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 * \file
29 * \brief Basic types and constants for UTF
30 *
31 * <h2> Basic types and constants for UTF </h2>
32 *   This file defines basic types and constants for utf.h to be
33 *   platform-independent. umachine.h and utf.h are included into
34 *   utypes.h to provide all the general definitions for ICU.
35 *   All of these definitions used to be in utypes.h before
36 *   the UTF-handling macros made this unmaintainable.
37 *
38 */
39/*==========================================================================*/
40/* Include platform-dependent definitions                                   */
41/* which are contained in the platform-specific file platform.h             */
42/*==========================================================================*/
43
44#include "unicode/ptypes.h" /* platform.h is included in ptypes.h */
45
46/*
47 * ANSI C headers:
48 * stddef.h defines wchar_t
49 */
50#include <stdbool.h>
51#include <stddef.h>
52
53/*==========================================================================*/
54/* For C wrappers, we use the symbol U_CAPI.                                */
55/* This works properly if the includer is C or C++.                         */
56/* Functions are declared   U_CAPI return-type U_EXPORT2 function-name()... */
57/*==========================================================================*/
58
59/**
60 * \def U_CFUNC
61 * This is used in a declaration of a library private ICU C function.
62 * @stable ICU 2.4
63 */
64
65/**
66 * \def U_CDECL_BEGIN
67 * This is used to begin a declaration of a library private ICU C API.
68 * @stable ICU 2.4
69 */
70
71/**
72 * \def U_CDECL_END
73 * This is used to end a declaration of a library private ICU C API
74 * @stable ICU 2.4
75 */
76
77#ifdef __cplusplus
78#   define U_CFUNC extern "C"
79#   define U_CDECL_BEGIN extern "C" {
80#   define U_CDECL_END   }
81#else
82#   define U_CFUNC extern
83#   define U_CDECL_BEGIN
84#   define U_CDECL_END
85#endif
86
87#ifndef U_ATTRIBUTE_DEPRECATED
88/**
89 * \def U_ATTRIBUTE_DEPRECATED
90 *  This is used for GCC specific attributes
91 * @internal
92 */
93#if U_GCC_MAJOR_MINOR >= 302
94#    define U_ATTRIBUTE_DEPRECATED __attribute__ ((deprecated))
95/**
96 * \def U_ATTRIBUTE_DEPRECATED
97 * This is used for Visual C++ specific attributes
98 * @internal
99 */
100#elif defined(_MSC_VER) && (_MSC_VER >= 1400)
101#    define U_ATTRIBUTE_DEPRECATED __declspec(deprecated)
102#else
103#    define U_ATTRIBUTE_DEPRECATED
104#endif
105#endif
106
107/** This is used to declare a function as a public ICU C API @stable ICU 2.0*/
108#define U_CAPI U_CFUNC U_EXPORT
109/** Obsolete/same as U_CAPI; was used to declare a function as a stable public ICU C API*/
110#define U_STABLE U_CAPI
111/** Obsolete/same as U_CAPI; was used to declare a function as a draft public ICU C API  */
112#define U_DRAFT  U_CAPI
113/** This is used to declare a function as a deprecated public ICU C API  */
114#define U_DEPRECATED U_CAPI U_ATTRIBUTE_DEPRECATED
115/** Obsolete/same as U_CAPI; was used to declare a function as an obsolete public ICU C API  */
116#define U_OBSOLETE U_CAPI
117/** Obsolete/same as U_CAPI; was used to declare a function as an internal ICU C API  */
118#define U_INTERNAL U_CAPI
119
120/**
121 * \def U_OVERRIDE
122 * Defined to the C++11 "override" keyword if available.
123 * Denotes a class or member which is an override of the base class.
124 * May result in an error if it applied to something not an override.
125 * @internal
126 */
127#ifndef U_OVERRIDE
128#define U_OVERRIDE override
129#endif
130
131/**
132 * \def U_FINAL
133 * Defined to the C++11 "final" keyword if available.
134 * Denotes a class or member which may not be overridden in subclasses.
135 * May result in an error if subclasses attempt to override.
136 * @internal
137 */
138#if !defined(U_FINAL) || defined(U_IN_DOXYGEN)
139#define U_FINAL final
140#endif
141
142// Before ICU 65, function-like, multi-statement ICU macros were just defined as
143// series of statements wrapped in { } blocks and the caller could choose to
144// either treat them as if they were actual functions and end the invocation
145// with a trailing ; creating an empty statement after the block or else omit
146// this trailing ; using the knowledge that the macro would expand to { }.
147//
148// But doing so doesn't work well with macros that look like functions and
149// compiler warnings about empty statements (ICU-20601) and ICU 65 therefore
150// switches to the standard solution of wrapping such macros in do { } while.
