icuSources/common/unicode/umachine.h

   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) 1999-2015, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 ******************************************************************************
  10 *   file name:  umachine.h
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 1999sep13
  16 *   created by: Markus W. Scherer
  17 *
  18 *   This file defines basic types and constants for ICU to be
  19 *   platform-independent. umachine.h and utf.h are included into
  20 *   utypes.h to provide all the general definitions for ICU.
  21 *   All of these definitions used to be in utypes.h before
  22 *   the UTF-handling macros made this unmaintainable.
  23 */
  24
  25 #ifndef __UMACHINE_H__
  26 #define __UMACHINE_H__
  27
  28
  29 /**
  30  * \file
  31  * \brief Basic types and constants for UTF
  32  *
  33  * <h2> Basic types and constants for UTF </h2>
  34  *   This file defines basic types and constants for utf.h to be
  35  *   platform-independent. umachine.h and utf.h are included into
  36  *   utypes.h to provide all the general definitions for ICU.
  37  *   All of these definitions used to be in utypes.h before
  38  *   the UTF-handling macros made this unmaintainable.
  39  *
  40  */
  41 /*==========================================================================*/
  42 /* Include platform-dependent definitions                                   */
  43 /* which are contained in the platform-specific file platform.h             */
  44 /*==========================================================================*/
  45
  46 #include "unicode/ptypes.h" /* platform.h is included in ptypes.h */
  47
  48 /*
  49  * ANSI C headers:
  50  * stddef.h defines wchar_t
  51  */
  52 #include <stddef.h>
  53
  54 /*==========================================================================*/
  55 /* For C wrappers, we use the symbol U_STABLE.                                */
  56 /* This works properly if the includer is C or C++.                         */
  57 /* Functions are declared   U_STABLE return-type U_EXPORT2 function-name()... */
  58 /*==========================================================================*/
  59
  60 /**
  61  * \def U_CFUNC
  62  * This is used in a declaration of a library private ICU C function.
  63  * @stable ICU 2.4
  64  */
  65
  66 /**
  67  * \def U_CDECL_BEGIN
  68  * This is used to begin a declaration of a library private ICU C API.
  69  * @stable ICU 2.4
  70  */
  71
  72 /**
  73  * \def U_CDECL_END
  74  * This is used to end a declaration of a library private ICU C API
  75  * @stable ICU 2.4
  76  */
  77
  78 #ifdef __cplusplus
  79 #   define U_CFUNC extern "C"
  80 #   define U_CDECL_BEGIN extern "C" {
  81 #   define U_CDECL_END   }
  82 #else
  83 #   define U_CFUNC extern
  84 #   define U_CDECL_BEGIN
  85 #   define U_CDECL_END
  86 #endif
  87
  88 #ifndef U_ATTRIBUTE_DEPRECATED
  89 /**
  90  * \def U_ATTRIBUTE_DEPRECATED
  91  *  This is used for GCC specific attributes
  92  * @internal
  93  */
  94 #if U_GCC_MAJOR_MINOR >= 302
  95 #    define U_ATTRIBUTE_DEPRECATED __attribute__ ((deprecated))
  96 /**
  97  * \def U_ATTRIBUTE_DEPRECATED
  98  * This is used for Visual C++ specific attributes
  99  * @internal
 100  */
 101 #elif defined(_MSC_VER) && (_MSC_VER >= 1400)
 102 #    define U_ATTRIBUTE_DEPRECATED __declspec(deprecated)
 103 #else
 104 #    define U_ATTRIBUTE_DEPRECATED
 105 #endif
 106 #endif
 107
 108 /** This is used to declare a function as a public ICU C API @stable ICU 2.0*/
 109 #define U_CAPI U_CFUNC U_EXPORT
 110 /** This is used to declare a function as a stable public ICU C API*/
 111 #define U_STABLE U_CAPI
 112 /** This is used to declare a function as a draft public ICU C API  */
 113 #define U_DRAFT  U_CAPI
 114 /** This is used to declare a function as a deprecated public ICU C API  */
 115 #define U_DEPRECATED U_CAPI U_ATTRIBUTE_DEPRECATED
 116 /** This is used to declare a function as an obsolete public ICU C API  */
 117 #define U_OBSOLETE U_CAPI
 118 /** This is used to declare a function as an internal ICU C API  */
 119 #define U_INTERNAL U_CAPI
 120
 121 /**
 122  * \def U_OVERRIDE
 123  * Defined to the C++11 "override" keyword if available.
