icuSources/common/unicode/filteredbrk.h

   1 /*
   2 ********************************************************************************
   3 *   Copyright (C) 1997-2015, 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/utypes.h"
  12 #include "unicode/brkiter.h"
  13
  14 #if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION
  15 #ifndef U_HIDE_INTERNAL_API
  16
  17 U_NAMESPACE_BEGIN
  18
  19 /**
  20  * \file
  21  * \brief C++ API: FilteredBreakIteratorBuilder
  22  */
  23
  24 /**
  25  * The BreakIteratorFilter is used to modify the behavior of a BreakIterator
  26  *  by constructing a new BreakIterator which suppresses certain segment boundaries.
  27  *  See  http://www.unicode.org/reports/tr35/tr35-general.html#Segmentation_Exceptions .
  28  *  For example, a typical English Sentence Break Iterator would break on the space
  29  *  in the string "Mr. Smith" (resulting in two segments),
  30  *  but with "Mr." as an exception, a filtered break iterator
  31  *  would consider the string "Mr. Smith" to be a single segment.
  32  *
  33  * <p><b>Note:</b> An instance of {@link BreakIterator} returned by this builder
  34  * class currently does not support following operations in this technology preview
  35  * version:
  36  * <ul>
  37  *   <li>{@link BreakIterator#next(int32_t) next(int32_t n)}</li>
  38  *   <li>{@link BreakIterator#previous(void) previous(void)}</li>
  39  *   <li>{@link BreakIterator#following(int32_t) following(int32_t offset)}</li>
  40  *   <li>{@link BreakIterator#preceding(int32_t) preceding(int32_t offset)}</li>
  41  * </ul>
  42  * When one of above methods is called, the corresponding method of the adopted
  43  * BreakIterator will be invoked (i.e. no segment suppressions will be used).
  44  * Note: This fallback behavior undoes r36410 in which these methods were changed
  45  * to just return BreakIterator.DONE immediately without updating the internal state.
  46  *
  47  * @internal technology preview
  48  */
  49 class U_COMMON_API FilteredBreakIteratorBuilder : public UObject {
  50  public:
  51   /**
  52    *  destructor.
  53    * @internal technology preview
  54    */
  55   virtual ~FilteredBreakIteratorBuilder();
  56
  57   /**
  58    * Construct a FilteredBreakIteratorBuilder based on rules in a locale.
  59    * The rules are taken from CLDR exception data for the locale,
  60    *  see http://www.unicode.org/reports/tr35/tr35-general.html#Segmentation_Exceptions
  61    *  This is the equivalent of calling createInstance(UErrorCode&)
  62    *    and then repeatedly calling addNoBreakAfter(...) with the contents
  63    *    of the CLDR exception data.
  64    * @param where the locale.
  65    * @param status The error code.
  66    * @return the new builder
  67    * @internal technology preview
  68    */
  69   static FilteredBreakIteratorBuilder *createInstance(const Locale& where, UErrorCode& status);
  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    * @internal technology preview
  77    */
  78   static FilteredBreakIteratorBuilder *createInstance(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    * @internal technology preview
  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 exception 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    * @internal technology preview
 102    */
 103   virtual UBool unsuppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0;
 104
 105   /**
 106    * Wrap (adopt) an existing break iterator in a new filtered instance.
 107    * The resulting BreakIterator is owned by the caller.
 108    * The BreakIteratorFilter may be destroyed before the BreakIterator is destroyed.
 109    * Note that the adoptBreakIterator is adopted by the new BreakIterator
 110    * and should no longer be used by the caller.
 111    * The FilteredBreakIteratorBuilder may be reused.
 112    * @param adoptBreakIterator the break iterator to adopt
 113    * @param status error code
 114    * @return the new BreakIterator, owned by the caller.
 115    * @internal technology preview
 116    */
 117   virtual BreakIterator *build(BreakIterator* adoptBreakIterator, UErrorCode& status) = 0;
 118
 119  protected:
 120   /**
 121    * For subclass use
 122    * @internal technology preview
 123    */
 124   FilteredBreakIteratorBuilder();
 125 };
 126
 127
 128 U_NAMESPACE_END
 129
 130 #endif  /* U_HIDE_INTERNAL_API */
 131 #endif // #if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION
 132
 133 #endif // #ifndef FILTEREDBRK_H