jpayne@69: // © 2016 and later: Unicode, Inc. and others.
jpayne@69: // License & terms of use: http://www.unicode.org/copyright.html
jpayne@69: /*
jpayne@69: *******************************************************************************
jpayne@69: *
jpayne@69: *   Copyright (C) 2005-2012, International Business Machines
jpayne@69: *   Corporation and others.  All Rights Reserved.
jpayne@69: *
jpayne@69: *******************************************************************************
jpayne@69: *   file name:  ucasemap.h
jpayne@69: *   encoding:   UTF-8
jpayne@69: *   tab size:   8 (not used)
jpayne@69: *   indentation:4
jpayne@69: *
jpayne@69: *   created on: 2005may06
jpayne@69: *   created by: Markus W. Scherer
jpayne@69: *
jpayne@69: *   Case mapping service object and functions using it.
jpayne@69: */
jpayne@69: 
jpayne@69: #ifndef __UCASEMAP_H__
jpayne@69: #define __UCASEMAP_H__
jpayne@69: 
jpayne@69: #include "unicode/utypes.h"
jpayne@69: #include "unicode/localpointer.h"
jpayne@69: #include "unicode/stringoptions.h"
jpayne@69: #include "unicode/ustring.h"
jpayne@69: 
jpayne@69: /**
jpayne@69:  * \file
jpayne@69:  * \brief C API: Unicode case mapping functions using a UCaseMap service object.
jpayne@69:  *
jpayne@69:  * The service object takes care of memory allocations, data loading, and setup
jpayne@69:  * for the attributes, as usual.
jpayne@69:  *
jpayne@69:  * Currently, the functionality provided here does not overlap with uchar.h
jpayne@69:  * and ustring.h, except for ucasemap_toTitle().
jpayne@69:  *
jpayne@69:  * ucasemap_utf8XYZ() functions operate directly on UTF-8 strings.
jpayne@69:  */
jpayne@69: 
jpayne@69: /**
jpayne@69:  * UCaseMap is an opaque service object for newer ICU case mapping functions.
jpayne@69:  * Older functions did not use a service object.
jpayne@69:  * @stable ICU 3.4
jpayne@69:  */
jpayne@69: struct UCaseMap;
jpayne@69: typedef struct UCaseMap UCaseMap; /**< C typedef for struct UCaseMap. @stable ICU 3.4 */
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Open a UCaseMap service object for a locale and a set of options.
jpayne@69:  * The locale ID and options are preprocessed so that functions using the
jpayne@69:  * service object need not process them in each call.
jpayne@69:  *
jpayne@69:  * @param locale ICU locale ID, used for language-dependent
jpayne@69:  *               upper-/lower-/title-casing according to the Unicode standard.
jpayne@69:  *               Usual semantics: ""=root, NULL=default locale, etc.
jpayne@69:  * @param options Options bit set, used for case folding and string comparisons.
jpayne@69:  *                Same flags as for u_foldCase(), u_strFoldCase(),
jpayne@69:  *                u_strCaseCompare(), etc.
jpayne@69:  *                Use 0 or U_FOLD_CASE_DEFAULT for default behavior.
jpayne@69:  * @param pErrorCode Must be a valid pointer to an error code value,
jpayne@69:  *                   which must not indicate a failure before the function call.
jpayne@69:  * @return Pointer to a UCaseMap service object, if successful.
jpayne@69:  *
jpayne@69:  * @see U_FOLD_CASE_DEFAULT
jpayne@69:  * @see U_FOLD_CASE_EXCLUDE_SPECIAL_I
jpayne@69:  * @see U_TITLECASE_NO_LOWERCASE
jpayne@69:  * @see U_TITLECASE_NO_BREAK_ADJUSTMENT
jpayne@69:  * @stable ICU 3.4
jpayne@69:  */
jpayne@69: U_STABLE UCaseMap * U_EXPORT2
jpayne@69: ucasemap_open(const char *locale, uint32_t options, UErrorCode *pErrorCode);
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Close a UCaseMap service object.
jpayne@69:  * @param csm Object to be closed.
