2001-01-31 18:13:16 +00:00
|
|
|
/*
|
|
|
|
**********************************************************************
|
2001-03-21 20:44:20 +00:00
|
|
|
* Copyright (C) 1999-2001, International Business Machines
|
2001-01-31 18:13:16 +00:00
|
|
|
* Corporation and others. All Rights Reserved.
|
|
|
|
**********************************************************************
|
|
|
|
* file name: ustr_imp.h
|
|
|
|
* encoding: US-ASCII
|
|
|
|
* tab size: 8 (not used)
|
|
|
|
* indentation:4
|
|
|
|
*
|
|
|
|
* created on: 2001jan30
|
|
|
|
* created by: Markus W. Scherer
|
|
|
|
*/
|
|
|
|
|
|
|
|
#ifndef __USTR_IMP_H__
|
|
|
|
#define __USTR_IMP_H__
|
|
|
|
|
|
|
|
#include "unicode/utypes.h"
|
2001-08-11 00:29:58 +00:00
|
|
|
#include "unicode/ucnv.h"
|
2002-02-20 02:07:29 +00:00
|
|
|
#include "unicode/uiter.h"
|
2001-01-31 18:13:16 +00:00
|
|
|
|
2002-02-21 17:58:51 +00:00
|
|
|
/** Simple declaration for u_strToTitle() to avoid including unicode/ubrk.h. */
|
2002-02-22 00:14:13 +00:00
|
|
|
#ifndef UBRK_TYPEDEF_UBREAK_ITERATOR
|
|
|
|
# define UBRK_TYPEDEF_UBREAK_ITERATOR
|
|
|
|
typedef void *UBreakIterator;
|
|
|
|
#endif
|
2002-02-21 17:58:51 +00:00
|
|
|
|
2002-05-13 23:01:38 +00:00
|
|
|
/**
|
2002-05-22 00:35:01 +00:00
|
|
|
* Bit mask for getting just the options from a string compare options word
|
|
|
|
* that are relevant for case-insensitive string comparison.
|
|
|
|
* See uchar.h. Also include _STRNCMP_STYLE and U_COMPARE_CODE_POINT_ORDER.
|
|
|
|
* @internal
|
|
|
|
*/
|
|
|
|
#define _STRCASECMP_OPTIONS_MASK 0xffff
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Bit mask for getting just the options from a string compare options word
|
|
|
|
* that are relevant for case folding (of a single string or code point).
|
|
|
|
* See uchar.h.
|
|
|
|
* @internal
|
|
|
|
*/
|
|
|
|
#define _FOLD_CASE_OPTIONS_MASK 0xff
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Compare two strings in code point order or code unit order.
|
2002-05-13 23:01:38 +00:00
|
|
|
* Works in strcmp style (both lengths -1),
|
|
|
|
* strncmp style (lengths equal and >=0, flag TRUE),
|
|
|
|
* and memcmp/UnicodeString style (at least one length >=0).
|
|
|
|
* @internal
|
|
|
|
*/
|
2002-05-22 00:35:01 +00:00
|
|
|
U_CAPI int32_t U_EXPORT2
|
2002-05-24 17:05:31 +00:00
|
|
|
uprv_strCompare(const UChar *s1, int32_t length1,
|
|
|
|
const UChar *s2, int32_t length2,
|
|
|
|
UBool strncmpStyle, UBool codePointOrder);
|
2002-05-13 23:01:38 +00:00
|
|
|
|
2001-02-14 00:47:36 +00:00
|
|
|
/**
|
|
|
|
* Are the Unicode properties loaded?
|
|
|
|
* This must be used before internal functions are called that do
|
|
|
|
* not perform this check.
|
|
|
|
* @internal
|
|
|
|
*/
|
|
|
|
U_CFUNC UBool
|
2001-02-27 23:45:16 +00:00
|
|
|
uprv_haveProperties(void);
|
2001-02-14 00:47:36 +00:00
|
|
|
|
2001-01-31 18:13:16 +00:00
|
|
|
/**
|
|
|
|
* Type of a function that may be passed to the internal case mapping functions
|
2001-06-25 22:59:23 +00:00
|
|
|
* or similar for growing the destination buffer.
|
2001-02-14 00:47:36 +00:00
|
|
|
* @internal
|
2001-01-31 18:13:16 +00:00
|
|
|
*/
|
2001-02-26 19:22:32 +00:00
|
|
|
typedef UBool U_CALLCONV
|
2001-08-17 00:18:25 +00:00
|
|
|
UGrowBuffer(void *context, /* opaque pointer for this function */
|
|
|
|
UChar **pBuffer, /* in/out destination buffer pointer */
|
|
|
|
int32_t *pCapacity, /* in/out buffer capacity in numbers of UChars */
|
|
|
|
int32_t reqCapacity,/* requested capacity */
|
|
|
|
int32_t length); /* number of UChars to be copied to new buffer */
|
2001-01-31 18:13:16 +00:00
|
|
|
|
2001-06-25 22:59:23 +00:00
|
|
|
/**
|
2001-08-17 00:18:25 +00:00
|
|
|
* Default implementation of UGrowBuffer.
