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
 129 /**
 130  * \def U_FINAL
 131  * Defined to the C++11 "final" keyword if available.
 132  * Denotes a class or member which may not be overridden in subclasses.
 133  * May result in an error if subclasses attempt to override.
 134  * @internal
 135  */
 136
 137 #if U_CPLUSPLUS_VERSION >= 11
 138 /* C++11 */
 139 #ifndef U_OVERRIDE
 140 #define U_OVERRIDE override
 141 #endif
 142 #ifndef U_FINAL
 143 #define U_FINAL final
 144 #endif
 145 #else
 146 /* not C++11 - define to nothing */
 147 #ifndef U_OVERRIDE
 148 #define U_OVERRIDE
 149 #endif
 150 #ifndef U_FINAL
 151 #define U_FINAL
 152 #endif
 153 #endif
 154
 155 /*==========================================================================*/
 156 /* limits for int32_t etc., like in POSIX inttypes.h                        */
 157 /*==========================================================================*/
 158
 159 #ifndef INT8_MIN
 160 /** The smallest value an 8 bit signed integer can hold @stable ICU 2.0 */
 161 #   define INT8_MIN        ((int8_t)(-128))
 162 #endif
 163 #ifndef INT16_MIN
 164 /** The smallest value a 16 bit signed integer can hold @stable ICU 2.0 */
 165 #   define INT16_MIN       ((int16_t)(-32767-1))
 166 #endif
 167 #ifndef INT32_MIN
 168 /** The smallest value a 32 bit signed integer can hold @stable ICU 2.0 */
 169 #   define INT32_MIN       ((int32_t)(-2147483647-1))
 170 #endif
 171
 172 #ifndef INT8_MAX
 173 /** The largest value an 8 bit signed integer can hold @stable ICU 2.0 */
 174 #   define INT8_MAX        ((int8_t)(127))
 175 #endif
 176 #ifndef INT16_MAX
 177 /** The largest value a 16 bit signed integer can hold @stable ICU 2.0 */
 178 #   define INT16_MAX       ((int16_t)(32767))
 179 #endif
 180 #ifndef INT32_MAX
 181 /** The largest value a 32 bit signed integer can hold @stable ICU 2.0 */
 182 #   define INT32_MAX       ((int32_t)(2147483647))
 183 #endif
 184
 185 #ifndef UINT8_MAX
 186 /** The largest value an 8 bit unsigned integer can hold @stable ICU 2.0 */
 187 #   define UINT8_MAX       ((uint8_t)(255U))
 188 #endif
 189 #ifndef UINT16_MAX
 190 /** The largest value a 16 bit unsigned integer can hold @stable ICU 2.0 */
 191 #   define UINT16_MAX      ((uint16_t)(65535U))
 192 #endif
 193 #ifndef UINT32_MAX
 194 /** The largest value a 32 bit unsigned integer can hold @stable ICU 2.0 */
 195 #   define UINT32_MAX      ((uint32_t)(4294967295U))
 196 #endif
 197
 198 #if defined(U_INT64_T_UNAVAILABLE)
 199 # error int64_t is required for decimal format and rule-based number format.
 200 #else
 201 # ifndef INT64_C
 202 /**
 203  * Provides a platform independent way to specify a signed 64-bit integer constant.
 204  * note: may be wrong for some 64 bit platforms - ensure your compiler provides INT64_C
 205  * @stable ICU 2.8
 206  */
 207 #   define INT64_C(c) c ## LL
 208 # endif
 209 # ifndef UINT64_C
 210 /**
 211  * Provides a platform independent way to specify an unsigned 64-bit integer constant.
