP];
|
---|
[1] | 132 | break;
|
---|
| 133 | }
|
---|
| 134 | pCur++;
|
---|
| 135 | } while (pCur->EndCP != RTUNICP_MAX);
|
---|
[33560] | 136 | return 0;
|
---|
[1] | 137 | }
|
---|
| 138 |
|
---|
| 139 |
|
---|
| 140 | /**
|
---|
| 141 | * Checks if a unicode code point is upper case.
|
---|
| 142 | *
|
---|
| 143 | * @returns true if it is.
|
---|
| 144 | * @returns false if it isn't.
|
---|
| 145 | * @param CodePoint The code point.
|
---|
| 146 | */
|
---|
| 147 | DECLINLINE(bool) RTUniCpIsUpper(RTUNICP CodePoint)
|
---|
| 148 | {
|
---|
| 149 | return (rtUniCpFlags(CodePoint) & RTUNI_UPPER) != 0;
|
---|
| 150 | }
|
---|
| 151 |
|
---|
| 152 |
|
---|
| 153 | /**
|
---|
| 154 | * Checks if a unicode code point is lower case.
|
---|
| 155 | *
|
---|
| 156 | * @returns true if it is.
|
---|
| 157 | * @returns false if it isn't.
|
---|
| 158 | * @param CodePoint The code point.
|
---|
| 159 | */
|
---|
| 160 | DECLINLINE(bool) RTUniCpIsLower(RTUNICP CodePoint)
|
---|
| 161 | {
|
---|
| 162 | return (rtUniCpFlags(CodePoint) & RTUNI_LOWER) != 0;
|
---|
| 163 | }
|
---|
| 164 |
|
---|
| 165 |
|
---|
| 166 | /**
|
---|
[46011] | 167 | * Checks if a unicode code point is case foldable.
|
---|
| 168 | *
|
---|
| 169 | * @returns true if it is.
|
---|
| 170 | * @returns false if it isn't.
|
---|
| 171 | * @param CodePoint The code point.
|
---|
| 172 | */
|
---|
| 173 | DECLINLINE(bool) RTUniCpIsFoldable(RTUNICP CodePoint)
|
---|
| 174 | {
|
---|
| 175 | /* Right enough. */
|
---|
| 176 | return (rtUniCpFlags(CodePoint) & (RTUNI_LOWER | RTUNI_UPPER)) != 0;
|
---|
| 177 | }
|
---|
| 178 |
|
---|
| 179 |
|
---|
| 180 | /**
|
---|
[1] | 181 | * Checks if a unicode code point is alphabetic.
|
---|
| 182 | *
|
---|
| 183 | * @returns true if it is.
|
---|
| 184 | * @returns false if it isn't.
|
---|
| 185 | * @param CodePoint The code point.
|
---|
| 186 | */
|
---|
| 187 | DECLINLINE(bool) RTUniCpIsAlphabetic(RTUNICP CodePoint)
|
---|
| 188 | {
|
---|
| 189 | return (rtUniCpFlags(CodePoint) & RTUNI_ALPHA) != 0;
|
---|
| 190 | }
|
---|
| 191 |
|
---|
| 192 |
|
---|
| 193 | /**
|
---|
| 194 | * Checks if a unicode code point is a decimal digit.
|
---|
| 195 | *
|
---|
| 196 | * @returns true if it is.
|
---|
| 197 | * @returns false if it isn't.
|
---|
| 198 | * @param CodePoint The code point.
|
---|
| 199 | */
|
---|
| 200 | DECLINLINE(bool) RTUniCpIsDecDigit(RTUNICP CodePoint)
|
---|
| 201 | {
|
---|
| 202 | return (rtUniCpFlags(CodePoint) & RTUNI_DDIGIT) != 0;
|
---|
| 203 | }
|
---|
| 204 |
|
---|
| 205 |
|
---|
| 206 | /**
|
---|
| 207 | * Checks if a unicode code point is a hexadecimal digit.
|
---|
| 208 | *
|
---|
| 209 | * @returns true if it is.
|
---|
| 210 | * @returns false if it isn't.
|
---|
| 211 | * @param CodePoint The code point.
|
---|
| 212 | */
|
---|
| 213 | DECLINLINE(bool) RTUniCpIsHexDigit(RTUNICP CodePoint)
|
---|
| 214 | {
|
---|
| 215 | return (rtUniCpFlags(CodePoint) & RTUNI_XDIGIT) != 0;
|
---|
| 216 | }
|
---|
| 217 |
|
---|
| 218 |
|
---|
| 219 | /**
|
---|
| 220 | * Checks if a unicode code point is white space.
