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) 1997-2016, International Business Machines
7 *   Corporation and others.  All Rights Reserved.
8 *
9 ******************************************************************************
10 *
11 *  FILE NAME : putilimp.h
12 *
13 *   Date        Name        Description
14 *   10/17/04    grhoten     Move internal functions from putil.h to this file.
15 ******************************************************************************
16 */
17 
18 #ifndef PUTILIMP_H
19 #define PUTILIMP_H
20 
21 #include "unicode/utypes.h"
22 #include "unicode/putil.h"
23 
24 /**
25  * \def U_SIGNED_RIGHT_SHIFT_IS_ARITHMETIC
26  * Nearly all CPUs and compilers implement a right-shift of a signed integer
27  * as an Arithmetic Shift Right which copies the sign bit (the Most Significant Bit (MSB))
28  * into the vacated bits (sign extension).
29  * For example, (int32_t)0xfff5fff3>>4 becomes 0xffff5fff and -1>>1=-1.
30  *
31  * This can be useful for storing a signed value in the upper bits
32  * and another bit field in the lower bits.
33  * The signed value can be retrieved by simple right-shifting.
34  *
35  * This is consistent with the Java language.
36  *
37  * However, the C standard allows compilers to implement a right-shift of a signed integer
38  * as a Logical Shift Right which copies a 0 into the vacated bits.
39  * For example, (int32_t)0xfff5fff3>>4 becomes 0x0fff5fff and -1>>1=0x7fffffff.
40  *
41  * Code that depends on the natural behavior should be guarded with this macro,
42  * with an alternate path for unusual platforms.
43  * @internal
44  */
45 #ifdef U_SIGNED_RIGHT_SHIFT_IS_ARITHMETIC
46     /* Use the predefined value. */
47 #else
48     /*
49      * Nearly all CPUs & compilers implement a right-shift of a signed integer
50      * as an Arithmetic Shift Right (with sign extension).
51      */
52 #   define U_SIGNED_RIGHT_SHIFT_IS_ARITHMETIC 1
53 #endif
54 
55 /** Define this to 1 if your platform supports IEEE 754 floating point,
56    to 0 if it does not. */
57 #ifndef IEEE_754
58 #   define IEEE_754 1
59 #endif
60 
61 /**
62  * uintptr_t is an optional part of the standard definitions in stdint.h.
63  * The opengroup.org documentation for stdint.h says
64  * "On XSI-conformant systems, the intptr_t and uintptr_t types are required;
65  * otherwise, they are optional."
66  * We assume that when uintptr_t is defined, UINTPTR_MAX is defined as well.
67  *
68  * Do not use ptrdiff_t since it is signed. size_t is unsigned.
69  */
70 /* TODO: This check fails on some z environments. Filed a ticket #9357 for this. */
71 #if !defined(__intptr_t_defined) && !defined(UINTPTR_MAX) && (U_PLATFORM != U_PF_OS390)
72 typedef size_t uintptr_t;
73 #endif
74 
75 /*===========================================================================*/
76 /** @{ Information about POSIX support                                       */
77 /*===========================================================================*/
78 
79 #ifdef U_HAVE_NL_LANGINFO_CODESET
80     /* Use the predefined value. */
81 #elif U_PLATFORM_USES_ONLY_WIN32_API || U_PLATFORM == U_PF_ANDROID || U_PLATFORM == U_PF_QNX
82 #   define U_HAVE_NL_LANGINFO_CODESET 0
83 #else
84 #   define U_HAVE_NL_LANGINFO_CODESET 1
85 #endif
86 
87 #ifdef U_NL_LANGINFO_CODESET
88     /* Use the predefined value. */
89 #elif !U_HAVE_NL_LANGINFO_CODESET
90 #   define U_NL_LANGINFO_CODESET -1
91 #elif U_PLATFORM == U_PF_OS400
92    /* not defined */
93 #else
94 #   define U_NL_LANGINFO_CODESET CODESET
95 #endif
96 
97 #if defined(U_TZSET) || defined(U_HAVE_TZSET)
98     /* Use the predefined value. */
99 #elif U_PLATFORM_USES_ONLY_WIN32_API
100     // UWP doesn't support tzset or environment variables for tz
101 #if U_PLATFORM_HAS_WINUWP_API == 0
102 #   define U_TZSET _tzset
103 #endif
104 #elif U_PLATFORM == U_PF_OS400
105    /* not defined */
106 #elif defined(__wasi__)
107    /* not defined */
108 #else
109 #   define U_TZSET tzset
110 #endif
111 
112 #if defined(U_TIMEZONE) || defined(U_HAVE_TIMEZONE)
113     /* Use the predefined value. */
114 #elif U_PLATFORM == U_PF_ANDROID
115 #   define U_TIMEZONE timezone
116 #elif defined(__UCLIBC__)
117     // uClibc does not have __timezone or _timezone.