jpayne@69:  * @stable ICU 3.4
jpayne@69:  */
jpayne@69: U_STABLE void U_EXPORT2
jpayne@69: ucasemap_close(UCaseMap *csm);
jpayne@69: 
jpayne@69: #if U_SHOW_CPLUSPLUS_API
jpayne@69: 
jpayne@69: U_NAMESPACE_BEGIN
jpayne@69: 
jpayne@69: /**
jpayne@69:  * \class LocalUCaseMapPointer
jpayne@69:  * "Smart pointer" class, closes a UCaseMap via ucasemap_close().
jpayne@69:  * For most methods see the LocalPointerBase base class.
jpayne@69:  *
jpayne@69:  * @see LocalPointerBase
jpayne@69:  * @see LocalPointer
jpayne@69:  * @stable ICU 4.4
jpayne@69:  */
jpayne@69: U_DEFINE_LOCAL_OPEN_POINTER(LocalUCaseMapPointer, UCaseMap, ucasemap_close);
jpayne@69: 
jpayne@69: U_NAMESPACE_END
jpayne@69: 
jpayne@69: #endif
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Get the locale ID that is used for language-dependent case mappings.
jpayne@69:  * @param csm UCaseMap service object.
jpayne@69:  * @return locale ID
jpayne@69:  * @stable ICU 3.4
jpayne@69:  */
jpayne@69: U_STABLE const char * U_EXPORT2
jpayne@69: ucasemap_getLocale(const UCaseMap *csm);
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Get the options bit set that is used for case folding and string comparisons.
jpayne@69:  * @param csm UCaseMap service object.
jpayne@69:  * @return options bit set
jpayne@69:  * @stable ICU 3.4
jpayne@69:  */
jpayne@69: U_STABLE uint32_t U_EXPORT2
jpayne@69: ucasemap_getOptions(const UCaseMap *csm);
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Set the locale ID that is used for language-dependent case mappings.
jpayne@69:  *
jpayne@69:  * @param csm UCaseMap service object.
jpayne@69:  * @param locale Locale ID, see ucasemap_open().
jpayne@69:  * @param pErrorCode Must be a valid pointer to an error code value,
jpayne@69:  *                   which must not indicate a failure before the function call.
jpayne@69:  *
jpayne@69:  * @see ucasemap_open
jpayne@69:  * @stable ICU 3.4
jpayne@69:  */
jpayne@69: U_STABLE void U_EXPORT2
jpayne@69: ucasemap_setLocale(UCaseMap *csm, const char *locale, UErrorCode *pErrorCode);
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Set the options bit set that is used for case folding and string comparisons.
jpayne@69:  *
jpayne@69:  * @param csm UCaseMap service object.
jpayne@69:  * @param options Options bit set, see ucasemap_open().
jpayne@69:  * @param pErrorCode Must be a valid pointer to an error code value,
jpayne@69:  *                   which must not indicate a failure before the function call.
jpayne@69:  *
jpayne@69:  * @see ucasemap_open
jpayne@69:  * @stable ICU 3.4
jpayne@69:  */
jpayne@69: U_STABLE void U_EXPORT2
jpayne@69: ucasemap_setOptions(UCaseMap *csm, uint32_t options, UErrorCode *pErrorCode);
jpayne@69: 
jpayne@69: #if !UCONFIG_NO_BREAK_ITERATION
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Get the break iterator that is used for titlecasing.
jpayne@69:  * Do not modify the returned break iterator.
jpayne@69:  * @param csm UCaseMap service object.
jpayne@69:  * @return titlecasing break iterator
jpayne@69:  * @stable ICU 3.8
jpayne@69:  */
jpayne@69: U_STABLE const UBreakIterator * U_EXPORT2
jpayne@69: ucasemap_getBreakIterator(const UCaseMap *csm);
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Set the break iterator that is used for titlecasing.
jpayne@69:  * The UCaseMap service object releases a previously set break iterator
jpayne@69:  * and "adopts" this new one, taking ownership of it.
jpayne@69:  * It will be released in a subsequent call to ucasemap_setBreakIterator()
jpayne@69:  * or ucasemap_close().
jpayne@69:  *
jpayne@69:  * Break iterator operations are not thread-safe. Therefore, titlecasing
jpayne@69:  * functions use non-const UCaseMap objects. It is not possible to titlecase
jpayne@69:  * strings concurrently using the same UCaseMap.
