michael@0: /*  
michael@0: **********************************************************************
michael@0: *   Copyright (C) 1999-2011, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: **********************************************************************
michael@0: *   file name:  ustr_imp.h
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 2001jan30
michael@0: *   created by: Markus W. Scherer
michael@0: */
michael@0: 
michael@0: #ifndef __USTR_IMP_H__
michael@0: #define __USTR_IMP_H__
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/uiter.h"
michael@0: #include "ucase.h"
michael@0: 
michael@0: /** Simple declaration to avoid including unicode/ubrk.h. */
michael@0: #ifndef UBRK_TYPEDEF_UBREAK_ITERATOR
michael@0: #   define UBRK_TYPEDEF_UBREAK_ITERATOR
michael@0:     typedef struct UBreakIterator UBreakIterator;
michael@0: #endif
michael@0: 
michael@0: #ifndef U_COMPARE_IGNORE_CASE
michael@0: /* see also unorm.h */
michael@0: /**
michael@0:  * Option bit for unorm_compare:
michael@0:  * Perform case-insensitive comparison.
michael@0:  */
michael@0: #define U_COMPARE_IGNORE_CASE       0x10000
michael@0: #endif
michael@0: 
michael@0: /**
michael@0:  * Internal option for unorm_cmpEquivFold() for strncmp style.
michael@0:  * If set, checks for both string length and terminating NUL.
michael@0:  */
michael@0: #define _STRNCMP_STYLE 0x1000
michael@0: 
michael@0: /**
michael@0:  * Compare two strings in code point order or code unit order.
michael@0:  * Works in strcmp style (both lengths -1),
michael@0:  * strncmp style (lengths equal and >=0, flag TRUE),
michael@0:  * and memcmp/UnicodeString style (at least one length >=0).
michael@0:  */
michael@0: U_CFUNC int32_t U_EXPORT2
michael@0: uprv_strCompare(const UChar *s1, int32_t length1,
michael@0:                 const UChar *s2, int32_t length2,
michael@0:                 UBool strncmpStyle, UBool codePointOrder);
michael@0: 
michael@0: /**
michael@0:  * Internal API, used by u_strcasecmp() etc.
michael@0:  * Compare strings case-insensitively,
michael@0:  * in code point order or code unit order.
michael@0:  */
michael@0: U_CFUNC int32_t
michael@0: u_strcmpFold(const UChar *s1, int32_t length1,
michael@0:              const UChar *s2, int32_t length2,
michael@0:              uint32_t options,
michael@0:              UErrorCode *pErrorCode);
michael@0: 
michael@0: /**
michael@0:  * Are the Unicode properties loaded?
michael@0:  * This must be used before internal functions are called that do
michael@0:  * not perform this check.
michael@0:  * Generate a debug assertion failure if data is not loaded.
michael@0:  */
michael@0: U_CFUNC UBool
michael@0: uprv_haveProperties(UErrorCode *pErrorCode);
michael@0: 
michael@0: /**
michael@0:   * Load the Unicode property data.
michael@0:   * Intended primarily for use from u_init().
michael@0:   * Has no effect if property data is already loaded.
michael@0:   * NOT thread safe.
michael@0:   */
michael@0: /*U_CFUNC int8_t
michael@0: uprv_loadPropsData(UErrorCode *errorCode);*/
michael@0: 
michael@0: /*
michael@0:  * Internal string casing functions implementing
michael@0:  * ustring.h/ustrcase.c and UnicodeString case mapping functions.
michael@0:  */
michael@0: 
michael@0: struct UCaseMap {
michael@0:     const UCaseProps *csp;
michael@0: #if !UCONFIG_NO_BREAK_ITERATION
michael@0:     UBreakIterator *iter;  /* We adopt the iterator, so we own it. */
michael@0: #endif
michael@0:     char locale[32];
michael@0:     int32_t locCache;
michael@0:     uint32_t options;
michael@0: };
michael@0: 
michael@0: #ifndef __UCASEMAP_H__
michael@0: typedef struct UCaseMap UCaseMap;
michael@0: #endif
michael@0: 
michael@0: #if UCONFIG_NO_BREAK_ITERATION
michael@0: #   define UCASEMAP_INITIALIZER { NULL, { 0 }, 0, 0 }
michael@0: #else
michael@0: #   define UCASEMAP_INITIALIZER { NULL, NULL, { 0 }, 0, 0 }
michael@0: #endif
michael@0: 
michael@0: U_CFUNC void
michael@0: ustrcase_setTempCaseMapLocale(UCaseMap *csm, const char *locale);
michael@0: 
michael@0: #ifndef U_STRING_CASE_MAPPER_DEFINED
michael@0: #define U_STRING_CASE_MAPPER_DEFINED
michael@0: 
michael@0: /**
michael@0:  * String case mapping function type, used by ustrcase_map().
michael@0:  * All error checking must be done.
michael@0:  * The UCaseMap must be fully initialized, with locale and/or iter set as needed.
michael@0:  * src and dest must not overlap.