151//
152// This will however break existing code that depends on being able to invoke
153// these macros without a trailing ; so to be able to remain compatible with
154// such code the wrapper is itself defined as macros so that it's possible to
155// build ICU 65 and later with the old macro behaviour, like this:
156//
157// export CPPFLAGS='-DUPRV_BLOCK_MACRO_BEGIN="" -DUPRV_BLOCK_MACRO_END=""'
158// runConfigureICU ...
159//
160
161/**
162 * \def UPRV_BLOCK_MACRO_BEGIN
163 * Defined as the "do" keyword by default.
164 * @internal
165 */
166#ifndef UPRV_BLOCK_MACRO_BEGIN
167#define UPRV_BLOCK_MACRO_BEGIN do
168#endif
169
170/**
171 * \def UPRV_BLOCK_MACRO_END
172 * Defined as "while (false)" by default.
173 * @internal
174 */
175#ifndef UPRV_BLOCK_MACRO_END
176#define UPRV_BLOCK_MACRO_END while (false)
177#endif
178
179/*==========================================================================*/
180/* limits for int32_t etc., like in POSIX inttypes.h                        */
181/*==========================================================================*/
182
183#ifndef INT8_MIN
184/** The smallest value an 8 bit signed integer can hold @stable ICU 2.0 */
185#   define INT8_MIN        ((int8_t)(-128))
186#endif
187#ifndef INT16_MIN
188/** The smallest value a 16 bit signed integer can hold @stable ICU 2.0 */
189#   define INT16_MIN       ((int16_t)(-32767-1))
190#endif
191#ifndef INT32_MIN
192/** The smallest value a 32 bit signed integer can hold @stable ICU 2.0 */
193#   define INT32_MIN       ((int32_t)(-2147483647-1))
194#endif
195
196#ifndef INT8_MAX
197/** The largest value an 8 bit signed integer can hold @stable ICU 2.0 */
198#   define INT8_MAX        ((int8_t)(127))
199#endif
200#ifndef INT16_MAX
201/** The largest value a 16 bit signed integer can hold @stable ICU 2.0 */
202#   define INT16_MAX       ((int16_t)(32767))
203#endif
204#ifndef INT32_MAX
205/** The largest value a 32 bit signed integer can hold @stable ICU 2.0 */
206#   define INT32_MAX       ((int32_t)(2147483647))
207#endif
208
209#ifndef UINT8_MAX
210/** The largest value an 8 bit unsigned integer can hold @stable ICU 2.0 */
211#   define UINT8_MAX       ((uint8_t)(255U))
212#endif
213#ifndef UINT16_MAX
214/** The largest value a 16 bit unsigned integer can hold @stable ICU 2.0 */
215#   define UINT16_MAX      ((uint16_t)(65535U))
216#endif
217#ifndef UINT32_MAX
218/** The largest value a 32 bit unsigned integer can hold @stable ICU 2.0 */
219#   define UINT32_MAX      ((uint32_t)(4294967295U))
220#endif
221
222#if defined(U_INT64_T_UNAVAILABLE)
223# error int64_t is required for decimal format and rule-based number format.
224#else
225# ifndef INT64_C
226/**
227 * Provides a platform independent way to specify a signed 64-bit integer constant.
228 * note: may be wrong for some 64 bit platforms - ensure your compiler provides INT64_C
229 * @stable ICU 2.8
230 */
231#   define INT64_C(c) c ## LL
232# endif
233# ifndef UINT64_C
234/**
235 * Provides a platform independent way to specify an unsigned 64-bit integer constant.
236 * note: may be wrong for some 64 bit platforms - ensure your compiler provides UINT64_C
237 * @stable ICU 2.8
238 */
239#   define UINT64_C(c) c ## ULL
240# endif
241# ifndef U_INT64_MIN
242/** The smallest value a 64 bit signed integer can hold @stable ICU 2.8 */
243#     define U_INT64_MIN       ((int64_t)(INT64_C(-9223372036854775807)-1))
244# endif
245# ifndef U_INT64_MAX
246/** The largest value a 64 bit signed integer can hold @stable ICU 2.8 */
247#     define U_INT64_MAX       ((int64_t)(INT64_C(9223372036854775807)))
248# endif
249# ifndef U_UINT64_MAX
250/** The largest value a 64 bit unsigned integer can hold @stable ICU 2.8 */
251#     define U_UINT64_MAX      ((uint64_t)(UINT64_C(18446744073709551615)))
252# endif
253#endif
254
255/*==========================================================================*/
256/* Boolean data type                                                        */
257/*==========================================================================*/
258
259/**
260 * The ICU boolean type, a signed-byte integer.
261 * ICU-specific for historical reasons: The C and C++ standards used to not define type bool.