|
---|
| 221 | *
|
---|
| 222 | * @returns true if it is.
|
---|
| 223 | * @returns false if it isn't.
|
---|
| 224 | * @param CodePoint The code point.
|
---|
| 225 | */
|
---|
| 226 | DECLINLINE(bool) RTUniCpIsSpace(RTUNICP CodePoint)
|
---|
| 227 | {
|
---|
| 228 | return (rtUniCpFlags(CodePoint) & RTUNI_WSPACE) != 0;
|
---|
| 229 | }
|
---|
| 230 |
|
---|
| 231 |
|
---|
| 232 |
|
---|
| 233 | /**
|
---|
| 234 | * Array of uppercase ranges.
|
---|
| 235 | * @internal
|
---|
| 236 | */
|
---|
| 237 | extern RTDATADECL(const RTUNICASERANGE) g_aRTUniUpperRanges[];
|
---|
| 238 |
|
---|
| 239 | /**
|
---|
| 240 | * Array of lowercase ranges.
|
---|
| 241 | * @internal
|
---|
| 242 | */
|
---|
| 243 | extern RTDATADECL(const RTUNICASERANGE) g_aRTUniLowerRanges[];
|
---|
| 244 |
|
---|
| 245 |
|
---|
| 246 | /**
|
---|
| 247 | * Folds a unicode code point using the specified range array.
|
---|
| 248 | *
|
---|
| 249 | * @returns FOlded code point.
|
---|
| 250 | * @param CodePoint The unicode code point to fold.
|
---|
| 251 | * @param pCur The case folding range to use.
|
---|
| 252 | */
|
---|
| 253 | DECLINLINE(RTUNICP) rtUniCpFold(RTUNICP CodePoint, PCRTUNICASERANGE pCur)
|
---|
| 254 | {
|
---|
| 255 | do
|
---|
| 256 | {
|
---|
| 257 | if (pCur->EndCP > CodePoint)
|
---|
| 258 | {
|
---|
| 259 | if (pCur->BeginCP <= CodePoint)
|
---|
| 260 | CodePoint = pCur->paFoldedCPs[CodePoint - pCur->BeginCP];
|
---|
| 261 | break;
|
---|
| 262 | }
|
---|
| 263 | pCur++;
|
---|
| 264 | } while (pCur->EndCP != RTUNICP_MAX);
|
---|
| 265 | return CodePoint;
|
---|
| 266 | }
|
---|
| 267 |
|
---|
| 268 |
|
---|
| 269 | /**
|
---|
| 270 | * Folds a unicode code point to upper case.
|
---|
| 271 | *
|
---|
| 272 | * @returns Folded code point.
|
---|
| 273 | * @param CodePoint The unicode code point to fold.
|
---|
| 274 | */
|
---|
| 275 | DECLINLINE(RTUNICP) RTUniCpToUpper(RTUNICP CodePoint)
|
---|
| 276 | {
|
---|
| 277 | return rtUniCpFold(CodePoint, &g_aRTUniUpperRanges[0]);
|
---|
| 278 | }
|
---|
| 279 |
|
---|
| 280 |
|
---|
| 281 | /**
|
---|
| 282 | * Folds a unicode code point to lower case.
|
---|
| 283 | *
|
---|
| 284 | * @returns Folded code point.
|
---|
| 285 | * @param CodePoint The unicode code point to fold.
|
---|
| 286 | */
|
---|
| 287 | DECLINLINE(RTUNICP) RTUniCpToLower(RTUNICP CodePoint)
|
---|
| 288 | {
|
---|
| 289 | return rtUniCpFold(CodePoint, &g_aRTUniLowerRanges[0]);
|
---|
| 290 | }
|
---|
| 291 |
|
---|
| 292 |
|
---|
| 293 | #else /* RTUNI_USE_WCTYPE */
|
---|
| 294 |
|
---|
| 295 |
|
---|
| 296 | /**
|
---|
| 297 | * Checks if a unicode code point is upper case.
|
---|
| 298 | *
|
---|
| 299 | * @returns true if it is.
|
---|
| 300 | * @returns false if it isn't.
|
---|
| 301 | * @param CodePoint The code point.