 212  * note: may be wrong for some 64 bit platforms - ensure your compiler provides UINT64_C
 213  * @stable ICU 2.8
 214  */
 215 #   define UINT64_C(c) c ## ULL
 216 # endif
 217 # ifndef U_INT64_MIN
 218 /** The smallest value a 64 bit signed integer can hold @stable ICU 2.8 */
 219 #     define U_INT64_MIN       ((int64_t)(INT64_C(-9223372036854775807)-1))
 220 # endif
 221 # ifndef U_INT64_MAX
 222 /** The largest value a 64 bit signed integer can hold @stable ICU 2.8 */
 223 #     define U_INT64_MAX       ((int64_t)(INT64_C(9223372036854775807)))
 224 # endif
 225 # ifndef U_UINT64_MAX
 226 /** The largest value a 64 bit unsigned integer can hold @stable ICU 2.8 */
 227 #     define U_UINT64_MAX      ((uint64_t)(UINT64_C(18446744073709551615)))
 228 # endif
 229 #endif
 230
 231 /*==========================================================================*/
 232 /* Boolean data type                                                        */
 233 /*==========================================================================*/
 234
 235 /** The ICU boolean type @stable ICU 2.0 */
 236 typedef int8_t UBool;
 237
 238 #ifndef TRUE
 239 /** The TRUE value of a UBool @stable ICU 2.0 */
 240 #   define TRUE  1
 241 #endif
 242 #ifndef FALSE
 243 /** The FALSE value of a UBool @stable ICU 2.0 */
 244 #   define FALSE 0
 245 #endif
 246
 247
 248 /*==========================================================================*/
 249 /* Unicode data types                                                       */
 250 /*==========================================================================*/
 251
 252 /* wchar_t-related definitions -------------------------------------------- */
 253
 254 /*
 255  * \def U_WCHAR_IS_UTF16
 256  * Defined if wchar_t uses UTF-16.
 257  *
 258  * @stable ICU 2.0
 259  */
 260 /*
 261  * \def U_WCHAR_IS_UTF32
 262  * Defined if wchar_t uses UTF-32.
 263  *
 264  * @stable ICU 2.0
 265  */
 266 #if !defined(U_WCHAR_IS_UTF16) && !defined(U_WCHAR_IS_UTF32)
 267 #   ifdef __STDC_ISO_10646__
 268 #       if (U_SIZEOF_WCHAR_T==2)
 269 #           define U_WCHAR_IS_UTF16
 270 #       elif (U_SIZEOF_WCHAR_T==4)
 271 #           define  U_WCHAR_IS_UTF32
 272 #       endif
 273 #   elif defined __UCS2__
 274 #       if (U_PF_OS390 <= U_PLATFORM && U_PLATFORM <= U_PF_OS400) && (U_SIZEOF_WCHAR_T==2)
 275 #           define U_WCHAR_IS_UTF16
 276 #       endif
 277 #   elif defined(__UCS4__) || (U_PLATFORM == U_PF_OS400 && defined(__UTF32__))
 278 #       if (U_SIZEOF_WCHAR_T==4)
 279 #           define U_WCHAR_IS_UTF32
 280 #       endif
 281 #   elif U_PLATFORM_IS_DARWIN_BASED || (U_SIZEOF_WCHAR_T==4 && U_PLATFORM_IS_LINUX_BASED)
 282 #       define U_WCHAR_IS_UTF32
 283 #   elif U_PLATFORM_HAS_WIN32_API
 284 #       define U_WCHAR_IS_UTF16
 285 #   endif
 286 #endif
 287
 288 /* UChar and UChar32 definitions -------------------------------------------- */
 289
 290 /** Number of bytes in a UChar. @stable ICU 2.0 */
 291 #define U_SIZEOF_UCHAR 2
 292
 293 /**
 294  * \def U_CHAR16_IS_TYPEDEF
 295  * If 1, then char16_t is a typedef and not a real type (yet)
 296  * @internal
 297  */
 298 #if (U_PLATFORM == U_PF_AIX) && defined(__cplusplus) &&(U_CPLUSPLUS_VERSION < 11)
 299 // for AIX, uchar.h needs to be included
 300 # include <uchar.h>
 301 # define U_CHAR16_IS_TYPEDEF 1
 302 #else
 303 # define U_CHAR16_IS_TYPEDEF 0
 304 #endif
 305
 306
 307 /**
 308  * \var UChar
 309  *
 310  * The base type for UTF-16 code units and pointers.
 311  * Unsigned 16-bit integer.
 312  * Starting with ICU 59, C++ API uses char16_t directly, while C API continues to use UChar.
 313  *
 314  * UChar is configurable by defining the macro UCHAR_TYPE
 315  * on the preprocessor or compiler command line:
 316  * -DUCHAR_TYPE=uint16_t or -DUCHAR_TYPE=wchar_t (if U_SIZEOF_WCHAR_T==2) etc.
 317  * (The UCHAR_TYPE can also be #defined earlier in this file, for outside the ICU library code.)
