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 "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// Before ICU 65, function-like, multi-statement ICU macros were just defined as
144// series of statements wrapped in { } blocks and the caller could choose to
145// either treat them as if they were actual functions and end the invocation
146// with a trailing ; creating an empty statement after the block or else omit
147// this trailing ; using the knowledge that the macro would expand to { }.
148//
149// But doing so doesn't work well with macros that look like functions and
150// compiler warnings about empty statements (ICU-20601) and ICU 65 therefore
151// switches to the standard solution of wrapping such macros in do { } while.
152//
153// This will however break existing code that depends on being able to invoke
154// these macros without a trailing ; so to be able to remain compatible with
155// such code the wrapper is itself defined as macros so that it's possible to
156// build ICU 65 and later with the old macro behaviour, like this:
157//
158// CPPFLAGS='-DUPRV_BLOCK_MACRO_BEGIN="" -DUPRV_BLOCK_MACRO_END=""'
159// runConfigureICU ...
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/** The ICU boolean type @stable ICU 2.0 */
260typedef int8_t UBool;
261
262#ifndef TRUE
263/** The TRUE value of a UBool @stable ICU 2.0 */
264# define TRUE 1
265#endif
266#ifndef FALSE
267/** The FALSE value of a UBool @stable ICU 2.0 */
268# define FALSE 0
269#endif
270
271
272/*==========================================================================*/
273/* Unicode data types */
274/*==========================================================================*/
275
276/* wchar_t-related definitions -------------------------------------------- */
277
278/*
279 * \def U_WCHAR_IS_UTF16
280 * Defined if wchar_t uses UTF-16.
281 *
282 * @stable ICU 2.0
283 */
284/*
285 * \def U_WCHAR_IS_UTF32
286 * Defined if wchar_t uses UTF-32.
287 *
288 * @stable ICU 2.0
289 */
290#if !defined(U_WCHAR_IS_UTF16) && !defined(U_WCHAR_IS_UTF32)
291# ifdef __STDC_ISO_10646__
292# if (U_SIZEOF_WCHAR_T==2)
293# define U_WCHAR_IS_UTF16
294# elif (U_SIZEOF_WCHAR_T==4)
295# define U_WCHAR_IS_UTF32
296# endif
297# elif defined __UCS2__
298# if (U_PF_OS390 <= U_PLATFORM && U_PLATFORM <= U_PF_OS400) && (U_SIZEOF_WCHAR_T==2)
299# define U_WCHAR_IS_UTF16
300# endif
301# elif defined(__UCS4__) || (U_PLATFORM == U_PF_OS400 && defined(__UTF32__))
302# if (U_SIZEOF_WCHAR_T==4)
303# define U_WCHAR_IS_UTF32
304# endif
305# elif U_PLATFORM_IS_DARWIN_BASED || (U_SIZEOF_WCHAR_T==4 && U_PLATFORM_IS_LINUX_BASED)
306# define U_WCHAR_IS_UTF32
307# elif U_PLATFORM_HAS_WIN32_API
308# define U_WCHAR_IS_UTF16
309# endif
310#endif
311
312/* UChar and UChar32 definitions -------------------------------------------- */
313
314/** Number of bytes in a UChar. @stable ICU 2.0 */
315#define U_SIZEOF_UCHAR 2
316
317/**
318 * \def U_CHAR16_IS_TYPEDEF
319 * If 1, then char16_t is a typedef and not a real type (yet)
320 * @internal
321 */
322#if (U_PLATFORM == U_PF_AIX) && defined(__cplusplus) &&(U_CPLUSPLUS_VERSION < 11)
323// for AIX, uchar.h needs to be included
324# include <uchar.h>
325# define U_CHAR16_IS_TYPEDEF 1
326#elif defined(_MSC_VER) && (_MSC_VER < 1900)
327// Versions of Visual Studio/MSVC below 2015 do not support char16_t as a real type,
328// and instead use a typedef. https://msdn.microsoft.com/library/bb531344.aspx
329# define U_CHAR16_IS_TYPEDEF 1
330#else
331# define U_CHAR16_IS_TYPEDEF 0
332#endif
333
334
335/**
336 * \var UChar
337 *
338 * The base type for UTF-16 code units and pointers.
339 * Unsigned 16-bit integer.
340 * Starting with ICU 59, C++ API uses char16_t directly, while C API continues to use UChar.