|
---|
| 302 | */
|
---|
| 303 | DECLINLINE(bool) RTUniCpIsUpper(RTUNICP CodePoint)
|
---|
| 304 | {
|
---|
| 305 | return !!iswupper(CodePoint);
|
---|
| 306 | }
|
---|
| 307 |
|
---|
| 308 |
|
---|
| 309 | /**
|
---|
| 310 | * Checks if a unicode code point is lower case.
|
---|
| 311 | *
|
---|
| 312 | * @returns true if it is.
|
---|
| 313 | * @returns false if it isn't.
|
---|
| 314 | * @param CodePoint The code point.
|
---|
| 315 | */
|
---|
| 316 | DECLINLINE(bool) RTUniCpIsLower(RTUNICP CodePoint)
|
---|
| 317 | {
|
---|
| 318 | return !!iswlower(CodePoint);
|
---|
| 319 | }
|
---|
| 320 |
|
---|
| 321 |
|
---|
| 322 | /**
|
---|
[46011] | 323 | * Checks if a unicode code point is case foldable.
|
---|
| 324 | *
|
---|
| 325 | * @returns true if it is.
|
---|
| 326 | * @returns false if it isn't.
|
---|
| 327 | * @param CodePoint The code point.
|
---|
| 328 | */
|
---|
| 329 | DECLINLINE(bool) RTUniCpIsFoldable(RTUNICP CodePoint)
|
---|
| 330 | {
|
---|
| 331 | /* Right enough. */
|
---|
| 332 | return iswupper(CodePoint) || iswlower(CodePoint);
|
---|
| 333 | }
|
---|
| 334 |
|
---|
| 335 |
|
---|
| 336 | /**
|
---|
[1] | 337 | * Checks if a unicode code point is alphabetic.
|
---|
| 338 | *
|
---|
| 339 | * @returns true if it is.
|
---|
| 340 | * @returns false if it isn't.
|
---|
| 341 | * @param CodePoint The code point.
|
---|
| 342 | */
|
---|
| 343 | DECLINLINE(bool) RTUniCpIsAlphabetic(RTUNICP CodePoint)
|
---|
| 344 | {
|
---|
| 345 | return !!iswalpha(CodePoint);
|
---|
| 346 | }
|
---|
| 347 |
|
---|
| 348 |
|
---|
| 349 | /**
|
---|
| 350 | * Checks if a unicode code point is a decimal digit.
|
---|
| 351 | *
|
---|
| 352 | * @returns true if it is.
|
---|
| 353 | * @returns false if it isn't.
|
---|
| 354 | * @param CodePoint The code point.
|
---|
| 355 | */
|
---|
| 356 | DECLINLINE(bool) RTUniCpIsDecDigit(RTUNICP CodePoint)
|
---|
| 357 | {
|
---|
| 358 | return !!iswdigit(CodePoint);
|
---|
| 359 | }
|
---|
| 360 |
|
---|
| 361 |
|
---|
| 362 | /**
|
---|
| 363 | * Checks if a unicode code point is a hexadecimal digit.
|
---|
| 364 | *
|
---|
| 365 | * @returns true if it is.
|
---|
| 366 | * @returns false if it isn't.
|
---|
| 367 | * @param CodePoint The code point.
|
---|
| 368 | */
|
---|
| 369 | DECLINLINE(bool) RTUniCpIsHexDigit(RTUNICP CodePoint)
|
---|
| 370 | {
|
---|
| 371 | return !!iswxdigit(CodePoint);
|
---|
| 372 | }
|
---|
| 373 |
|
---|
| 374 |
|
---|
| 375 | /**
|
---|
| 376 | * Checks if a unicode code point is white space.
|
---|
| 377 | *
|
---|
| 378 | * @returns true if it is.
|
---|
| 379 | * @returns false if it isn't.
|
---|
| 380 | * @param CodePoint The code point.
|
---|
| 381 | */
|
---|
| 382 | DECLINLINE(bool) RTUniCpIsSpace(RTUNICP CodePoint)
|
---|
| 383 | {
|
---|
| 384 | return !!iswspace(CodePoint);
|
---|
| 385 | }
|
---|
| 386 |
|
---|
| 387 |
|
---|
| 388 | /**
|
---|
| 389 | * Folds a unicode code point to upper case.
|
---|
| 390 | *
|
---|
| 391 | * @returns Folded code point.
|
---|
| 392 | * @param CodePoint The unicode code point to fold.