 124  * Denotes a class or member which is an override of the base class.
 125  * May result in an error if it applied to something not an override.
 126  * @internal
 127  */
 128 #ifndef U_OVERRIDE
 129 #define U_OVERRIDE override
 130 #endif
 131
 132 /**
 133  * \def U_FINAL
 134  * Defined to the C++11 "final" keyword if available.
 135  * Denotes a class or member which may not be overridden in subclasses.
 136  * May result in an error if subclasses attempt to override.
 137  * @internal
 138  */
 139 #if !defined(U_FINAL) || defined(U_IN_DOXYGEN)
 140 #define U_FINAL final
 141 #endif
 142
 143
 144 /*==========================================================================*/
 145 /* limits for int32_t etc., like in POSIX inttypes.h                        */
 146 /*==========================================================================*/
 147
 148 #ifndef INT8_MIN
 149 /** The smallest value an 8 bit signed integer can hold @stable ICU 2.0 */
 150 #   define INT8_MIN        ((int8_t)(-128))
 151 #endif
 152 #ifndef INT16_MIN
 153 /** The smallest value a 16 bit signed integer can hold @stable ICU 2.0 */
 154 #   define INT16_MIN       ((int16_t)(-32767-1))
 155 #endif
 156 #ifndef INT32_MIN
 157 /** The smallest value a 32 bit signed integer can hold @stable ICU 2.0 */
 158 #   define INT32_MIN       ((int32_t)(-2147483647-1))
 159 #endif
 160
 161 #ifndef INT8_MAX
 162 /** The largest value an 8 bit signed integer can hold @stable ICU 2.0 */
 163 #   define INT8_MAX        ((int8_t)(127))
 164 #endif
 165 #ifndef INT16_MAX
 166 /** The largest value a 16 bit signed integer can hold @stable ICU 2.0 */
 167 #   define INT16_MAX       ((int16_t)(32767))
 168 #endif
 169 #ifndef INT32_MAX
 170 /** The largest value a 32 bit signed integer can hold @stable ICU 2.0 */
 171 #   define INT32_MAX       ((int32_t)(2147483647))
 172 #endif
 173
 174 #ifndef UINT8_MAX
 175 /** The largest value an 8 bit unsigned integer can hold @stable ICU 2.0 */
 176 #   define UINT8_MAX       ((uint8_t)(255U))
 177 #endif
 178 #ifndef UINT16_MAX
 179 /** The largest value a 16 bit unsigned integer can hold @stable ICU 2.0 */
 180 #   define UINT16_MAX      ((uint16_t)(65535U))
 181 #endif
 182 #ifndef UINT32_MAX
 183 /** The largest value a 32 bit unsigned integer can hold @stable ICU 2.0 */
 184 #   define UINT32_MAX      ((uint32_t)(4294967295U))
 185 #endif
 186
 187 #if defined(U_INT64_T_UNAVAILABLE)
 188 # error int64_t is required for decimal format and rule-based number format.
 189 #else
 190 # ifndef INT64_C
 191 /**
 192  * Provides a platform independent way to specify a signed 64-bit integer constant.
 193  * note: may be wrong for some 64 bit platforms - ensure your compiler provides INT64_C
 194  * @stable ICU 2.8
 195  */
 196 #   define INT64_C(c) c ## LL
 197 # endif
 198 # ifndef UINT64_C
 199 /**
 200  * Provides a platform independent way to specify an unsigned 64-bit integer constant.