262 * Also provides a fixed type definition, as opposed to
263 * type bool whose details (e.g., sizeof) may vary by compiler and between C and C++.
264 *
265 * @stable ICU 2.0
266 */
267typedef int8_t UBool;
268
269/**
270 * \def U_DEFINE_FALSE_AND_TRUE
271 * Normally turns off defining macros FALSE=0 & TRUE=1 in public ICU headers.
272 * These obsolete macros sometimes break compilation of other code that
273 * defines enum constants or similar with these names.
274 * C++ has long defined bool/false/true.
275 * C99 also added definitions for these, although as macros; see stdbool.h.
276 *
277 * You may transitionally define U_DEFINE_FALSE_AND_TRUE=1 if you need time to migrate code.
278 *
279 * @internal ICU 68
280 */
281#ifdef U_DEFINE_FALSE_AND_TRUE
282    // Use the predefined value.
283#elif defined(U_COMBINED_IMPLEMENTATION) || \
284        defined(U_COMMON_IMPLEMENTATION) || defined(U_I18N_IMPLEMENTATION) || \
285        defined(U_IO_IMPLEMENTATION) || defined(U_LAYOUTEX_IMPLEMENTATION) || \
286        defined(U_TOOLUTIL_IMPLEMENTATION)
287    // Inside ICU: Keep FALSE & TRUE available.
288#   define U_DEFINE_FALSE_AND_TRUE 1
289#else
290    // Outside ICU: Avoid collision with non-macro definitions of FALSE & TRUE.
291#   define U_DEFINE_FALSE_AND_TRUE 1
292#endif
293
294#if U_DEFINE_FALSE_AND_TRUE || defined(U_IN_DOXYGEN)
295#ifndef TRUE
296/**
297 * The TRUE value of a UBool.
298 *
299 * @deprecated ICU 68 Use standard "true" instead.
300 */
301#   define TRUE  1
302#endif
303#ifndef FALSE
304/**
305 * The FALSE value of a UBool.
306 *
307 * @deprecated ICU 68 Use standard "false" instead.
308 */
309#   define FALSE 0
310#endif
311#endif  // U_DEFINE_FALSE_AND_TRUE
312
313/*==========================================================================*/
314/* Unicode data types                                                       */
315/*==========================================================================*/
316
317/* wchar_t-related definitions -------------------------------------------- */
318
319/*
320 * \def U_WCHAR_IS_UTF16
321 * Defined if wchar_t uses UTF-16.
322 *
323 * @stable ICU 2.0
324 */
325/*
326 * \def U_WCHAR_IS_UTF32
327 * Defined if wchar_t uses UTF-32.
328 *
329 * @stable ICU 2.0
330 */
331#if !defined(U_WCHAR_IS_UTF16) && !defined(U_WCHAR_IS_UTF32)
332#   ifdef __STDC_ISO_10646__
333#       if (U_SIZEOF_WCHAR_T==2)
334#           define U_WCHAR_IS_UTF16
335#       elif (U_SIZEOF_WCHAR_T==4)
336#           define  U_WCHAR_IS_UTF32
337#       endif
338#   elif defined __UCS2__
339#       if (U_PF_OS390 <= U_PLATFORM && U_PLATFORM <= U_PF_OS400) && (U_SIZEOF_WCHAR_T==2)
340#           define U_WCHAR_IS_UTF16
341#       endif
342#   elif defined(__UCS4__) || (U_PLATFORM == U_PF_OS400 && defined(__UTF32__))
343#       if (U_SIZEOF_WCHAR_T==4)
344#           define U_WCHAR_IS_UTF32
345#       endif
346#   elif U_PLATFORM_IS_DARWIN_BASED || (U_SIZEOF_WCHAR_T==4 && U_PLATFORM_IS_LINUX_BASED)
347#       define U_WCHAR_IS_UTF32
348#   elif U_PLATFORM_HAS_WIN32_API
349#       define U_WCHAR_IS_UTF16
350#   endif
351#endif
352
353/* UChar and UChar32 definitions -------------------------------------------- */
354
355/** Number of bytes in a UChar. @stable ICU 2.0 */
356#define U_SIZEOF_UCHAR 2
357
358/**
359 * \def U_CHAR16_IS_TYPEDEF
360 * If 1, then char16_t is a typedef and not a real type (yet)
361 * @internal
362 */
363#if (U_PLATFORM == U_PF_AIX) && defined(__cplusplus) &&(U_CPLUSPLUS_VERSION < 11)
364// for AIX, uchar.h needs to be included
365# include <uchar.h>
366# define U_CHAR16_IS_TYPEDEF 1
367#elif defined(_MSC_VER) && (_MSC_VER < 1900)
368// Versions of Visual Studio/MSVC below 2015 do not support char16_t as a real type,
369// and instead use a typedef.  https://msdn.microsoft.com/library/bb531344.aspx
370# define U_CHAR16_IS_TYPEDEF 1
371#else
372# define U_CHAR16_IS_TYPEDEF 0
373#endif
374/**
375 * \var UChar
376 *
377 * The base type for UTF-16 code units and pointers.