|
2001-06-25 22:59:23 +00:00
|
|
|
* Takes a static buffer as context, allocates a new buffer,
|
|
|
|
* and releases the old one if it is not the same as the one passed as context.
|
|
|
|
* @internal
|
|
|
|
*/
|
2001-07-10 16:43:51 +00:00
|
|
|
U_CAPI UBool /* U_CALLCONV U_EXPORT2 */
|
2001-06-25 22:59:23 +00:00
|
|
|
u_growBufferFromStatic(void *context,
|
|
|
|
UChar **pBuffer, int32_t *pCapacity, int32_t reqCapacity,
|
|
|
|
int32_t length);
|
|
|
|
|
2001-01-31 18:13:16 +00:00
|
|
|
/*
|
2002-02-21 17:58:51 +00:00
|
|
|
* Internal string casing functions implementing
|
|
|
|
* ustring.h/ustrcase.c and UnicodeString case mapping functions.
|
|
|
|
*
|
|
|
|
* Lowercases [srcStart..srcLimit[ but takes
|
|
|
|
* context [0..srcLength[ into account.
|
2001-06-25 22:59:23 +00:00
|
|
|
* @internal
|
2001-01-31 18:13:16 +00:00
|
|
|
*/
|
|
|
|
U_CFUNC int32_t
|
|
|
|
u_internalStrToLower(UChar *dest, int32_t destCapacity,
|
|
|
|
const UChar *src, int32_t srcLength,
|
2002-02-21 17:58:51 +00:00
|
|
|
int32_t srcStart, int32_t srcLimit,
|
2001-01-31 18:13:16 +00:00
|
|
|
const char *locale,
|
|
|
|
UErrorCode *pErrorCode);
|
|
|
|
|
2001-06-25 22:59:23 +00:00
|
|
|
/**
|
|
|
|
* @internal
|
|
|
|
*/
|
2001-01-31 18:13:16 +00:00
|
|
|
U_CFUNC int32_t
|
|
|
|
u_internalStrToUpper(UChar *dest, int32_t destCapacity,
|
|
|
|
const UChar *src, int32_t srcLength,
|
|
|
|
const char *locale,
|
|
|
|
UErrorCode *pErrorCode);
|
|
|
|
|
2002-09-20 01:54:48 +00:00
|
|
|
#if !UCONFIG_NO_BREAK_ITERATION
|
|
|
|
|
2002-02-21 17:58:51 +00:00
|
|
|
/**
|
|
|
|
* @internal
|
|
|
|
*/
|
|
|
|
U_CFUNC int32_t
|
|
|
|
u_internalStrToTitle(UChar *dest, int32_t destCapacity,
|
|
|
|
const UChar *src, int32_t srcLength,
|
|
|
|
UBreakIterator *titleIter,
|
|
|
|
const char *locale,
|
|
|
|
UErrorCode *pErrorCode);
|
|
|
|
|
2002-09-20 01:54:48 +00:00
|
|
|
#endif
|
|
|
|
|
2001-02-14 00:47:36 +00:00
|
|
|
/**
|
|
|
|
* Internal case folding function.
|
|
|
|
* @internal
|
|
|
|
*/
|
|
|
|
U_CFUNC int32_t
|
|
|
|
u_internalStrFoldCase(UChar *dest, int32_t destCapacity,
|
|
|
|
const UChar *src, int32_t srcLength,
|
|
|
|
uint32_t options,
|
|
|
|
UErrorCode *pErrorCode);
|
|
|
|
|
2001-11-29 20:44:54 +00:00
|
|
|
/**
|
2002-02-20 02:07:29 +00:00
|
|
|
* Get the full lowercase mapping for c.
|
|
|
|
* @param iter Character iterator to check for context for SpecialCasing.
|
|
|
|
* The current index must be on the character after c.
|
|
|
|
* This function may or may not change the iterator index.
|
|
|
|
* If iter==NULL then a context-independent result is returned.
|
2001-11-29 20:44:54 +00:00
|
|
|
* @return the length of the output, negative if same as c
|
|
|
|
* @internal
|
|
|
|
*/
|
|
|
|
U_CAPI int32_t U_EXPORT2
|
2002-02-20 02:07:29 +00:00
|
|
|
u_internalToLower(UChar32 c, UCharIterator *iter,
|
|
|
|
UChar *dest, int32_t destCapacity,
|
|
|
|
const char *locale);
|
2001-11-29 20:44:54 +00:00
|
|
|
|
2001-02-14 00:47:36 +00:00
|
|
|
/**
|
2002-02-20 02:07:29 +00:00
|
|
|
* Get the full uppercase mapping for c.
|
|
|
|
* @param iter Character iterator to check for context for SpecialCasing.
|
|
|
|
* The current index must be on the character after c.
|
|
|
|
* This function may or may not change the iterator index.
|
|
|
|
* If iter==NULL then a context-independent result is returned.