118 #elif defined(_NEWLIB_VERSION)
119 #   define U_TIMEZONE _timezone
120 #elif defined(__GLIBC__)
121     // glibc
122 #   define U_TIMEZONE __timezone
123 #elif U_PLATFORM_IS_LINUX_BASED
124     // not defined
125 #elif U_PLATFORM_USES_ONLY_WIN32_API
126 #   define U_TIMEZONE _timezone
127 #elif U_PLATFORM == U_PF_BSD && !defined(__NetBSD__)
128    /* not defined */
129 #elif U_PLATFORM == U_PF_OS400
130    /* not defined */
131 #elif U_PLATFORM == U_PF_IPHONE
132    /* not defined */
133 #elif defined(__wasi__)
134    /* not defined */
135 #else
136 #   define U_TIMEZONE timezone
137 #endif
138 
139 #if defined(U_TZNAME) || defined(U_HAVE_TZNAME)
140     /* Use the predefined value. */
141 #elif U_PLATFORM_USES_ONLY_WIN32_API
142     /* not usable on all windows platforms */
143 #if U_PLATFORM_HAS_WINUWP_API == 0
144 #   define U_TZNAME _tzname
145 #endif
146 #elif U_PLATFORM == U_PF_OS400
147    /* not defined */
148 #elif defined(__wasi__)
149    /* not defined */
150 #else
151 #   define U_TZNAME tzname
152 #endif
153 
154 #ifdef U_HAVE_MMAP
155     /* Use the predefined value. */
156 #elif U_PLATFORM_USES_ONLY_WIN32_API
157 #   define U_HAVE_MMAP 0
158 #else
159 #   define U_HAVE_MMAP 1
160 #endif
161 
162 #ifdef U_HAVE_POPEN
163     /* Use the predefined value. */
164 #elif U_PLATFORM_USES_ONLY_WIN32_API
165 #   define U_HAVE_POPEN 0
166 #elif U_PLATFORM == U_PF_OS400
167 #   define U_HAVE_POPEN 0
168 #else
169 #   define U_HAVE_POPEN 1
170 #endif
171 
172 /**
173  * \def U_HAVE_DIRENT_H
174  * Defines whether dirent.h is available.
175  * @internal
176  */
177 #ifdef U_HAVE_DIRENT_H
178     /* Use the predefined value. */
179 #elif U_PLATFORM_USES_ONLY_WIN32_API
180 #   define U_HAVE_DIRENT_H 0
181 #else
182 #   define U_HAVE_DIRENT_H 1
183 #endif
184 
185 /** @} */
186 
187 /*===========================================================================*/
188 /** @{ Programs used by ICU code                                             */
189 /*===========================================================================*/
190 
191 /**
192  * \def U_MAKE_IS_NMAKE
193  * Defines whether the "make" program is Windows nmake.
194  */
195 #ifdef U_MAKE_IS_NMAKE
196     /* Use the predefined value. */
197 #elif U_PLATFORM == U_PF_WINDOWS
198 #   define U_MAKE_IS_NMAKE 1
199 #else
200 #   define U_MAKE_IS_NMAKE 0
201 #endif
202 
203 /** @} */
204 
205 /*==========================================================================*/
206 /* Platform utilities                                                       */
207 /*==========================================================================*/
208 
209 /**
210  * Platform utilities isolates the platform dependencies of the
211  * library.  For each platform which this code is ported to, these
212  * functions may have to be re-implemented.
