Back to home page

EIC code displayed by LXR

 
 

    


Warning, file /include/unicode/filteredbrk.h was not indexed or was modified since last indexation (in which case cross-reference links may be missing, inaccurate or erroneous).

0001 // © 2016 and later: Unicode, Inc. and others.
0002 // License & terms of use: http://www.unicode.org/copyright.html
0003 /*
0004 ********************************************************************************
0005 *   Copyright (C) 1997-2015, International Business Machines
0006 *   Corporation and others.  All Rights Reserved.
0007 ********************************************************************************
0008 */
0009 
0010 #ifndef FILTEREDBRK_H
0011 #define FILTEREDBRK_H
0012 
0013 #include "unicode/utypes.h"
0014 
0015 #if U_SHOW_CPLUSPLUS_API
0016 
0017 #include "unicode/brkiter.h"
0018 
0019 #if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION
0020 
0021 U_NAMESPACE_BEGIN
0022 
0023 /**
0024  * \file
0025  * \brief C++ API: FilteredBreakIteratorBuilder
0026  */
0027 
0028 /**
0029  * The BreakIteratorFilter is used to modify the behavior of a BreakIterator
0030  *  by constructing a new BreakIterator which suppresses certain segment boundaries.
0031  *  See  http://www.unicode.org/reports/tr35/tr35-general.html#Segmentation_Exceptions .
0032  *  For example, a typical English Sentence Break Iterator would break on the space
0033  *  in the string "Mr. Smith" (resulting in two segments),
0034  *  but with "Mr." as an exception, a filtered break iterator
0035  *  would consider the string "Mr. Smith" to be a single segment.
0036  *
0037  * @stable ICU 56
0038  */
0039 class U_COMMON_API FilteredBreakIteratorBuilder : public UObject {
0040  public:
0041   /**
0042    *  destructor.
0043    * @stable ICU 56
0044    */
0045   virtual ~FilteredBreakIteratorBuilder();
0046 
0047   /**
0048    * Construct a FilteredBreakIteratorBuilder based on rules in a locale.
0049    * The rules are taken from CLDR exception data for the locale,
0050    *  see http://www.unicode.org/reports/tr35/tr35-general.html#Segmentation_Exceptions
0051    *  This is the equivalent of calling createInstance(UErrorCode&)
0052    *    and then repeatedly calling addNoBreakAfter(...) with the contents
0053    *    of the CLDR exception data.
0054    * @param where the locale.
0055    * @param status The error code.
0056    * @return the new builder
0057    * @stable ICU 56
0058    */
0059   static FilteredBreakIteratorBuilder *createInstance(const Locale& where, UErrorCode& status);
0060 
0061 #ifndef U_HIDE_DEPRECATED_API
0062   /**
0063    * This function has been deprecated in favor of createEmptyInstance, which has
0064    * identical behavior.
0065    * @param status The error code.
0066    * @return the new builder
0067    * @deprecated ICU 60 use createEmptyInstance instead
0068    * @see createEmptyInstance()
0069    */
0070   static FilteredBreakIteratorBuilder *createInstance(UErrorCode &status);
0071 #endif  /* U_HIDE_DEPRECATED_API */
0072 
0073   /**
0074    * Construct an empty FilteredBreakIteratorBuilder.
0075    * In this state, it will not suppress any segment boundaries.
0076    * @param status The error code.
0077    * @return the new builder
0078    * @stable ICU 60
0079    */
0080   static FilteredBreakIteratorBuilder *createEmptyInstance(UErrorCode &status);
0081 
0082   /**
0083    * Suppress a certain string from being the end of a segment.
0084    * For example, suppressing "Mr.", then segments ending in "Mr." will not be returned
0085    * by the iterator.
0086    * @param string the string to suppress, such as "Mr."
0087    * @param status error code
0088    * @return returns true if the string was not present and now added,
0089    * false if the call was a no-op because the string was already being suppressed.
0090    * @stable ICU 56
0091    */
0092   virtual UBool suppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0;
0093 
0094   /**
0095    * Stop suppressing a certain string from being the end of the segment.
0096    * This function does not create any new segment boundaries, but only serves to un-do
0097    * the effect of earlier calls to suppressBreakAfter, or to un-do the effect of
0098    * locale data which may be suppressing certain strings.
0099    * @param string the exception to remove
0100    * @param status error code
0101    * @return returns true if the string was present and now removed,
0102    * false if the call was a no-op because the string was not being suppressed.
0103    * @stable ICU 56
0104    */
0105   virtual UBool unsuppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0;
0106 
0107 #ifndef U_FORCE_HIDE_DEPRECATED_API
0108   /**
0109    * This function has been deprecated in favor of wrapIteratorWithFilter()
0110    * The behavior is identical.
0111    * @param adoptBreakIterator the break iterator to adopt
0112    * @param status error code
0113    * @return the new BreakIterator, owned by the caller.
0114    * @deprecated ICU 60 use wrapIteratorWithFilter() instead
0115    * @see wrapBreakIteratorWithFilter()
0116    */
0117   virtual BreakIterator *build(BreakIterator* adoptBreakIterator, UErrorCode& status) = 0;
0118 #endif  // U_FORCE_HIDE_DEPRECATED_API
0119 
0120   /**
0121    * Wrap (adopt) an existing break iterator in a new filtered instance.
0122    * The resulting BreakIterator is owned by the caller.
0123    * The BreakIteratorFilter may be destroyed before the BreakIterator is destroyed.
0124    * Note that the adoptBreakIterator is adopted by the new BreakIterator
0125    * and should no longer be used by the caller.
0126    * The FilteredBreakIteratorBuilder may be reused.
0127    * This function is an alias for build()
0128    * @param adoptBreakIterator the break iterator to adopt
0129    * @param status error code
0130    * @return the new BreakIterator, owned by the caller.
0131    * @stable ICU 60
0132    */
0133   inline BreakIterator *wrapIteratorWithFilter(BreakIterator* adoptBreakIterator, UErrorCode& status) {
0134     return build(adoptBreakIterator, status);
0135   }
0136 
0137  protected:
0138   /**
0139    * For subclass use
0140    * @stable ICU 56
0141    */
0142   FilteredBreakIteratorBuilder();
0143 };
0144 
0145 
0146 U_NAMESPACE_END
0147 
0148 #endif // #if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION
0149 
0150 #endif /* U_SHOW_CPLUSPLUS_API */
0151 
0152 #endif // #ifndef FILTEREDBRK_H