|
|
|
|
* @return the length of the output, negative if same as c
|
2001-02-14 00:47:36 +00:00
|
|
|
* @internal
|
|
|
|
*/
|
2002-02-20 02:07:29 +00:00
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
|
|
u_internalToUpper(UChar32 c, UCharIterator *iter,
|
|
|
|
UChar *dest, int32_t destCapacity,
|
|
|
|
const char *locale);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get the full titlecase mapping for c.
|
|
|
|
* @param iter Character iterator to check for context for SpecialCasing.
|
|
|
|
* The current index must be on the character after c.
|
|
|
|
* This function may or may not change the iterator index.
|
|
|
|
* If iter==NULL then a context-independent result is returned.
|
|
|
|
* @return the length of the output, negative if same as c
|
|
|
|
* @internal
|
|
|
|
*/
|
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
|
|
u_internalToTitle(UChar32 c, UCharIterator *iter,
|
|
|
|
UChar *dest, int32_t destCapacity,
|
|
|
|
const char *locale);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get the full case folding mapping for c.
|
|
|
|
* @return the length of the output, negative if same as c
|
|
|
|
* @internal
|
|
|
|
*/
|
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
|
|
u_internalFoldCase(UChar32 c,
|
|
|
|
UChar *dest, int32_t destCapacity,
|
|
|
|
uint32_t options);
|
2001-02-14 00:47:36 +00:00
|
|
|
|
2001-08-11 00:29:58 +00:00
|
|
|
/**
|
|
|
|
* Get the default converter. This is a commonly used converter
|
|
|
|
* that is used for the ustring and UnicodeString API.
|
|
|
|
* Remember to use the u_releaseDefaultConverter when you are done.
|
|
|
|
* @internal
|
|
|
|
*/
|
2001-09-05 23:37:16 +00:00
|
|
|
U_CAPI UConverter* U_EXPORT2
|
2001-08-11 00:29:58 +00:00
|
|
|
u_getDefaultConverter(UErrorCode *status);
|
|
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Release the default converter to the converter cache.
|
|
|
|
* @internal
|
|
|
|
*/
|
2001-09-05 23:37:16 +00:00
|
|
|
U_CAPI void U_EXPORT2
|
2001-08-11 00:29:58 +00:00
|
|
|
u_releaseDefaultConverter(UConverter *converter);
|
|
|
|
|
2001-09-05 23:37:16 +00:00
|
|
|
/**
|
|
|
|
* NUL-terminate a UChar * string if possible.
|
|
|
|
* If length < destCapacity then NUL-terminate.
|
|
|
|
* If length == destCapacity then do not terminate but set U_STRING_NOT_TERMINATED_WARNING.
|
|
|
|
* If length > destCapacity then do not terminate but set U_BUFFER_OVERFLOW_ERROR.
|
|
|
|
*
|
|
|
|
* @param dest Destination buffer, can be NULL if destCapacity==0.
|
|
|
|
* @param destCapacity Number of UChars available at dest.
|
|
|
|
* @param length Number of UChars that were (to be) written to dest.
|
|
|
|
* @param pErrorCode ICU error code.
|
|
|
|
* @return length
|
|
|
|
* @internal
|
|
|
|
*/
|
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
|
|
u_terminateUChars(UChar *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
|
|
|
|
|
2001-09-18 00:56:43 +00:00
|
|
|
/**
|
|
|
|
* NUL-terminate a char * string if possible.
|
|
|
|
* Same as u_terminateUChars() but for a different string type.
|
|
|
|
*/
|
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
|
|
u_terminateChars(char *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* NUL-terminate a UChar32 * string if possible.
|
|
|
|
* Same as u_terminateUChars() but for a different string type.
|
|
|
|
*/
|
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
|
|
u_terminateUChar32s(UChar32 *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* NUL-terminate a wchar_t * string if possible.
|
|
|
|
* Same as u_terminateUChars() but for a different string type.
|
|
|
|
*/
|
|
|
|
U_CAPI int32_t U_EXPORT2
|
|
|
|
u_terminateWChars(wchar_t *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
|
|
|
|
|
2001-11-29 19:27:40 +00:00
|
|
|
#define u_getMaxCaseExpansion() 10
|
|
|
|
|
2002-03-18 23:19:41 +00:00
|
|
|
/**
|
|
|
|
* Find a single (unmatched) surrogate code point in the string s[0..length[ .
|
|
|
|
* Find the first such surrogate.
|
|
|
|
* @internal
|
|
|
|
*/
|
|
|
|
U_CFUNC const UChar *
|
|
|
|
uprv_strFindSurrogate(const UChar *s, int32_t length, UChar surrogate);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Find a single (unmatched) surrogate code point in the string s[0..length[ .
|
|
|
|
* Find the last such surrogate.
|
|
|
|
* @internal
|
|
|
|
*/
|
|
|
|
U_CFUNC const UChar *
|
|
|
|
uprv_strFindLastSurrogate(const UChar *s, int32_t length, UChar surrogate);
|
|
|
|
|
2001-01-31 18:13:16 +00:00
|
|
|
#endif
|