213  */
214 
215 /**
216  * Floating point utility to determine if a double is Not a Number (NaN).
217  * @internal
218  */
219 U_CAPI UBool   U_EXPORT2 uprv_isNaN(double d);
220 /**
221  * Floating point utility to determine if a double has an infinite value.
222  * @internal
223  */
224 U_CAPI UBool   U_EXPORT2 uprv_isInfinite(double d);
225 /**
226  * Floating point utility to determine if a double has a positive infinite value.
227  * @internal
228  */
229 U_CAPI UBool   U_EXPORT2 uprv_isPositiveInfinity(double d);
230 /**
231  * Floating point utility to determine if a double has a negative infinite value.
232  * @internal
233  */
234 U_CAPI UBool   U_EXPORT2 uprv_isNegativeInfinity(double d);
235 /**
236  * Floating point utility that returns a Not a Number (NaN) value.
237  * @internal
238  */
239 U_CAPI double  U_EXPORT2 uprv_getNaN(void);
240 /**
241  * Floating point utility that returns an infinite value.
242  * @internal
243  */
244 U_CAPI double  U_EXPORT2 uprv_getInfinity(void);
245 
246 /**
247  * Floating point utility to truncate a double.
248  * @internal
249  */
250 U_CAPI double  U_EXPORT2 uprv_trunc(double d);
251 /**
252  * Floating point utility to calculate the floor of a double.
253  * @internal
254  */
255 U_CAPI double  U_EXPORT2 uprv_floor(double d);
256 /**
257  * Floating point utility to calculate the ceiling of a double.
258  * @internal
259  */
260 U_CAPI double  U_EXPORT2 uprv_ceil(double d);
261 /**
262  * Floating point utility to calculate the absolute value of a double.
263  * @internal
264  */
265 U_CAPI double  U_EXPORT2 uprv_fabs(double d);
266 /**
267  * Floating point utility to calculate the fractional and integer parts of a double.
268  * @internal
269  */
270 U_CAPI double  U_EXPORT2 uprv_modf(double d, double* pinteger);
271 /**
272  * Floating point utility to calculate the remainder of a double divided by another double.
273  * @internal
274  */
275 U_CAPI double  U_EXPORT2 uprv_fmod(double d, double y);
276 /**
277  * Floating point utility to calculate d to the power of exponent (d^exponent).
278  * @internal
279  */
280 U_CAPI double  U_EXPORT2 uprv_pow(double d, double exponent);
281 /**
282  * Floating point utility to calculate 10 to the power of exponent (10^exponent).
283  * @internal
284  */
285 U_CAPI double  U_EXPORT2 uprv_pow10(int32_t exponent);
286 /**
287  * Floating point utility to calculate the maximum value of two doubles.
288  * @internal
289  */
290 U_CAPI double  U_EXPORT2 uprv_fmax(double d, double y);
291 /**
292  * Floating point utility to calculate the minimum value of two doubles.
293  * @internal
294  */
295 U_CAPI double  U_EXPORT2 uprv_fmin(double d, double y);
296 /**
297  * Private utility to calculate the maximum value of two integers.
298  * @internal
299  */
300 U_CAPI int32_t U_EXPORT2 uprv_max(int32_t d, int32_t y);
301 /**
302  * Private utility to calculate the minimum value of two integers.
303  * @internal
304  */
305 U_CAPI int32_t U_EXPORT2 uprv_min(int32_t d, int32_t y);
306 
307 #if U_IS_BIG_ENDIAN
308 #   define uprv_isNegative(number) (*((signed char *)&(number))<0)
309 #else
310 #   define uprv_isNegative(number) (*((signed char *)&(number)+sizeof(number)-1)<0)
311 #endif
312 
313 /**
314  * Return the largest positive number that can be represented by an integer
315  * type of arbitrary bit length.
316  * @internal
317  */
318 U_CAPI double  U_EXPORT2 uprv_maxMantissa(void);
319 
320 /**
321  * Floating point utility to calculate the logarithm of a double.
