1 // Copyright (C) 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 **********************************************************************
5 * Copyright (C) 2008-2015, International Business Machines
6 * Corporation and others. All Rights Reserved.
7 **********************************************************************
8 * Date Name Description
9 * 05/11/2008 Andy Heninger Port from Java
10 **********************************************************************
11 */
12
13 #include "unicode/utypes.h"
14
15 #if !UCONFIG_NO_TRANSLITERATION && !UCONFIG_NO_BREAK_ITERATION
16
17 #include "unicode/brkiter.h"
18 #include "unicode/localpointer.h"
19 #include "unicode/uchar.h"
20 #include "unicode/unifilt.h"
21 #include "unicode/uniset.h"
22
23 #include "brktrans.h"
24 #include "cmemory.h"
25 #include "mutex.h"
26 #include "uprops.h"
27 #include "uinvchar.h"
28 #include "util.h"
29 #include "uvectr32.h"
30
31 U_NAMESPACE_BEGIN
32
33 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(BreakTransliterator)
34
35 static const UChar SPACE = 32; // ' '
36
37
38 /**
39 * Constructs a transliterator with the default delimiters '{' and
40 * '}'.
41 */
BreakTransliterator(UnicodeFilter * adoptedFilter)42 BreakTransliterator::BreakTransliterator(UnicodeFilter* adoptedFilter) :
43 Transliterator(UNICODE_STRING("Any-BreakInternal", 17), adoptedFilter),
44 cachedBI(NULL), cachedBoundaries(NULL), fInsertion(SPACE) {
45 }
46
47
48 /**
49 * Destructor.
50 */
~BreakTransliterator()51 BreakTransliterator::~BreakTransliterator() {
52 }
53
54 /**
55 * Copy constructor.
56 */
BreakTransliterator(const BreakTransliterator & o)57 BreakTransliterator::BreakTransliterator(const BreakTransliterator& o) :
58 Transliterator(o), cachedBI(NULL), cachedBoundaries(NULL), fInsertion(o.fInsertion) {
59 }
60
61
62 /**
63 * Transliterator API.
64 */
clone(void) const65 Transliterator* BreakTransliterator::clone(void) const {
66 return new BreakTransliterator(*this);
67 }
68
69 /**
70 * Implements {@link Transliterator#handleTransliterate}.
71 */
handleTransliterate(Replaceable & text,UTransPosition & offsets,UBool isIncremental) const72 void BreakTransliterator::handleTransliterate(Replaceable& text, UTransPosition& offsets,
73 UBool isIncremental ) const {
74
75 UErrorCode status = U_ZERO_ERROR;
76 LocalPointer<BreakIterator> bi;
77 LocalPointer<UVector32> boundaries;
78
79 {
80 Mutex m;
81 BreakTransliterator *nonConstThis = const_cast<BreakTransliterator *>(this);
82 boundaries.moveFrom(nonConstThis->cachedBoundaries);
83 bi.moveFrom(nonConstThis->cachedBI);
84 }
85 if (bi.isNull()) {
86 bi.adoptInstead(BreakIterator::createWordInstance(Locale::getEnglish(), status));
87 }
88 if (boundaries.isNull()) {
89 boundaries.adoptInstead(new UVector32(status));
90 }
91
92 if (bi.isNull() || boundaries.isNull() || U_FAILURE(status)) {
93 return;
94 }
95
96 boundaries->removeAllElements();
97 UnicodeString sText = replaceableAsString(text);
98 bi->setText(sText);
99 bi->preceding(offsets.start);
100
101 // To make things much easier, we will stack the boundaries, and then insert at the end.
102 // generally, we won't need too many, since we will be filtered.
103
104 int32_t boundary;
105 for(boundary = bi->next(); boundary != UBRK_DONE && boundary < offsets.limit; boundary = bi->next()) {
106 if (boundary == 0) continue;
107 // HACK: Check to see that preceeding item was a letter
108
109 UChar32 cp = sText.char32At(boundary-1);
110 int type = u_charType(cp);
111 //System.out.println(Integer.toString(cp,16) + " (before): " + type);
112 if ((U_MASK(type) & (U_GC_L_MASK | U_GC_M_MASK)) == 0) continue;
113
114 cp = sText.char32At(boundary);
115 type = u_charType(cp);
116 //System.out.println(Integer.toString(cp,16) + " (after): " + type);
117 if ((U_MASK(type) & (U_GC_L_MASK | U_GC_M_MASK)) == 0) continue;
118
119 boundaries->addElement(boundary, status);
120 // printf("Boundary at %d\n", boundary);
121 }
122
123 int delta = 0;
124 int lastBoundary = 0;
125
126 if (boundaries->size() != 0) { // if we found something, adjust
127 delta = boundaries->size() * fInsertion.length();
128 lastBoundary = boundaries->lastElementi();
129
130 // we do this from the end backwards, so that we don't have to keep updating.
131
132 while (boundaries->size() > 0) {
133 boundary = boundaries->popi();
134 text.handleReplaceBetween(boundary, boundary, fInsertion);
135 }
136 }
137
138 // Now fix up the return values
139 offsets.contextLimit += delta;
140 offsets.limit += delta;
141 offsets.start = isIncremental ? lastBoundary + delta : offsets.limit;
142
143 // Return break iterator & boundaries vector to the cache.
144 {
145 Mutex m;
146 BreakTransliterator *nonConstThis = const_cast<BreakTransliterator *>(this);
147 if (nonConstThis->cachedBI.isNull()) {
148 nonConstThis->cachedBI.moveFrom(bi);
149 }
150 if (nonConstThis->cachedBoundaries.isNull()) {
151 nonConstThis->cachedBoundaries.moveFrom(boundaries);
152 }
153 }
154
155 // TODO: do something with U_FAILURE(status);
156 // (need to look at transliterators overall, not just here.)
157 }
158
159 //
160 // getInsertion()
161 //
getInsertion() const162 const UnicodeString &BreakTransliterator::getInsertion() const {
163 return fInsertion;
164 }
165
166 //
167 // setInsertion()
168 //
setInsertion(const UnicodeString & insertion)169 void BreakTransliterator::setInsertion(const UnicodeString &insertion) {
170 this->fInsertion = insertion;
171 }
172
173 //
174 // replaceableAsString Hack to let break iterators work
175 // on the replaceable text from transliterators.
176 // In practice, the only real Replaceable type that we
177 // will be seeing is UnicodeString, so this function
178 // will normally be efficient.
179 //
replaceableAsString(Replaceable & r)180 UnicodeString BreakTransliterator::replaceableAsString(Replaceable &r) {
181 UnicodeString s;
182 UnicodeString *rs = dynamic_cast<UnicodeString *>(&r);
183 if (rs != NULL) {
184 s = *rs;
185 } else {
186 r.extractBetween(0, r.length(), s);
187 }
188 return s;
189 }
190
191 U_NAMESPACE_END
192
193 #endif /* #if !UCONFIG_NO_TRANSLITERATION */
194