 201  * note: may be wrong for some 64 bit platforms - ensure your compiler provides UINT64_C
 202  * @stable ICU 2.8
 203  */
 204 #   define UINT64_C(c) c ## ULL
 205 # endif
 206 # ifndef U_INT64_MIN
 207 /** The smallest value a 64 bit signed integer can hold @stable ICU 2.8 */
 208 #     define U_INT64_MIN       ((int64_t)(INT64_C(-9223372036854775807)-1))
 209 # endif
 210 # ifndef U_INT64_MAX
 211 /** The largest value a 64 bit signed integer can hold @stable ICU 2.8 */
 212 #     define U_INT64_MAX       ((int64_t)(INT64_C(9223372036854775807)))
 213 # endif
 214 # ifndef U_UINT64_MAX
 215 /** The largest value a 64 bit unsigned integer can hold @stable ICU 2.8 */
 216 #     define U_UINT64_MAX      ((uint64_t)(UINT64_C(18446744073709551615)))
 217 # endif
 218 #endif
 219
 220 /*==========================================================================*/
 221 /* Boolean data type                                                        */
 222 /*==========================================================================*/
 223
 224 /** The ICU boolean type @stable ICU 2.0 */
 225 typedef int8_t UBool;
 226
 227 #ifndef TRUE
 228 /** The TRUE value of a UBool @stable ICU 2.0 */
 229 #   define TRUE  1
 230 #endif
 231 #ifndef FALSE
 232 /** The FALSE value of a UBool @stable ICU 2.0 */
 233 #   define FALSE 0
 234 #endif
 235
 236
 237 /*==========================================================================*/
 238 /* Unicode data types                                                       */
 239 /*==========================================================================*/
 240
 241 /* wchar_t-related definitions -------------------------------------------- */
 242
 243 /*
 244  * \def U_WCHAR_IS_UTF16
 245  * Defined if wchar_t uses UTF-16.
 246  *
 247  * @stable ICU 2.0
 248  */
 249 /*
 250  * \def U_WCHAR_IS_UTF32
 251  * Defined if wchar_t uses UTF-32.
 252  *
 253  * @stable ICU 2.0
 254  */
 255 #if !defined(U_WCHAR_IS_UTF16) && !defined(U_WCHAR_IS_UTF32)
 256 #   ifdef __STDC_ISO_10646__
 257 #       if (U_SIZEOF_WCHAR_T==2)
 258 #           define U_WCHAR_IS_UTF16
 259 #       elif (U_SIZEOF_WCHAR_T==4)
 260 #           define  U_WCHAR_IS_UTF32
 261 #       endif
 262 #   elif defined __UCS2__
 263 #       if (U_PF_OS390 <= U_PLATFORM && U_PLATFORM <= U_PF_OS400) && (U_SIZEOF_WCHAR_T==2)
 264 #           define U_WCHAR_IS_UTF16
 265 #       endif
 266 #   elif defined(__UCS4__) || (U_PLATFORM == U_PF_OS400 && defined(__UTF32__))
 267 #       if (U_SIZEOF_WCHAR_T==4)
 268 #           define U_WCHAR_IS_UTF32
 269 #       endif
 270 #   elif U_PLATFORM_IS_DARWIN_BASED || (U_SIZEOF_WCHAR_T==4 && U_PLATFORM_IS_LINUX_BASED)
 271 #       define U_WCHAR_IS_UTF32
 272 #   elif U_PLATFORM_HAS_WIN32_API
 273 #       define U_WCHAR_IS_UTF16
 274 #   endif
 275 #endif
 276
 277 /* UChar and UChar32 definitions -------------------------------------------- */
 278
 279 /** Number of bytes in a UChar. @stable ICU 2.0 */
 280 #define U_SIZEOF_UCHAR 2
 281
 282 /**
 283  * \def U_CHAR16_IS_TYPEDEF
 284  * If 1, then char16_t is a typedef and not a real type (yet)
 285  * @internal
 286  */
 287 #if (U_PLATFORM == U_PF_AIX) && defined(__cplusplus) &&(U_CPLUSPLUS_VERSION < 11)
 288 // for AIX, uchar.h needs to be included
 289 # include <uchar.h>
 290 # define U_CHAR16_IS_TYPEDEF 1
 291 #elif defined(_MSC_VER) && (_MSC_VER < 1900)
 292 // Versions of Visual Studio/MSVC below 2015 do not support char16_t as a real type,
 293 // and instead use a typedef.  https://msdn.microsoft.com/library/bb531344.aspx
 294 # define U_CHAR16_IS_TYPEDEF 1
 295 #else
 296 # define U_CHAR16_IS_TYPEDEF 0
 297 #endif
 298
 299
 300 /**
 301  * \var UChar
 302  *
 303  * The base type for UTF-16 code units and pointers.
 304  * Unsigned 16-bit integer.
 305  * Starting with ICU 59, C++ API uses char16_t directly, while C API continues to use UChar.
 306  *
 307  * UChar is configurable by defining the macro UCHAR_TYPE
 308  * on the preprocessor or compiler command line:
 309  * -DUCHAR_TYPE=uint16_t or -DUCHAR_TYPE=wchar_t (if U_SIZEOF_WCHAR_T==2) etc.