322  * @internal
323  */
324 U_CAPI double  U_EXPORT2 uprv_log(double d);
325 
326 /**
327  * Does common notion of rounding e.g. uprv_floor(x + 0.5);
328  * @param x the double number
329  * @return the rounded double
330  * @internal
331  */
332 U_CAPI double  U_EXPORT2 uprv_round(double x);
333 
334 /**
335  * Adds the signed integers a and b, storing the result in res.
336  * Checks for signed integer overflow.
337  * Similar to the GCC/Clang extension __builtin_add_overflow
338  *
339  * @param a The first operand.
340  * @param b The second operand.
341  * @param res a + b
342  * @return true if overflow occurred; false if no overflow occurred.
343  * @internal
344  */
345 U_CAPI UBool U_EXPORT2 uprv_add32_overflow(int32_t a, int32_t b, int32_t* res);
346 
347 /**
348  * Multiplies the signed integers a and b, storing the result in res.
349  * Checks for signed integer overflow.
350  * Similar to the GCC/Clang extension __builtin_mul_overflow
351  *
352  * @param a The first multiplicand.
353  * @param b The second multiplicand.
354  * @param res a * b
355  * @return true if overflow occurred; false if no overflow occurred.
356  * @internal
357  */
358 U_CAPI UBool U_EXPORT2 uprv_mul32_overflow(int32_t a, int32_t b, int32_t* res);
359 
360 #if 0
361 /**
362  * Returns the number of digits after the decimal point in a double number x.
363  *
364  * @param x the double number
365  * @return the number of digits after the decimal point in a double number x.
366  * @internal
367  */
368 /*U_CAPI int32_t  U_EXPORT2 uprv_digitsAfterDecimal(double x);*/
369 #endif
370 
371 #if !U_CHARSET_IS_UTF8
372 /**
373  * Please use ucnv_getDefaultName() instead.
374  * Return the default codepage for this platform and locale.
375  * This function can call setlocale() on Unix platforms. Please read the
376  * platform documentation on setlocale() before calling this function.
377  * @return the default codepage for this platform
378  * @internal
379  */
380 U_CAPI const char*  U_EXPORT2 uprv_getDefaultCodepage(void);
381 #endif
382 
383 /**
384  * Please use uloc_getDefault() instead.
385  * Return the default locale ID string by querying the system, or
386  *     zero if one cannot be found.
387  * This function can call setlocale() on Unix platforms. Please read the
388  * platform documentation on setlocale() before calling this function.
389  * @return the default locale ID string
390  * @internal
391  */
392 U_CAPI const char*  U_EXPORT2 uprv_getDefaultLocaleID(void);
393 
394 /**
395  * Time zone utilities
396  *
397  * Wrappers for C runtime library functions relating to timezones.
398  * The t_tzset() function (similar to tzset) uses the current setting
399  * of the environment variable TZ to assign values to three global
400  * variables: daylight, timezone, and tzname. These variables have the
401  * following meanings, and are declared in &lt;time.h&gt;.
402  *
403  *   daylight   Nonzero if daylight-saving-time zone (DST) is specified
404  *              in TZ; otherwise, 0. Default value is 1.
405  *   timezone   Difference in seconds between coordinated universal
406  *              time and local time. E.g., -28,800 for PST (GMT-8hrs)
407  *   tzname(0)  Three-letter time-zone name derived from TZ environment
408  *              variable. E.g., "PST".
409  *   tzname(1)  Three-letter DST zone name derived from TZ environment
410  *              variable.  E.g., "PDT". If DST zone is omitted from TZ,
411  *              tzname(1) is an empty string.
412  *
413  * Notes: For example, to set the TZ environment variable to correspond
414  * to the current time zone in Germany, you can use one of the
415  * following statements:
416  *
417  *   set TZ=GST1GDT
418  *   set TZ=GST+1GDT
419  *
420  * If the TZ value is not set, t_tzset() attempts to use the time zone
421  * information specified by the operating system. Under Windows NT
422  * and Windows 95, this information is specified in the Control Panel's
423  * Date/Time application.