|
---|
| 393 | */
|
---|
| 394 | DECLINLINE(RTUNICP) RTUniCpToUpper(RTUNICP CodePoint)
|
---|
| 395 | {
|
---|
| 396 | return towupper(CodePoint);
|
---|
| 397 | }
|
---|
| 398 |
|
---|
| 399 |
|
---|
| 400 | /**
|
---|
| 401 | * Folds a unicode code point to lower case.
|
---|
| 402 | *
|
---|
| 403 | * @returns Folded code point.
|
---|
| 404 | * @param CodePoint The unicode code point to fold.
|
---|
| 405 | */
|
---|
| 406 | DECLINLINE(RTUNICP) RTUniCpToLower(RTUNICP CodePoint)
|
---|
| 407 | {
|
---|
| 408 | return towlower(CodePoint);
|
---|
| 409 | }
|
---|
| 410 |
|
---|
| 411 |
|
---|
| 412 | #endif /* RTUNI_USE_WCTYPE */
|
---|
| 413 |
|
---|
| 414 |
|
---|
| 415 | /**
|
---|
| 416 | * Frees a unicode string.
|
---|
| 417 | *
|
---|
| 418 | * @param pusz The string to free.
|
---|
| 419 | */
|
---|
| 420 | RTDECL(void) RTUniFree(PRTUNICP pusz);
|
---|
| 421 |
|
---|
| 422 |
|
---|
[51770] | 423 | /**
|
---|
| 424 | * Checks if a code point valid.
|
---|
| 425 | *
|
---|
| 426 | * Any code point (defined or not) within the 17 unicode planes (0 thru 16),
|
---|
| 427 | * except surrogates will be considered valid code points by this function.
|
---|
| 428 | *
|
---|
| 429 | * @returns true if in range, false if not.
|
---|
| 430 | * @param CodePoint The unicode code point to validate.
|
---|
| 431 | */
|
---|
| 432 | DECLINLINE(bool) RTUniCpIsValid(RTUNICP CodePoint)
|
---|
| 433 | {
|
---|
| 434 | return CodePoint <= 0x00d7ff
|
---|
| 435 | || ( CodePoint <= 0x10ffff
|
---|
| 436 | && CodePoint >= 0x00e000);
|
---|
| 437 | }
|
---|
| 438 |
|
---|
| 439 |
|
---|
| 440 | /**
|
---|
| 441 | * Checks if the given code point is in the BMP range.
|
---|
| 442 | *
|
---|
| 443 | * Surrogates are not considered in the BMP range by this function.
|
---|
| 444 | *
|
---|
| 445 | * @returns true if in BMP, false if not.
|
---|
| 446 | * @param CodePoint The unicode code point to consider.
|
---|
| 447 | */
|
---|
| 448 | DECLINLINE(bool) RTUniCpIsBMP(RTUNICP CodePoint)
|
---|
| 449 | {
|
---|
| 450 | return CodePoint <= 0xd7ff
|
---|
| 451 | || ( CodePoint <= 0xffff
|
---|
| 452 | && CodePoint >= 0xe000);
|
---|
| 453 | }
|
---|
| 454 |
|
---|
| 455 |
|
---|
| 456 | /**
|
---|
| 457 | * Folds a unicode code point to lower case.
|
---|
| 458 | *
|
---|
| 459 | * @returns Folded code point.
|
---|
| 460 | * @param CodePoint The unicode code point to fold.
|
---|
| 461 | */
|
---|
| 462 | DECLINLINE(size_t) RTUniCpCalcUtf8Len(RTUNICP CodePoint)
|
---|
| 463 | {
|
---|
| 464 | if (CodePoint < 0x80)
|
---|
| 465 | return 1;
|
---|
| 466 | return 2
|
---|
| 467 | + (CodePoint >= 0x00000800)
|
---|
| 468 | + (CodePoint >= 0x00010000)
|
---|
| 469 | + (CodePoint >= 0x00200000)
|
---|
| 470 | + (CodePoint >= 0x04000000)
|
---|
| 471 | + (CodePoint >= 0x80000000) /* illegal */;
|
---|
| 472 | }
|
---|
| 473 |
|
---|
| 474 |
|
---|
| 475 |
|
---|
[20374] | 476 | RT_C_DECLS_END
|
---|
[1] | 477 | /** @} */
|
---|
| 478 |
|
---|
| 479 |
|
---|
[76585] | 480 | #endif /* !IPRT_INCLUDED_uni_h */
|
---|
[1] | 481 |
|
---|