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// 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// export CPPFLAGS='-DUPRV_BLOCK_MACRO_BEGIN="" -DUPRV_BLOCK_MACRO_END=""'
159// runConfigureICU ...
160//
161
162/**
163 * \def UPRV_BLOCK_MACRO_BEGIN
164 * Defined as the "do" keyword by default.
165 * @internal
166 */
167#ifndef UPRV_BLOCK_MACRO_BEGIN
168#define UPRV_BLOCK_MACRO_BEGIN do
169#endif
170
171/**
172 * \def UPRV_BLOCK_MACRO_END
173 * Defined as "while (FALSE)" by default.
174 * @internal
175 */
176#ifndef UPRV_BLOCK_MACRO_END
177#define UPRV_BLOCK_MACRO_END while (FALSE)
178#endif
179
180/*==========================================================================*/
181/* limits for int32_t etc., like in POSIX inttypes.h */
182/*==========================================================================*/
183
184#ifndef INT8_MIN
185/** The smallest value an 8 bit signed integer can hold @stable ICU 2.0 */
186# define INT8_MIN ((int8_t)(-128))
187#endif
188#ifndef INT16_MIN
189/** The smallest value a 16 bit signed integer can hold @stable ICU 2.0 */
190# define INT16_MIN ((int16_t)(-32767-1))
191#endif
192#ifndef INT32_MIN
193/** The smallest value a 32 bit signed integer can hold @stable ICU 2.0 */
194# define INT32_MIN ((int32_t)(-2147483647-1))
195#endif
196
197#ifndef INT8_MAX
198/** The largest value an 8 bit signed integer can hold @stable ICU 2.0 */
199# define INT8_MAX ((int8_t)(127))
200#endif
201#ifndef INT16_MAX
202/** The largest value a 16 bit signed integer can hold @stable ICU 2.0 */
203# define INT16_MAX ((int16_t)(32767))
204#endif
205#ifndef INT32_MAX
206/** The largest value a 32 bit signed integer can hold @stable ICU 2.0 */
207# define INT32_MAX ((int32_t)(2147483647))
208#endif
209
210#ifndef UINT8_MAX
211/** The largest value an 8 bit unsigned integer can hold @stable ICU 2.0 */
212# define UINT8_MAX ((uint8_t)(255U))
213#endif
214#ifndef UINT16_MAX
215/** The largest value a 16 bit unsigned integer can hold @stable ICU 2.0 */
216# define UINT16_MAX ((uint16_t)(65535U))
217#endif
218#ifndef UINT32_MAX
219/** The largest value a 32 bit unsigned integer can hold @stable ICU 2.0 */
220# define UINT32_MAX ((uint32_t)(4294967295U))
221#endif
222
223#if defined(U_INT64_T_UNAVAILABLE)
224# error int64_t is required for decimal format and rule-based number format.
225#else
226# ifndef INT64_C
227/**
228 * Provides a platform independent way to specify a signed 64-bit integer constant.
229 * note: may be wrong for some 64 bit platforms - ensure your compiler provides INT64_C
230 * @stable ICU 2.8
231 */
232# define INT64_C(c) c ## LL
233# endif
234# ifndef UINT64_C
235/**
236 * Provides a platform independent way to specify an unsigned 64-bit integer constant.