 310  * (The UCHAR_TYPE can also be \#defined earlier in this file, for outside the ICU library code.)
 311  * This is for transitional use from application code that uses uint16_t or wchar_t for UTF-16.
 312  *
 313  * The default is UChar=char16_t.
 314  *
 315  * C++11 defines char16_t as bit-compatible with uint16_t, but as a distinct type.
 316  *
 317  * In C, char16_t is a simple typedef of uint_least16_t.
 318  * ICU requires uint_least16_t=uint16_t for data memory mapping.
 319  * On macOS, char16_t is not available because the uchar.h standard header is missing.
 320  *
 321  * @stable ICU 4.4
 322  */
 323
 324 #if 0
 325     // #if 1 is normal. UChar defaults to char16_t in C++.
 326     // For configuration testing of UChar=uint16_t temporarily change this to #if 0.
 327     // The intltest Makefile #defines UCHAR_TYPE=char16_t,
 328     // so we only #define it to uint16_t if it is undefined so far.
 329 #elif !defined(UCHAR_TYPE)
 330 #   define UCHAR_TYPE uint16_t
 331 #endif
 332
 333 #if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || \
 334         defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION)
 335     // Inside the ICU library code, never configurable.
 336     typedef char16_t UChar;
 337 #elif defined(UCHAR_TYPE)
 338     typedef UCHAR_TYPE UChar;
 339 #elif defined(__cplusplus)
 340     typedef char16_t UChar;
 341 #else
 342     typedef uint16_t UChar;
 343 #endif
 344
 345 /**
 346  * \var OldUChar
 347  * Default ICU 58 definition of UChar.
 348  * A base type for UTF-16 code units and pointers.
 349  * Unsigned 16-bit integer.
 350  *
 351  * Define OldUChar to be wchar_t if that is 16 bits wide.
 352  * If wchar_t is not 16 bits wide, then define UChar to be uint16_t.
 353  *
 354  * This makes the definition of OldUChar platform-dependent
 355  * but allows direct string type compatibility with platforms with
 356  * 16-bit wchar_t types.
 357  *
 358  * This is how UChar was defined in ICU 58, for transition convenience.
 359  * Exception: ICU 58 UChar was defined to UCHAR_TYPE if that macro was defined.
 360  * The current UChar responds to UCHAR_TYPE but OldUChar does not.
 361  *
 362  * @stable ICU 59
 363  */
 364 #if U_SIZEOF_WCHAR_T==2
 365     typedef wchar_t OldUChar;
 366 #elif defined(__CHAR16_TYPE__)
 367     typedef __CHAR16_TYPE__ OldUChar;
 368 #else
 369     typedef uint16_t OldUChar;
 370 #endif
 371
 372 /**
 373  * Define UChar32 as a type for single Unicode code points.
 374  * UChar32 is a signed 32-bit integer (same as int32_t).
 375  *
 376  * The Unicode code point range is 0..0x10ffff.
 377  * All other values (negative or >=0x110000) are illegal as Unicode code points.
 378  * They may be used as sentinel values to indicate "done", "error"
 379  * or similar non-code point conditions.
 380  *
 381  * Before ICU 2.4 (Jitterbug 2146), UChar32 was defined
 382  * to be wchar_t if that is 32 bits wide (wchar_t may be signed or unsigned)
 383  * or else to be uint32_t.
 384  * That is, the definition of UChar32 was platform-dependent.
 385  *
 386  * @see U_SENTINEL
 387  * @stable ICU 2.4
 388  */
 389 typedef int32_t UChar32;
 390
 391 /**
 392  * This value is intended for sentinel values for APIs that
 393  * (take or) return single code points (UChar32).
 394  * It is outside of the Unicode code point range 0..0x10ffff.
 395  *
 396  * For example, a "done" or "error" value in a new API
 397  * could be indicated with U_SENTINEL.
 398  *
 399  * ICU APIs designed before ICU 2.4 usually define service-specific "done"
 400  * values, mostly 0xffff.
 401  * Those may need to be distinguished from
 402  * actual U+ffff text contents by calling functions like
 403  * CharacterIterator::hasNext() or UnicodeString::length().
 404  *
 405  * @return -1
 406  * @see UChar32
 407  * @stable ICU 2.4
 408  */
 409 #define U_SENTINEL (-1)
 410
 411 #include "unicode/urename.h"
 412
 413 #endif