 318  * This is for transitional use from application code that uses uint16_t or wchar_t for UTF-16.
 319  *
 320  * The default is UChar=char16_t.
 321  *
 322  * C++11 defines char16_t as bit-compatible with uint16_t, but as a distinct type.
 323  *
 324  * In C, char16_t is a simple typedef of uint_least16_t.
 325  * ICU requires uint_least16_t=uint16_t for data memory mapping.
 326  * On macOS, char16_t is not available because the uchar.h standard header is missing.
 327  *
 328  * @stable ICU 4.4
 329  */
 330
 331 #if 0
 332     // #if 1 is normal. UChar defaults to char16_t in C++.
 333     // For configuration testing of UChar=uint16_t temporarily change this to #if 0.
 334     // The intltest Makefile #defines UCHAR_TYPE=char16_t,
 335     // so we only #define it to uint16_t if it is undefined so far.
 336 #elif !defined(UCHAR_TYPE)
 337 #   define UCHAR_TYPE uint16_t
 338 #endif
 339
 340 #if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || \
 341         defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION)
 342     // Inside the ICU library code, never configurable.
 343     typedef char16_t UChar;
 344 #elif defined(UCHAR_TYPE)
 345     typedef UCHAR_TYPE UChar;
 346 #elif defined(__cplusplus)
 347     typedef char16_t UChar;
 348 #else
 349     typedef uint16_t UChar;
 350 #endif
 351
 352 /**
 353  * \var OldUChar
 354  * Default ICU 58 definition of UChar.
 355  * A base type for UTF-16 code units and pointers.
 356  * Unsigned 16-bit integer.
 357  *
 358  * Define OldUChar to be wchar_t if that is 16 bits wide.
 359  * If wchar_t is not 16 bits wide, then define UChar to be uint16_t.
 360  *
 361  * This makes the definition of OldUChar platform-dependent
 362  * but allows direct string type compatibility with platforms with
 363  * 16-bit wchar_t types.
 364  *
 365  * This is how UChar was defined in ICU 58, for transition convenience.
 366  * Exception: ICU 58 UChar was defined to UCHAR_TYPE if that macro was defined.
 367  * The current UChar responds to UCHAR_TYPE but OldUChar does not.
 368  *
 369  * @draft ICU 59
 370  */
 371 #if U_SIZEOF_WCHAR_T==2
 372     typedef wchar_t OldUChar;
 373 #elif defined(__CHAR16_TYPE__)
 374     typedef __CHAR16_TYPE__ OldUChar;
 375 #else
 376     typedef uint16_t OldUChar;
 377 #endif
 378
 379 /**
 380  * Define UChar32 as a type for single Unicode code points.
 381  * UChar32 is a signed 32-bit integer (same as int32_t).
 382  *
 383  * The Unicode code point range is 0..0x10ffff.
 384  * All other values (negative or >=0x110000) are illegal as Unicode code points.
 385  * They may be used as sentinel values to indicate "done", "error"
 386  * or similar non-code point conditions.
 387  *
 388  * Before ICU 2.4 (Jitterbug 2146), UChar32 was defined
 389  * to be wchar_t if that is 32 bits wide (wchar_t may be signed or unsigned)
 390  * or else to be uint32_t.
 391  * That is, the definition of UChar32 was platform-dependent.
 392  *
 393  * @see U_SENTINEL
 394  * @stable ICU 2.4
 395  */
 396 typedef int32_t UChar32;
 397
 398 /**
 399  * This value is intended for sentinel values for APIs that
 400  * (take or) return single code points (UChar32).
 401  * It is outside of the Unicode code point range 0..0x10ffff.
 402  *
 403  * For example, a "done" or "error" value in a new API
 404  * could be indicated with U_SENTINEL.
 405  *
 406  * ICU APIs designed before ICU 2.4 usually define service-specific "done"
 407  * values, mostly 0xffff.
 408  * Those may need to be distinguished from
 409  * actual U+ffff text contents by calling functions like
 410  * CharacterIterator::hasNext() or UnicodeString::length().
 411  *
 412  * @return -1
 413  * @see UChar32
 414  * @stable ICU 2.4
 415  */
 416 #define U_SENTINEL (-1)
 417
 418 #include "unicode/urename.h"
 419
 420 #endif