icuSources/common/unicode/filteredbrk.h

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 ********************************************************************************
   5 *   Copyright (C) 1997-2015, International Business Machines
   6 *   Corporation and others.  All Rights Reserved.
   7 ********************************************************************************
   8 */
   9
  10 #ifndef FILTEREDBRK_H
  11 #define FILTEREDBRK_H
  12
  13 #include "unicode/utypes.h"
  14 #include "unicode/brkiter.h"
  15
  16 #if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION
  17
  18 #if U_SHOW_CPLUSPLUS_API
  19 U_NAMESPACE_BEGIN
  20
  21 /**
  22  * \file
  23  * \brief C++ API: FilteredBreakIteratorBuilder
  24  */
  25
  26 /**
  27  * The BreakIteratorFilter is used to modify the behavior of a BreakIterator
  28  *  by constructing a new BreakIterator which suppresses certain segment boundaries.
  29  *  See  http://www.unicode.org/reports/tr35/tr35-general.html#Segmentation_Exceptions .
  30  *  For example, a typical English Sentence Break Iterator would break on the space
  31  *  in the string "Mr. Smith" (resulting in two segments),
  32  *  but with "Mr." as an exception, a filtered break iterator
  33  *  would consider the string "Mr. Smith" to be a single segment.
  34  *
  35  * @stable ICU 56
  36  */
  37 class U_COMMON_API FilteredBreakIteratorBuilder : public UObject {
  38  public:
  39   /**
  40    *  destructor.
  41    * @stable ICU 56
  42    */
  43   virtual ~FilteredBreakIteratorBuilder();
  44
  45   /**
  46    * Construct a FilteredBreakIteratorBuilder based on rules in a locale.
  47    * The rules are taken from CLDR exception data for the locale,
  48    *  see http://www.unicode.org/reports/tr35/tr35-general.html#Segmentation_Exceptions
  49    *  This is the equivalent of calling createInstance(UErrorCode&)
  50    *    and then repeatedly calling addNoBreakAfter(...) with the contents
  51    *    of the CLDR exception data.
  52    * @param where the locale.
  53    * @param status The error code.
  54    * @return the new builder
  55    * @stable ICU 56
  56    */
  57   static FilteredBreakIteratorBuilder *createInstance(const Locale& where, UErrorCode& status);
  58
  59   /**
  60    * Construct an empty FilteredBreakIteratorBuilder.
  61    * In this state, it will not suppress any segment boundaries.
  62    * @param status The error code.
  63    * @return the new builder
  64    * @stable ICU 56
  65    */
  66   static FilteredBreakIteratorBuilder *createInstance(UErrorCode &status);
  67
  68   /**
  69    * Suppress a certain string from being the end of a segment.
  70    * For example, suppressing "Mr.", then segments ending in "Mr." will not be returned
  71    * by the iterator.
  72    * @param string the string to suppress, such as "Mr."
  73    * @param status error code
  74    * @return returns TRUE if the string was not present and now added,
  75    * FALSE if the call was a no-op because the string was already being suppressed.
  76    * @stable ICU 56
  77    */
  78   virtual UBool suppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0;
  79
  80   /**
  81    * Stop suppressing a certain string from being the end of the segment.
  82    * This function does not create any new segment boundaries, but only serves to un-do
  83    * the effect of earlier calls to suppressBreakAfter, or to un-do the effect of
  84    * locale data which may be suppressing certain strings.
  85    * @param exception the exception to remove
  86    * @param status error code
  87    * @return returns TRUE if the string was present and now removed,
  88    * FALSE if the call was a no-op because the string was not being suppressed.
  89    * @stable ICU 56
  90    */
  91   virtual UBool unsuppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0;
  92
  93   /**
  94    * Wrap (adopt) an existing break iterator in a new filtered instance.
  95    * The resulting BreakIterator is owned by the caller.
  96    * The BreakIteratorFilter may be destroyed before the BreakIterator is destroyed.
  97    * Note that the adoptBreakIterator is adopted by the new BreakIterator
  98    * and should no longer be used by the caller.
  99    * The FilteredBreakIteratorBuilder may be reused.
 100    * @param adoptBreakIterator the break iterator to adopt
 101    * @param status error code
 102    * @return the new BreakIterator, owned by the caller.
 103    * @stable ICU 56
 104    */
 105   virtual BreakIterator *build(BreakIterator* adoptBreakIterator, UErrorCode& status) = 0;
 106
 107  protected:
 108   /**
 109    * For subclass use
 110    * @stable ICU 56
 111    */
 112   FilteredBreakIteratorBuilder();
 113 };
 114
 115
 116 U_NAMESPACE_END
 117 #endif // U_SHOW_CPLUSPLUS_API
 118
 119 #endif // #if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION
 120
 121 #endif // #ifndef FILTEREDBRK_H