341 *
342 * UChar is configurable by defining the macro UCHAR_TYPE
343 * on the preprocessor or compiler command line:
344 * -DUCHAR_TYPE=uint16_t or -DUCHAR_TYPE=wchar_t (if U_SIZEOF_WCHAR_T==2) etc.
345 * (The UCHAR_TYPE can also be \#defined earlier in this file, for outside the ICU library code.)
346 * This is for transitional use from application code that uses uint16_t or wchar_t for UTF-16.
347 *
348 * The default is UChar=char16_t.
349 *
350 * C++11 defines char16_t as bit-compatible with uint16_t, but as a distinct type.
351 *
352 * In C, char16_t is a simple typedef of uint_least16_t.
353 * ICU requires uint_least16_t=uint16_t for data memory mapping.
354 * On macOS, char16_t is not available because the uchar.h standard header is missing.
355 *
356 * @stable ICU 4.4
357 */
358
359#if 1
360 // #if 1 is normal. UChar defaults to char16_t in C++.
361 // For configuration testing of UChar=uint16_t temporarily change this to #if 0.
362 // The intltest Makefile #defines UCHAR_TYPE=char16_t,
363 // so we only #define it to uint16_t if it is undefined so far.
364#elif !defined(UCHAR_TYPE)
365# define UCHAR_TYPE uint16_t
366#endif
367
368#if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || \
369 defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION)
370 // Inside the ICU library code, never configurable.
371 typedef char16_t UChar;
372#elif defined(UCHAR_TYPE)
373 typedef UCHAR_TYPE UChar;
374#elif defined(__cplusplus)
375 typedef char16_t UChar;
376#else
377 typedef uint16_t UChar;
378#endif
379
380/**
381 * \var OldUChar
382 * Default ICU 58 definition of UChar.
383 * A base type for UTF-16 code units and pointers.
384 * Unsigned 16-bit integer.
385 *
386 * Define OldUChar to be wchar_t if that is 16 bits wide.
387 * If wchar_t is not 16 bits wide, then define UChar to be uint16_t.
388 *
389 * This makes the definition of OldUChar platform-dependent
390 * but allows direct string type compatibility with platforms with
391 * 16-bit wchar_t types.
392 *
393 * This is how UChar was defined in ICU 58, for transition convenience.
394 * Exception: ICU 58 UChar was defined to UCHAR_TYPE if that macro was defined.
395 * The current UChar responds to UCHAR_TYPE but OldUChar does not.
396 *
397 * @stable ICU 59
398 */
399#if U_SIZEOF_WCHAR_T==2
400 typedef wchar_t OldUChar;
401#elif defined(__CHAR16_TYPE__)
402 typedef __CHAR16_TYPE__ OldUChar;
403#else
404 typedef uint16_t OldUChar;
405#endif
406
407/**
408 * Define UChar32 as a type for single Unicode code points.
409 * UChar32 is a signed 32-bit integer (same as int32_t).
410 *
411 * The Unicode code point range is 0..0x10ffff.
412 * All other values (negative or >=0x110000) are illegal as Unicode code points.
413 * They may be used as sentinel values to indicate "done", "error"
414 * or similar non-code point conditions.
415 *
416 * Before ICU 2.4 (Jitterbug 2146), UChar32 was defined
417 * to be wchar_t if that is 32 bits wide (wchar_t may be signed or unsigned)
418 * or else to be uint32_t.
419 * That is, the definition of UChar32 was platform-dependent.
420 *
421 * @see U_SENTINEL
422 * @stable ICU 2.4
423 */
424typedef int32_t UChar32;
425
426/**
427 * This value is intended for sentinel values for APIs that
428 * (take or) return single code points (UChar32).
429 * It is outside of the Unicode code point range 0..0x10ffff.
430 *
431 * For example, a "done" or "error" value in a new API
432 * could be indicated with U_SENTINEL.
433 *
434 * ICU APIs designed before ICU 2.4 usually define service-specific "done"
435 * values, mostly 0xffff.
436 * Those may need to be distinguished from
437 * actual U+ffff text contents by calling functions like
438 * CharacterIterator::hasNext() or UnicodeString::length().
439 *
440 * @return -1
441 * @see UChar32
442 * @stable ICU 2.4
443 */
444#define U_SENTINEL (-1)
445
446#include "urename.h"
447
448#endif