424  * @internal
425  */
426 U_CAPI void     U_EXPORT2 uprv_tzset(void);
427 
428 /**
429  * Difference in seconds between coordinated universal
430  * time and local time. E.g., -28,800 for PST (GMT-8hrs)
431  * @return the difference in seconds between coordinated universal time and local time.
432  * @internal
433  */
434 U_CAPI int32_t  U_EXPORT2 uprv_timezone(void);
435 
436 /**
437  *   tzname(0)  Three-letter time-zone name derived from TZ environment
438  *              variable. E.g., "PST".
439  *   tzname(1)  Three-letter DST zone name derived from TZ environment
440  *              variable.  E.g., "PDT". If DST zone is omitted from TZ,
441  *              tzname(1) is an empty string.
442  * @internal
443  */
444 U_CAPI const char* U_EXPORT2 uprv_tzname(int n);
445 
446 /**
447  * Reset the global tzname cache.
448  * @internal
449  */
450 U_CAPI void uprv_tzname_clear_cache(void);
451 
452 /**
453  * Get UTC (GMT) time measured in milliseconds since 0:00 on 1/1/1970.
454  * This function is affected by 'faketime' and should be the bottleneck for all user-visible ICU time functions.
455  * @return the UTC time measured in milliseconds
456  * @internal
457  */
458 U_CAPI UDate U_EXPORT2 uprv_getUTCtime(void);
459 
460 /**
461  * Get UTC (GMT) time measured in milliseconds since 0:00 on 1/1/1970.
462  * This function is not affected by 'faketime', so it should only be used by low level test functions- not by anything that
463  * exposes time to the end user.
464  * @return the UTC time measured in milliseconds
465  * @internal
466  */
467 U_CAPI UDate U_EXPORT2 uprv_getRawUTCtime(void);
468 
469 /**
470  * Determine whether a pathname is absolute or not, as defined by the platform.
471  * @param path Pathname to test
472  * @return true if the path is absolute
473  * @internal (ICU 3.0)
474  */
475 U_CAPI UBool U_EXPORT2 uprv_pathIsAbsolute(const char *path);
476 
477 /**
478  * Use U_MAX_PTR instead of this function.
479  * @param void pointer to test
480  * @return the largest possible pointer greater than the base
481  * @internal (ICU 3.8)
482  */
483 U_CAPI void * U_EXPORT2 uprv_maximumPtr(void *base);
484 
485 /**
486  * Maximum value of a (void*) - use to indicate the limit of an 'infinite' buffer.
487  * In fact, buffer sizes must not exceed 2GB so that the difference between
488  * the buffer limit and the buffer start can be expressed in an int32_t.
489  *
490  * The definition of U_MAX_PTR must fulfill the following conditions:
491  * - return the largest possible pointer greater than base
492  * - return a valid pointer according to the machine architecture (AS/400, 64-bit, etc.)
493  * - avoid wrapping around at high addresses
494  * - make sure that the returned pointer is not farther from base than 0x7fffffff bytes
495  *
496  * @param base The beginning of a buffer to find the maximum offset from
497  * @internal
498  */
499 #ifndef U_MAX_PTR
500 #  if U_PLATFORM == U_PF_OS390 && !defined(_LP64)
501     /* We have 31-bit pointers. */
502 #    define U_MAX_PTR(base) ((void *)0x7fffffff)
503 #  elif U_PLATFORM == U_PF_OS400
504 #    define U_MAX_PTR(base) uprv_maximumPtr((void *)base)
505 #  elif 0
506     /*
507      * For platforms where pointers are scalar values (which is normal, but unlike i5/OS)
508      * but that do not define uintptr_t.
509      *
510      * However, this does not work on modern compilers:
511      * The C++ standard does not define pointer overflow, and allows compilers to
512      * assume that p+u>p for any pointer p and any integer u>0.
513      * Thus, modern compilers optimize away the ">" comparison.
514      * (See ICU tickets #7187 and #8096.)