jpayne@69:  *
jpayne@69:  * @param csm UCaseMap service object.
jpayne@69:  * @param iterToAdopt Break iterator to be adopted for titlecasing.
jpayne@69:  * @param pErrorCode Must be a valid pointer to an error code value,
jpayne@69:  *                   which must not indicate a failure before the function call.
jpayne@69:  *
jpayne@69:  * @see ucasemap_toTitle
jpayne@69:  * @see ucasemap_utf8ToTitle
jpayne@69:  * @stable ICU 3.8
jpayne@69:  */
jpayne@69: U_STABLE void U_EXPORT2
jpayne@69: ucasemap_setBreakIterator(UCaseMap *csm, UBreakIterator *iterToAdopt, UErrorCode *pErrorCode);
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Titlecase a UTF-16 string. This function is almost a duplicate of u_strToTitle(),
jpayne@69:  * except that it takes ucasemap_setOptions() into account and has performance
jpayne@69:  * advantages from being able to use a UCaseMap object for multiple case mapping
jpayne@69:  * operations, saving setup time.
jpayne@69:  *
jpayne@69:  * Casing is locale-dependent and context-sensitive.
jpayne@69:  * Titlecasing uses a break iterator to find the first characters of words
jpayne@69:  * that are to be titlecased. It titlecases those characters and lowercases
jpayne@69:  * all others. (This can be modified with ucasemap_setOptions().)
jpayne@69:  *
jpayne@69:  * Note: This function takes a non-const UCaseMap pointer because it will
jpayne@69:  * open a default break iterator if no break iterator was set yet,
jpayne@69:  * and effectively call ucasemap_setBreakIterator();
jpayne@69:  * also because the break iterator is stateful and will be modified during
jpayne@69:  * the iteration.
jpayne@69:  *
jpayne@69:  * The titlecase break iterator can be provided to customize for arbitrary
jpayne@69:  * styles, using rules and dictionaries beyond the standard iterators.
jpayne@69:  * The standard titlecase iterator for the root locale implements the
jpayne@69:  * algorithm of Unicode TR 21.
jpayne@69:  *
jpayne@69:  * This function uses only the setText(), first() and next() methods of the
jpayne@69:  * provided break iterator.
jpayne@69:  *
jpayne@69:  * The result may be longer or shorter than the original.
jpayne@69:  * The source string and the destination buffer must not overlap.
jpayne@69:  *
jpayne@69:  * @param csm       UCaseMap service object. This pointer is non-const!
jpayne@69:  *                  See the note above for details.
jpayne@69:  * @param dest      A buffer for the result string. The result will be NUL-terminated if
jpayne@69:  *                  the buffer is large enough.
jpayne@69:  *                  The contents is undefined in case of failure.
jpayne@69:  * @param destCapacity The size of the buffer (number of UChars). If it is 0, then
jpayne@69:  *                  dest may be NULL and the function will only return the length of the result
jpayne@69:  *                  without writing any of the result string.
jpayne@69:  * @param src       The original string.
jpayne@69:  * @param srcLength The length of the original string. If -1, then src must be NUL-terminated.
jpayne@69:  * @param pErrorCode Must be a valid pointer to an error code value,
jpayne@69:  *                  which must not indicate a failure before the function call.
jpayne@69:  * @return The length of the result string, if successful - or in case of a buffer overflow,
jpayne@69:  *         in which case it will be greater than destCapacity.
jpayne@69:  *
jpayne@69:  * @see u_strToTitle
jpayne@69:  * @stable ICU 3.8
jpayne@69:  */
jpayne@69: U_STABLE int32_t U_EXPORT2
jpayne@69: ucasemap_toTitle(UCaseMap *csm,
jpayne@69:                  UChar *dest, int32_t destCapacity,
jpayne@69:                  const UChar *src, int32_t srcLength,
jpayne@69:                  UErrorCode *pErrorCode);
jpayne@69: 
jpayne@69: #endif  // UCONFIG_NO_BREAK_ITERATION
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Lowercase the characters in a UTF-8 string.
jpayne@69:  * Casing is locale-dependent and context-sensitive.
jpayne@69:  * The result may be longer or shorter than the original.
jpayne@69:  * The source string and the destination buffer must not overlap.
jpayne@69:  *
jpayne@69:  * @param csm       UCaseMap service object.