378 * Unsigned 16-bit integer.
379 * Starting with ICU 59, C++ API uses char16_t directly, while C API continues to use UChar.
380 *
381 * UChar is configurable by defining the macro UCHAR_TYPE
382 * on the preprocessor or compiler command line:
383 * -DUCHAR_TYPE=uint16_t or -DUCHAR_TYPE=wchar_t (if U_SIZEOF_WCHAR_T==2) etc.
384 * (The UCHAR_TYPE can also be \#defined earlier in this file, for outside the ICU library code.)
385 * This is for transitional use from application code that uses uint16_t or wchar_t for UTF-16.
386 *
387 * The default is UChar=char16_t.
388 *
389 * C++11 defines char16_t as bit-compatible with uint16_t, but as a distinct type.
390 *
391 * In C, char16_t is a simple typedef of uint_least16_t.
392 * ICU requires uint_least16_t=uint16_t for data memory mapping.
393 * On macOS, char16_t is not available because the uchar.h standard header is missing.
394 *
395 * @stable ICU 4.4
396 */
397
398#if 1
399    // #if 1 is normal. UChar defaults to char16_t in C++.
400    // For configuration testing of UChar=uint16_t temporarily change this to #if 0.
401    // The intltest Makefile #defines UCHAR_TYPE=char16_t,
402    // so we only #define it to uint16_t if it is undefined so far.
403#elif !defined(UCHAR_TYPE)
404#   define UCHAR_TYPE uint16_t
405#endif
406
407#if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || \
408        defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION)
409    // Inside the ICU library code, never configurable.
410    typedef char16_t UChar;
411#elif defined(UCHAR_TYPE)
412    typedef UCHAR_TYPE UChar;
413#elif (U_CPLUSPLUS_VERSION >= 11)
414    typedef char16_t UChar;
415#else
416    typedef uint16_t UChar;
417#endif
418
419/**
420 * \var OldUChar
421 * Default ICU 58 definition of UChar.
422 * A base type for UTF-16 code units and pointers.
423 * Unsigned 16-bit integer.
424 *
425 * Define OldUChar to be wchar_t if that is 16 bits wide.
426 * If wchar_t is not 16 bits wide, then define UChar to be uint16_t.
427 *
428 * This makes the definition of OldUChar platform-dependent
429 * but allows direct string type compatibility with platforms with
430 * 16-bit wchar_t types.
431 *
432 * This is how UChar was defined in ICU 58, for transition convenience.
433 * Exception: ICU 58 UChar was defined to UCHAR_TYPE if that macro was defined.
434 * The current UChar responds to UCHAR_TYPE but OldUChar does not.
435 *
436 * @stable ICU 59
437 */
438#if U_SIZEOF_WCHAR_T==2
439    typedef wchar_t OldUChar;
440#elif defined(__CHAR16_TYPE__)
441    typedef __CHAR16_TYPE__ OldUChar;
442#else
443    typedef uint16_t OldUChar;
444#endif
445
446/**
447 * Define UChar32 as a type for single Unicode code points.
448 * UChar32 is a signed 32-bit integer (same as int32_t).
449 *
450 * The Unicode code point range is 0..0x10ffff.
451 * All other values (negative or >=0x110000) are illegal as Unicode code points.
452 * They may be used as sentinel values to indicate "done", "error"
453 * or similar non-code point conditions.
454 *
455 * Before ICU 2.4 (Jitterbug 2146), UChar32 was defined
456 * to be wchar_t if that is 32 bits wide (wchar_t may be signed or unsigned)
457 * or else to be uint32_t.
458 * That is, the definition of UChar32 was platform-dependent.
459 *
460 * @see U_SENTINEL
461 * @stable ICU 2.4
462 */
463typedef int32_t UChar32;
464
465/**
466 * This value is intended for sentinel values for APIs that
467 * (take or) return single code points (UChar32).
468 * It is outside of the Unicode code point range 0..0x10ffff.
469 *
470 * For example, a "done" or "error" value in a new API
471 * could be indicated with U_SENTINEL.
472 *
473 * ICU APIs designed before ICU 2.4 usually define service-specific "done"
474 * values, mostly 0xffff.
475 * Those may need to be distinguished from
476 * actual U+ffff text contents by calling functions like
477 * CharacterIterator::hasNext() or UnicodeString::length().
478 *
479 * @return -1
480 * @see UChar32
481 * @stable ICU 2.4
482 */
483#define U_SENTINEL (-1)
484
485#endif
486