1 /* 2 ******************************************************************************** 3 * Copyright (C) 1997-2014, International Business Machines 4 * Corporation and others. All Rights Reserved. 5 ******************************************************************************** 6 */ 7 8 #ifndef FILTEREDBRK_H 9 #define FILTEREDBRK_H 10 11 #include "unicode/brkiter.h" 12 13 #if !UCONFIG_NO_BREAK_ITERATION && U_HAVE_STD_STRING && !UCONFIG_NO_FILTERED_BREAK_ITERATION 14 15 U_NAMESPACE_BEGIN 16 17 /** 18 * \file 19 * \brief C++ API: FilteredBreakIteratorBuilder 20 */ 21 22 /** 23 * The BreakIteratorFilter is used to modify the behavior of a BreakIterator 24 * by constructing a new BreakIterator which suppresses certain segment boundaries. 25 * See http://www.unicode.org/reports/tr35/tr35-general.html#Segmentation_Exceptions . 26 * For example, a typical English Sentence Break Iterator would break on the space 27 * in the string "Mr. Smith" (resulting in two segments), 28 * but with "Mr." as an exception, a filtered break iterator 29 * would consider the string "Mr. Smith" to be a single segment. 30 * 31 * @internal technology preview 32 */ 33 class U_I18N_API FilteredBreakIteratorBuilder : public UObject { 34 public: 35 /** 36 * destructor. 37 * @internal technology preview 38 */ 39 virtual ~FilteredBreakIteratorBuilder(); 40 41 /** 42 * Construct a FilteredBreakIteratorBuilder based on rules in a locale. 43 * The rules are taken from CLDR exception data for the locale, 44 * see http://www.unicode.org/reports/tr35/tr35-general.html#Segmentation_Exceptions 45 * This is the equivalent of calling createInstance(UErrorCode&) 46 * and then repeatedly calling addNoBreakAfter(...) with the contents 47 * of the CLDR exception data. 48 * @param where the locale. 49 * @param status The error code. 50 * @return the new builder 51 * @internal technology preview 52 */ 53 static FilteredBreakIteratorBuilder *createInstance(const Locale& where, UErrorCode& status); 54 55 /** 56 * Construct an empty FilteredBreakIteratorBuilder. 57 * In this state, it will not suppress any segment boundaries. 58 * @param status The error code. 59 * @return the new builder 60 * @internal technology preview 61 */ 62 static FilteredBreakIteratorBuilder *createInstance(UErrorCode &status); 63 64 /** 65 * Suppress a certain string from being the end of a segment. 66 * For example, suppressing "Mr.", then segments ending in "Mr." will not be returned 67 * by the iterator. 68 * @param string the string to suppress, such as "Mr." 69 * @param status error code 70 * @return returns TRUE if the string was not present and now added, 71 * FALSE if the call was a no-op because the string was already being suppressed. 72 * @internal technology preview 73 */ 74 virtual UBool suppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0; 75 76 /** 77 * Stop suppressing a certain string from being the end of the segment. 78 * This function does not create any new segment boundaries, but only serves to un-do 79 * the effect of earlier calls to suppressBreakAfter, or to un-do the effect of 80 * locale data which may be suppressing certain strings. 81 * @param exception the exception to remove 82 * @param status error code 83 * @return returns TRUE if the string was present and now removed, 84 * FALSE if the call was a no-op because the string was not being suppressed. 85 * @internal technology preview 86 */ 87 virtual UBool unsuppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0; 88 89 /** 90 * Wrap (adopt) an existing break iterator in a new filtered instance. 91 * The resulting BreakIterator is owned by the caller. 92 * The BreakIteratorFilter may be destroyed before the BreakIterator is destroyed. 93 * Note that the adoptBreakIterator is adopted by the new BreakIterator 94 * and should no longer be used by the caller. 95 * The FilteredBreakIteratorBuilder may be reused. 96 * @param adoptBreakIterator the break iterator to adopt 97 * @param status error code 98 * @return the new BreakIterator, owned by the caller. 99 * @internal technology preview 100 */ 101 virtual BreakIterator *build(BreakIterator* adoptBreakIterator, UErrorCode& status) = 0; 102 103 protected: 104 /** 105 * For subclass use 106 * @internal technology preview 107 */ 108 FilteredBreakIteratorBuilder(); 109 }; 110 111 112 U_NAMESPACE_END 113 114 #endif // #if !UCONFIG_NO_BREAK_ITERATION && U_HAVE_STD_STRING && !UCONFIG_NO_FILTERED_BREAK_ITERATION 115 116 #endif // #ifndef FILTEREDBRK_H 117