237 * note: may be wrong for some 64 bit platforms - ensure your compiler provides UINT64_C
238 * @stable ICU 2.8
239 */
240# define UINT64_C(c) c ## ULL
241# endif
242# ifndef U_INT64_MIN
243/** The smallest value a 64 bit signed integer can hold @stable ICU 2.8 */
244# define U_INT64_MIN ((int64_t)(INT64_C(-9223372036854775807)-1))
245# endif
246# ifndef U_INT64_MAX
247/** The largest value a 64 bit signed integer can hold @stable ICU 2.8 */
248# define U_INT64_MAX ((int64_t)(INT64_C(9223372036854775807)))
249# endif
250# ifndef U_UINT64_MAX
251/** The largest value a 64 bit unsigned integer can hold @stable ICU 2.8 */
252# define U_UINT64_MAX ((uint64_t)(UINT64_C(18446744073709551615)))
253# endif
254#endif
255
256/*==========================================================================*/
257/* Boolean data type */
258/*==========================================================================*/
259
260/** The ICU boolean type @stable ICU 2.0 */
261typedef int8_t UBool;
262
263#ifndef TRUE
264/** The TRUE value of a UBool @stable ICU 2.0 */
265# define TRUE 1
266#endif
267#ifndef FALSE
268/** The FALSE value of a UBool @stable ICU 2.0 */
269# define FALSE 0
270#endif
271
272
273/*==========================================================================*/
274/* Unicode data types */
275/*==========================================================================*/
276
277/* wchar_t-related definitions -------------------------------------------- */
278
279/*
280 * \def U_WCHAR_IS_UTF16
281 * Defined if wchar_t uses UTF-16.
282 *
283 * @stable ICU 2.0
284 */
285/*
286 * \def U_WCHAR_IS_UTF32
287 * Defined if wchar_t uses UTF-32.
288 *
289 * @stable ICU 2.0
290 */
291#if !defined(U_WCHAR_IS_UTF16) && !defined(U_WCHAR_IS_UTF32)
292# ifdef __STDC_ISO_10646__
293# if (U_SIZEOF_WCHAR_T==2)
294# define U_WCHAR_IS_UTF16
295# elif (U_SIZEOF_WCHAR_T==4)
296# define U_WCHAR_IS_UTF32
297# endif
298# elif defined __UCS2__
299# if (U_PF_OS390 <= U_PLATFORM && U_PLATFORM <= U_PF_OS400) && (U_SIZEOF_WCHAR_T==2)
300# define U_WCHAR_IS_UTF16
301# endif
302# elif defined(__UCS4__) || (U_PLATFORM == U_PF_OS400 && defined(__UTF32__))
303# if (U_SIZEOF_WCHAR_T==4)
304# define U_WCHAR_IS_UTF32
305# endif
306# elif U_PLATFORM_IS_DARWIN_BASED || (U_SIZEOF_WCHAR_T==4 && U_PLATFORM_IS_LINUX_BASED)
307# define U_WCHAR_IS_UTF32
308# elif U_PLATFORM_HAS_WIN32_API
309# define U_WCHAR_IS_UTF16
310# endif
311#endif
312
313/* UChar and UChar32 definitions -------------------------------------------- */
314
315/** Number of bytes in a UChar. @stable ICU 2.0 */
316#define U_SIZEOF_UCHAR 2
317
318/**
319 * \def U_CHAR16_IS_TYPEDEF
320 * If 1, then char16_t is a typedef and not a real type (yet)
321 * @internal
322 */
323#if (U_PLATFORM == U_PF_AIX) && defined(__cplusplus) &&(U_CPLUSPLUS_VERSION < 11)
324// for AIX, uchar.h needs to be included
325# include <uchar.h>
326# define U_CHAR16_IS_TYPEDEF 1
327#elif defined(_MSC_VER) && (_MSC_VER < 1900)
328// Versions of Visual Studio/MSVC below 2015 do not support char16_t as a real type,
329// and instead use a typedef. https://msdn.microsoft.com/library/bb531344.aspx
330# define U_CHAR16_IS_TYPEDEF 1
331#else
332# define U_CHAR16_IS_TYPEDEF 0
333#endif
334
335
336/**
337 * \var UChar
338 *
339 * The base type for UTF-16 code units and pointers.
340 * Unsigned 16-bit integer.
341 * Starting with ICU 59, C++ API uses char16_t directly, while C API continues to use UChar.
342 *
343 * UChar is configurable by defining the macro UCHAR_TYPE
344 * on the preprocessor or compiler command line:
345 * -DUCHAR_TYPE=uint16_t or -DUCHAR_TYPE=wchar_t (if U_SIZEOF_WCHAR_T==2) etc.
346 * (The UCHAR_TYPE can also be \#defined earlier in this file, for outside the ICU library code.)
347 * This is for transitional use from application code that uses uint16_t or wchar_t for UTF-16.