michael@0:  */
michael@0: typedef int32_t U_CALLCONV
michael@0: UStringCaseMapper(const UCaseMap *csm,
michael@0:                   UChar *dest, int32_t destCapacity,
michael@0:                   const UChar *src, int32_t srcLength,
michael@0:                   UErrorCode *pErrorCode);
michael@0: 
michael@0: #endif
michael@0: 
michael@0: /** Implements UStringCaseMapper. */
michael@0: U_CFUNC int32_t U_CALLCONV
michael@0: ustrcase_internalToLower(const UCaseMap *csm,
michael@0:                          UChar *dest, int32_t destCapacity,
michael@0:                          const UChar *src, int32_t srcLength,
michael@0:                          UErrorCode *pErrorCode);
michael@0: 
michael@0: /** Implements UStringCaseMapper. */
michael@0: U_CFUNC int32_t U_CALLCONV
michael@0: ustrcase_internalToUpper(const UCaseMap *csm,
michael@0:                          UChar *dest, int32_t destCapacity,
michael@0:                          const UChar *src, int32_t srcLength,
michael@0:                          UErrorCode *pErrorCode);
michael@0: 
michael@0: #if !UCONFIG_NO_BREAK_ITERATION
michael@0: 
michael@0: /** Implements UStringCaseMapper. */
michael@0: U_CFUNC int32_t U_CALLCONV
michael@0: ustrcase_internalToTitle(const UCaseMap *csm,
michael@0:                          UChar *dest, int32_t destCapacity,
michael@0:                          const UChar *src, int32_t srcLength,
michael@0:                          UErrorCode *pErrorCode);
michael@0: 
michael@0: #endif
michael@0: 
michael@0: /** Implements UStringCaseMapper. */
michael@0: U_CFUNC int32_t U_CALLCONV
michael@0: ustrcase_internalFold(const UCaseMap *csm,
michael@0:                       UChar *dest, int32_t destCapacity,
michael@0:                       const UChar *src, int32_t srcLength,
michael@0:                       UErrorCode *pErrorCode);
michael@0: 
michael@0: /**
michael@0:  * Implements argument checking and buffer handling
michael@0:  * for string case mapping as a common function.
michael@0:  */
michael@0: U_CFUNC int32_t
michael@0: ustrcase_map(const UCaseMap *csm,
michael@0:              UChar *dest, int32_t destCapacity,
michael@0:              const UChar *src, int32_t srcLength,
michael@0:              UStringCaseMapper *stringCaseMapper,
michael@0:              UErrorCode *pErrorCode);
michael@0: 
michael@0: /**
michael@0:  * UTF-8 string case mapping function type, used by ucasemap_mapUTF8().
michael@0:  * UTF-8 version of UStringCaseMapper.
michael@0:  * All error checking must be done.
michael@0:  * The UCaseMap must be fully initialized, with locale and/or iter set as needed.
michael@0:  * src and dest must not overlap.
michael@0:  */
michael@0: typedef int32_t U_CALLCONV
michael@0: UTF8CaseMapper(const UCaseMap *csm,
michael@0:                uint8_t *dest, int32_t destCapacity,
michael@0:                const uint8_t *src, int32_t srcLength,
michael@0:                UErrorCode *pErrorCode);
michael@0: 
michael@0: /** Implements UTF8CaseMapper. */
michael@0: U_CFUNC int32_t U_CALLCONV
michael@0: ucasemap_internalUTF8ToTitle(const UCaseMap *csm,
michael@0:          uint8_t *dest, int32_t destCapacity,
michael@0:          const uint8_t *src, int32_t srcLength,
michael@0:          UErrorCode *pErrorCode);
michael@0: 
michael@0: /**
michael@0:  * Implements argument checking and buffer handling
michael@0:  * for UTF-8 string case mapping as a common function.
michael@0:  */
michael@0: U_CFUNC int32_t
michael@0: ucasemap_mapUTF8(const UCaseMap *csm,
michael@0:                  uint8_t *dest, int32_t destCapacity,
michael@0:                  const uint8_t *src, int32_t srcLength,
michael@0:                  UTF8CaseMapper *stringCaseMapper,
michael@0:                  UErrorCode *pErrorCode);
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2 
michael@0: ustr_hashUCharsN(const UChar *str, int32_t length);
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2 
michael@0: ustr_hashCharsN(const char *str, int32_t length);
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: ustr_hashICharsN(const char *str, int32_t length);
michael@0: 
michael@0: /**
michael@0:  * NUL-terminate a UChar * string if possible.
michael@0:  * If length  < destCapacity then NUL-terminate.
michael@0:  * If length == destCapacity then do not terminate but set U_STRING_NOT_TERMINATED_WARNING.
michael@0:  * If length  > destCapacity then do not terminate but set U_BUFFER_OVERFLOW_ERROR.
michael@0:  *
michael@0:  * @param dest Destination buffer, can be NULL if destCapacity==0.
michael@0:  * @param destCapacity Number of UChars available at dest.
michael@0:  * @param length Number of UChars that were (to be) written to dest.
michael@0:  * @param pErrorCode ICU error code.
michael@0:  * @return length
michael@0:  */
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_terminateUChars(UChar *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
michael@0: 
michael@0: /**
michael@0:  * NUL-terminate a char * string if possible.
michael@0:  * Same as u_terminateUChars() but for a different string type.
michael@0:  */
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_terminateChars(char *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
michael@0: 
michael@0: /**
michael@0:  * NUL-terminate a UChar32 * string if possible.
michael@0:  * Same as u_terminateUChars() but for a different string type.
michael@0:  */
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_terminateUChar32s(UChar32 *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
michael@0: 
michael@0: /**
michael@0:  * NUL-terminate a wchar_t * string if possible.
michael@0:  * Same as u_terminateUChars() but for a different string type.
michael@0:  */
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_terminateWChars(wchar_t *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode);
michael@0: 
michael@0: #endif