515      */
516 #    define U_MAX_PTR(base) \
517     ((void *)(((char *)(base)+0x7fffffffu) > (char *)(base) \
518         ? ((char *)(base)+0x7fffffffu) \
519         : (char *)-1))
520 #  else
521     /* Default version. C++ standard compliant for scalar pointers. */
522 #    define U_MAX_PTR(base) \
523     ((void *)(((uintptr_t)(base)+0x7fffffffu) > (uintptr_t)(base) \
524         ? ((uintptr_t)(base)+0x7fffffffu) \
525         : (uintptr_t)-1))
526 #  endif
527 #endif
528 
529 
530 #ifdef __cplusplus
531 /**
532  * Pin a buffer capacity such that doing pointer arithmetic
533  * on the destination pointer and capacity cannot overflow.
534  *
535  * The pinned capacity must fulfill the following conditions (for positive capacities):
536  *   - dest + capacity is a valid pointer according to the machine architecture (AS/400, 64-bit, etc.)
537  *   - (dest + capacity) >= dest
538  *   - The size (in bytes) of T[capacity] does not exceed 0x7fffffff
539  *
540  * @param dest the destination buffer pointer.
541  * @param capacity the requested buffer capacity, in units of type T.
542  * @return the pinned capacity.
543  * @internal
544  */
545 template <typename T>
pinCapacity(T * dest,int32_t capacity)546 inline int32_t pinCapacity(T *dest, int32_t capacity) {
547     if (capacity <= 0) { return capacity; }
548 
549     uintptr_t destInt = (uintptr_t)dest;
550     uintptr_t maxInt;
551 
552 #  if U_PLATFORM == U_PF_OS390 && !defined(_LP64)
553     // We have 31-bit pointers.
554     maxInt = 0x7fffffff;
555 #  elif U_PLATFORM == U_PF_OS400
556     maxInt = (uintptr_t)uprv_maximumPtr((void *)dest);
557 #  else
558     maxInt = destInt + 0x7fffffffu;
559     if (maxInt < destInt) {
560         // Less than 2GB to the end of the address space.
561         // Pin to that to prevent address overflow.
562         maxInt = (uintptr_t)-1;
563     }
564 #  endif
565 
566     uintptr_t maxBytes = maxInt - destInt;  // max. 2GB
567     int32_t maxCapacity = (int32_t)(maxBytes / sizeof(T));
568     return capacity <= maxCapacity ? capacity : maxCapacity;
569 }
570 #endif   // __cplusplus
571 
572 /*  Dynamic Library Functions */
573 
574 typedef void (UVoidFunction)(void);
575 
576 #if U_ENABLE_DYLOAD
577 /**
578  * Load a library
579  * @internal (ICU 4.4)
580  */
581 U_CAPI void * U_EXPORT2 uprv_dl_open(const char *libName, UErrorCode *status);
582 
583 /**
584  * Close a library
585  * @internal (ICU 4.4)
586  */
587 U_CAPI void U_EXPORT2 uprv_dl_close( void *lib, UErrorCode *status);
588 
589 /**
590  * Extract a symbol from a library (function)
591  * @internal (ICU 4.8)
592  */
593 U_CAPI UVoidFunction* U_EXPORT2 uprv_dlsym_func( void *lib, const char *symbolName, UErrorCode *status);
594 
595 /**
596  * Extract a symbol from a library (function)
597  * Not implemented, no clients.
598  * @internal
599  */
600 /* U_CAPI void * U_EXPORT2 uprv_dlsym_data( void *lib, const char *symbolName, UErrorCode *status); */
601 
602 #endif
603 
604 /**
605  * Define malloc and related functions
606  * @internal
607  */
608 #if U_PLATFORM == U_PF_OS400
609 # define uprv_default_malloc(x) _C_TS_malloc(x)
610 # define uprv_default_realloc(x,y) _C_TS_realloc(x,y)
611 # define uprv_default_free(x) _C_TS_free(x)
612 /* also _C_TS_calloc(x) */
613 #else
614 /* C defaults */
615 # define uprv_default_malloc(x) malloc(x)
616 # define uprv_default_realloc(x,y) realloc(x,y)
617 # define uprv_default_free(x) free(x)
618 #endif
619 
620 
621 #endif
622