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 #ifndef U_HIDE_DEPRECATED_API
  60   /**
  61    * This function has been deprecated in favor of createEmptyInstance, which has
  62    * identical behavior.
  63    * @param status The error code.
  64    * @return the new builder
  65    * @deprecated ICU 60 use createEmptyInstance instead
  66    * @see createEmptyInstance()
  67    */
  68   static FilteredBreakIteratorBuilder *createInstance(UErrorCode &status);
  69 #endif  /* U_HIDE_DEPRECATED_API */
  70
  71   /**
  72    * Construct an empty FilteredBreakIteratorBuilder.
  73    * In this state, it will not suppress any segment boundaries.
  74    * @param status The error code.
  75    * @return the new builder
  76    * @stable ICU 60
  77    */
  78   static FilteredBreakIteratorBuilder *createEmptyInstance(UErrorCode &status);
  79
  80   /**
  81    * Suppress a certain string from being the end of a segment.
  82    * For example, suppressing "Mr.", then segments ending in "Mr." will not be returned
  83    * by the iterator.
  84    * @param string the string to suppress, such as "Mr."
  85    * @param status error code
  86    * @return returns TRUE if the string was not present and now added,
  87    * FALSE if the call was a no-op because the string was already being suppressed.
  88    * @stable ICU 56
  89    */
  90   virtual UBool suppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0;
  91
  92   /**
  93    * Stop suppressing a certain string from being the end of the segment.
  94    * This function does not create any new segment boundaries, but only serves to un-do
  95    * the effect of earlier calls to suppressBreakAfter, or to un-do the effect of
  96    * locale data which may be suppressing certain strings.
  97    * @param string the exception to remove
  98    * @param status error code
  99    * @return returns TRUE if the string was present and now removed,
 100    * FALSE if the call was a no-op because the string was not being suppressed.
 101    * @stable ICU 56
 102    */
 103   virtual UBool unsuppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0;
 104
 105   /**
 106    * This function has been deprecated in favor of wrapIteratorWithFilter()
 107    * The behavior is identical.
 108    * @param adoptBreakIterator the break iterator to adopt
 109    * @param status error code
 110    * @return the new BreakIterator, owned by the caller.
 111    * @deprecated ICU 60 use wrapIteratorWithFilter() instead
 112    * @see wrapBreakIteratorWithFilter()
 113    */
 114   virtual BreakIterator *build(BreakIterator* adoptBreakIterator, UErrorCode& status) = 0;
 115
 116   /**
 117    * Wrap (adopt) an existing break iterator in a new filtered instance.
 118    * The resulting BreakIterator is owned by the caller.
 119    * The BreakIteratorFilter may be destroyed before the BreakIterator is destroyed.
 120    * Note that the adoptBreakIterator is adopted by the new BreakIterator
 121    * and should no longer be used by the caller.
 122    * The FilteredBreakIteratorBuilder may be reused.
 123    * This function is an alias for build()
 124    * @param adoptBreakIterator the break iterator to adopt
 125    * @param status error code
 126    * @return the new BreakIterator, owned by the caller.
 127    * @stable ICU 60
 128    */
 129   inline BreakIterator *wrapIteratorWithFilter(BreakIterator* adoptBreakIterator, UErrorCode& status) {
 130     return build(adoptBreakIterator, status);
 131   }
 132
 133  protected:
 134   /**
 135    * For subclass use
 136    * @stable ICU 56
 137    */
 138   FilteredBreakIteratorBuilder();
 139 };
 140
 141
 142 U_NAMESPACE_END
 143 #endif // U_SHOW_CPLUSPLUS_API
 144
 145 #endif // #if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION
 146
 147 #endif // #ifndef FILTEREDBRK_H