348 *
349 * The default is UChar=char16_t.
350 *
351 * C++11 defines char16_t as bit-compatible with uint16_t, but as a distinct type.
352 *
353 * In C, char16_t is a simple typedef of uint_least16_t.
354 * ICU requires uint_least16_t=uint16_t for data memory mapping.
355 * On macOS, char16_t is not available because the uchar.h standard header is missing.
356 *
357 * @stable ICU 4.4
358 */
359
360#if 1
361 // #if 1 is normal. UChar defaults to char16_t in C++.
362 // For configuration testing of UChar=uint16_t temporarily change this to #if 0.
363 // The intltest Makefile #defines UCHAR_TYPE=char16_t,
364 // so we only #define it to uint16_t if it is undefined so far.
365#elif !defined(UCHAR_TYPE)
366# define UCHAR_TYPE uint16_t
367#endif
368
369#if defined(U_COMBINED_IMPLEMENTATION) || defined(U_COMMON_IMPLEMENTATION) || \
370 defined(U_I18N_IMPLEMENTATION) || defined(U_IO_IMPLEMENTATION)
371 // Inside the ICU library code, never configurable.
372 typedef char16_t UChar;
373#elif defined(UCHAR_TYPE)
374 typedef UCHAR_TYPE UChar;
375#elif (U_CPLUSPLUS_VERSION >= 11)
376 typedef char16_t UChar;
377#else
378 typedef uint16_t UChar;
379#endif
380
381/**
382 * \var OldUChar
383 * Default ICU 58 definition of UChar.
384 * A base type for UTF-16 code units and pointers.
385 * Unsigned 16-bit integer.
386 *
387 * Define OldUChar to be wchar_t if that is 16 bits wide.
388 * If wchar_t is not 16 bits wide, then define UChar to be uint16_t.
389 *
390 * This makes the definition of OldUChar platform-dependent
391 * but allows direct string type compatibility with platforms with
392 * 16-bit wchar_t types.
393 *
394 * This is how UChar was defined in ICU 58, for transition convenience.
395 * Exception: ICU 58 UChar was defined to UCHAR_TYPE if that macro was defined.
396 * The current UChar responds to UCHAR_TYPE but OldUChar does not.
397 *
398 * @stable ICU 59
399 */
400#if U_SIZEOF_WCHAR_T==2
401 typedef wchar_t OldUChar;
402#elif defined(__CHAR16_TYPE__)
403 typedef __CHAR16_TYPE__ OldUChar;
404#else
405 typedef uint16_t OldUChar;
406#endif
407
408/**
409 * Define UChar32 as a type for single Unicode code points.
410 * UChar32 is a signed 32-bit integer (same as int32_t).
411 *
412 * The Unicode code point range is 0..0x10ffff.
413 * All other values (negative or >=0x110000) are illegal as Unicode code points.
414 * They may be used as sentinel values to indicate "done", "error"
415 * or similar non-code point conditions.
416 *
417 * Before ICU 2.4 (Jitterbug 2146), UChar32 was defined
418 * to be wchar_t if that is 32 bits wide (wchar_t may be signed or unsigned)
419 * or else to be uint32_t.
420 * That is, the definition of UChar32 was platform-dependent.
421 *
422 * @see U_SENTINEL
423 * @stable ICU 2.4
424 */
425typedef int32_t UChar32;
426
427/**
428 * This value is intended for sentinel values for APIs that
429 * (take or) return single code points (UChar32).
430 * It is outside of the Unicode code point range 0..0x10ffff.
431 *
432 * For example, a "done" or "error" value in a new API
433 * could be indicated with U_SENTINEL.
434 *
435 * ICU APIs designed before ICU 2.4 usually define service-specific "done"
436 * values, mostly 0xffff.
437 * Those may need to be distinguished from
438 * actual U+ffff text contents by calling functions like
439 * CharacterIterator::hasNext() or UnicodeString::length().
440 *
441 * @return -1
442 * @see UChar32
443 * @stable ICU 2.4
444 */
445#define U_SENTINEL (-1)
446
447#include "unicode/urename.h"
448
449#endif
450