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
18U_NAMESPACE_BEGIN
19
20/**
21 * \file
22 * \brief C++ API: FilteredBreakIteratorBuilder
23 */
24
25/**
26 * The BreakIteratorFilter is used to modify the behavior of a BreakIterator
27 *  by constructing a new BreakIterator which suppresses certain segment boundaries.
28 *  See  http://www.unicode.org/reports/tr35/tr35-general.html#Segmentation_Exceptions .
29 *  For example, a typical English Sentence Break Iterator would break on the space
30 *  in the string "Mr. Smith" (resulting in two segments),
31 *  but with "Mr." as an exception, a filtered break iterator
32 *  would consider the string "Mr. Smith" to be a single segment.
33 *
34 * @stable ICU 56
35 */
36class U_COMMON_API FilteredBreakIteratorBuilder : public UObject {
37 public:
38  /**
39   *  destructor.
40   * @stable ICU 56
41   */
42  virtual ~FilteredBreakIteratorBuilder();
43
44  /**
45   * Construct a FilteredBreakIteratorBuilder based on rules in a locale.
46   * The rules are taken from CLDR exception data for the locale,
47   *  see http://www.unicode.org/reports/tr35/tr35-general.html#Segmentation_Exceptions
48   *  This is the equivalent of calling createInstance(UErrorCode&)
49   *    and then repeatedly calling addNoBreakAfter(...) with the contents
50   *    of the CLDR exception data.
51   * @param where the locale.
52   * @param status The error code.
53   * @return the new builder
54   * @stable ICU 56
55   */
56  static FilteredBreakIteratorBuilder *createInstance(const Locale& where, UErrorCode& status);
57
58#ifndef U_HIDE_DEPRECATED_API
59  /**
60   * This function has been deprecated in favor of createEmptyInstance, which has
61   * identical behavior.
62   * @param status The error code.
63   * @return the new builder
64   * @deprecated ICU 60 use createEmptyInstance instead
65   * @see createEmptyInstance()
66   */
67  static inline FilteredBreakIteratorBuilder *createInstance(UErrorCode &status) {
68    return createEmptyInstance(status);
69  }
70#endif  /* U_HIDE_DEPRECATED_API */
71
72#ifndef U_HIDE_DRAFT_API
73  /**
74   * Construct an empty FilteredBreakIteratorBuilder.
75   * In this state, it will not suppress any segment boundaries.
76   * @param status The error code.
77   * @return the new builder
78   * @draft ICU 60
79   */
80  static FilteredBreakIteratorBuilder *createEmptyInstance(UErrorCode &status);
81#endif  /* U_HIDE_DRAFT_API */
82
83  /**
84   * Suppress a certain string from being the end of a segment.
85   * For example, suppressing "Mr.", then segments ending in "Mr." will not be returned
86   * by the iterator.
87   * @param string the string to suppress, such as "Mr."
88   * @param status error code
89   * @return returns TRUE if the string was not present and now added,
90   * FALSE if the call was a no-op because the string was already being suppressed.
91   * @stable ICU 56
92   */
93  virtual UBool suppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0;
94
95  /**
96   * Stop suppressing a certain string from being the end of the segment.
97   * This function does not create any new segment boundaries, but only serves to un-do
98   * the effect of earlier calls to suppressBreakAfter, or to un-do the effect of
99   * locale data which may be suppressing certain strings.
100   * @param exception the exception to remove
101   * @param status error code
102   * @return returns TRUE if the string was present and now removed,
103   * FALSE if the call was a no-op because the string was not being suppressed.
104   * @stable ICU 56
105   */
106  virtual UBool unsuppressBreakAfter(const UnicodeString& string, UErrorCode& status) = 0;
107
108#ifndef U_HIDE_DEPRECATED_API
109  /**
110   * This function has been deprecated in favor of wrapIteratorWithFilter()
111   * The behavior is identical.
112   * @param adoptBreakIterator the break iterator to adopt
113   * @param status error code
114   * @return the new BreakIterator, owned by the caller.
115   * @deprecated ICU 60 use wrapIteratorWithFilter() instead
116   * @see wrapBreakIteratorWithFilter()
117   */
118  virtual BreakIterator *build(BreakIterator* adoptBreakIterator, UErrorCode& status) = 0;
119#endif  /* U_HIDE_DEPRECATED_API */
120
121#ifndef U_HIDE_DRAFT_API
122  /**
123   * Wrap (adopt) an existing break iterator in a new filtered instance.
124   * The resulting BreakIterator is owned by the caller.
125   * The BreakIteratorFilter may be destroyed before the BreakIterator is destroyed.
126   * Note that the adoptBreakIterator is adopted by the new BreakIterator
127   * and should no longer be used by the caller.
128   * The FilteredBreakIteratorBuilder may be reused.
129   * This function is an alias for build()
130   * @param adoptBreakIterator the break iterator to adopt
131   * @param status error code
132   * @return the new BreakIterator, owned by the caller.
133   * @draft ICU 60
134   */
135  inline BreakIterator *wrapIteratorWithFilter(BreakIterator* adoptBreakIterator, UErrorCode& status) {
136    return build(adoptBreakIterator, status);
137  }
138#endif  /* U_HIDE_DRAFT_API */
139
140 protected:
141  /**
142   * For subclass use
143   * @stable ICU 56
144   */
145  FilteredBreakIteratorBuilder();
146};
147
148
149U_NAMESPACE_END
150
151#endif // #if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION
152
153#endif // #ifndef FILTEREDBRK_H
154