jpayne@69:  * @param dest      A buffer for the result string. The result will be NUL-terminated if
jpayne@69:  *                  the buffer is large enough.
jpayne@69:  *                  The contents is undefined in case of failure.
jpayne@69:  * @param destCapacity The size of the buffer (number of bytes). If it is 0, then
jpayne@69:  *                  dest may be NULL and the function will only return the length of the result
jpayne@69:  *                  without writing any of the result string.
jpayne@69:  * @param src       The original string.
jpayne@69:  * @param srcLength The length of the original string. If -1, then src must be NUL-terminated.
jpayne@69:  * @param pErrorCode Must be a valid pointer to an error code value,
jpayne@69:  *                  which must not indicate a failure before the function call.
jpayne@69:  * @return The length of the result string, if successful - or in case of a buffer overflow,
jpayne@69:  *         in which case it will be greater than destCapacity.
jpayne@69:  *
jpayne@69:  * @see u_strToLower
jpayne@69:  * @stable ICU 3.4
jpayne@69:  */
jpayne@69: U_STABLE int32_t U_EXPORT2
jpayne@69: ucasemap_utf8ToLower(const UCaseMap *csm,
jpayne@69:                      char *dest, int32_t destCapacity,
jpayne@69:                      const char *src, int32_t srcLength,
jpayne@69:                      UErrorCode *pErrorCode);
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Uppercase the characters in a UTF-8 string.
jpayne@69:  * Casing is locale-dependent and context-sensitive.
jpayne@69:  * The result may be longer or shorter than the original.
jpayne@69:  * The source string and the destination buffer must not overlap.
jpayne@69:  *
jpayne@69:  * @param csm       UCaseMap service object.
jpayne@69:  * @param dest      A buffer for the result string. The result will be NUL-terminated if
jpayne@69:  *                  the buffer is large enough.
jpayne@69:  *                  The contents is undefined in case of failure.
jpayne@69:  * @param destCapacity The size of the buffer (number of bytes). If it is 0, then
jpayne@69:  *                  dest may be NULL and the function will only return the length of the result
jpayne@69:  *                  without writing any of the result string.
jpayne@69:  * @param src       The original string.
jpayne@69:  * @param srcLength The length of the original string. If -1, then src must be NUL-terminated.
jpayne@69:  * @param pErrorCode Must be a valid pointer to an error code value,
jpayne@69:  *                  which must not indicate a failure before the function call.
jpayne@69:  * @return The length of the result string, if successful - or in case of a buffer overflow,
jpayne@69:  *         in which case it will be greater than destCapacity.
jpayne@69:  *
jpayne@69:  * @see u_strToUpper
jpayne@69:  * @stable ICU 3.4
jpayne@69:  */
jpayne@69: U_STABLE int32_t U_EXPORT2
jpayne@69: ucasemap_utf8ToUpper(const UCaseMap *csm,
jpayne@69:                      char *dest, int32_t destCapacity,
jpayne@69:                      const char *src, int32_t srcLength,
jpayne@69:                      UErrorCode *pErrorCode);
jpayne@69: 
jpayne@69: #if !UCONFIG_NO_BREAK_ITERATION
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Titlecase a UTF-8 string.
jpayne@69:  * Casing is locale-dependent and context-sensitive.
jpayne@69:  * Titlecasing uses a break iterator to find the first characters of words
jpayne@69:  * that are to be titlecased. It titlecases those characters and lowercases
jpayne@69:  * all others. (This can be modified with ucasemap_setOptions().)
jpayne@69:  *
jpayne@69:  * Note: This function takes a non-const UCaseMap pointer because it will
jpayne@69:  * open a default break iterator if no break iterator was set yet,
jpayne@69:  * and effectively call ucasemap_setBreakIterator();
jpayne@69:  * also because the break iterator is stateful and will be modified during
jpayne@69:  * the iteration.
jpayne@69:  *
jpayne@69:  * The titlecase break iterator can be provided to customize for arbitrary
jpayne@69:  * styles, using rules and dictionaries beyond the standard iterators.
jpayne@69:  * The standard titlecase iterator for the root locale implements the
jpayne@69:  * algorithm of Unicode TR 21.
jpayne@69:  *
jpayne@69:  * This function uses only the setUText(), first(), next() and close() methods of the
jpayne@69:  * provided break iterator.
jpayne@69:  *
jpayne@69:  * The result may be longer or shorter than the original.
jpayne@69:  * The source string and the destination buffer must not overlap.
jpayne@69:  *
jpayne@69:  * @param csm       UCaseMap service object. This pointer is non-const!
jpayne@69:  *                  See the note above for details.
jpayne@69:  * @param dest      A buffer for the result string. The result will be NUL-terminated if
jpayne@69:  *                  the buffer is large enough.
jpayne@69:  *                  The contents is undefined in case of failure.
jpayne@69:  * @param destCapacity The size of the buffer (number of bytes). If it is 0, then
jpayne@69:  *                  dest may be NULL and the function will only return the length of the result
jpayne@69:  *                  without writing any of the result string.
jpayne@69:  * @param src       The original string.
jpayne@69:  * @param srcLength The length of the original string. If -1, then src must be NUL-terminated.
jpayne@69:  * @param pErrorCode Must be a valid pointer to an error code value,
jpayne@69:  *                  which must not indicate a failure before the function call.
jpayne@69:  * @return The length of the result string, if successful - or in case of a buffer overflow,
jpayne@69:  *         in which case it will be greater than destCapacity.
jpayne@69:  *
jpayne@69:  * @see u_strToTitle
jpayne@69:  * @see U_TITLECASE_NO_LOWERCASE
jpayne@69:  * @see U_TITLECASE_NO_BREAK_ADJUSTMENT
jpayne@69:  * @stable ICU 3.8
jpayne@69:  */
jpayne@69: U_STABLE int32_t U_EXPORT2
jpayne@69: ucasemap_utf8ToTitle(UCaseMap *csm,
jpayne@69:                     char *dest, int32_t destCapacity,
jpayne@69:                     const char *src, int32_t srcLength,
jpayne@69:                     UErrorCode *pErrorCode);
jpayne@69: 
jpayne@69: #endif
jpayne@69: 
jpayne@69: /**
jpayne@69:  * Case-folds the characters in a UTF-8 string.
jpayne@69:  *
jpayne@69:  * Case-folding is locale-independent and not context-sensitive,
jpayne@69:  * but there is an option for whether to include or exclude mappings for dotted I
jpayne@69:  * and dotless i that are marked with 'T' in CaseFolding.txt.
jpayne@69:  *
jpayne@69:  * The result may be longer or shorter than the original.
jpayne@69:  * The source string and the destination buffer must not overlap.
jpayne@69:  *
jpayne@69:  * @param csm       UCaseMap service object.
jpayne@69:  * @param dest      A buffer for the result string. The result will be NUL-terminated if
jpayne@69:  *                  the buffer is large enough.
jpayne@69:  *                  The contents is undefined in case of failure.
jpayne@69:  * @param destCapacity The size of the buffer (number of bytes). If it is 0, then
jpayne@69:  *                  dest may be NULL and the function will only return the length of the result
jpayne@69:  *                  without writing any of the result string.
jpayne@69:  * @param src       The original string.
jpayne@69:  * @param srcLength The length of the original string. If -1, then src must be NUL-terminated.
jpayne@69:  * @param pErrorCode Must be a valid pointer to an error code value,
jpayne@69:  *                  which must not indicate a failure before the function call.
jpayne@69:  * @return The length of the result string, if successful - or in case of a buffer overflow,
jpayne@69:  *         in which case it will be greater than destCapacity.
jpayne@69:  *
jpayne@69:  * @see u_strFoldCase
jpayne@69:  * @see ucasemap_setOptions
jpayne@69:  * @see U_FOLD_CASE_DEFAULT
jpayne@69:  * @see U_FOLD_CASE_EXCLUDE_SPECIAL_I
jpayne@69:  * @stable ICU 3.8
jpayne@69:  */
jpayne@69: U_STABLE int32_t U_EXPORT2
jpayne@69: ucasemap_utf8FoldCase(const UCaseMap *csm,
jpayne@69:                       char *dest, int32_t destCapacity,
jpayne@69:                       const char *src, int32_t srcLength,
jpayne@69:                       UErrorCode *pErrorCode);
jpayne@69: 
jpayne@69: #endif