1 /*
2 **********************************************************************
3 * Copyright (C) 2001-2011, International Business Machines
4 * Corporation and others. All Rights Reserved.
5 **********************************************************************
6 * Date Name Description
7 * 06/07/01 aliu Creation.
8 **********************************************************************
9 */
10
11 #include "unicode/utypes.h"
12
13 #if !UCONFIG_NO_TRANSLITERATION
14
15 #include "unicode/unifilt.h"
16 #include "unicode/uchar.h"
17 #include "unicode/uniset.h"
18 #include "unicode/utf16.h"
19 #include "cmemory.h"
20 #include "name2uni.h"
21 #include "patternprops.h"
22 #include "uprops.h"
23 #include "uinvchar.h"
24 #include "util.h"
25
26 U_NAMESPACE_BEGIN
27
28 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(NameUnicodeTransliterator)
29
30 static const UChar OPEN[] = {92,78,126,123,126,0}; // "\N~{~"
31 static const UChar OPEN_DELIM = 92; // '\\' first char of OPEN
32 static const UChar CLOSE_DELIM = 125; // '}'
33 static const UChar SPACE = 32; // ' '
34
35 U_CDECL_BEGIN
36
37 // USetAdder implementation
38 // Does not use uset.h to reduce code dependencies
39 static void U_CALLCONV
_set_add(USet * set,UChar32 c)40 _set_add(USet *set, UChar32 c) {
41 uset_add(set, c);
42 }
43
44 // These functions aren't used.
45 /*static void U_CALLCONV
46 _set_addRange(USet *set, UChar32 start, UChar32 end) {
47 ((UnicodeSet *)set)->add(start, end);
48 }
49
50 static void U_CALLCONV
51 _set_addString(USet *set, const UChar *str, int32_t length) {
52 ((UnicodeSet *)set)->add(UnicodeString((UBool)(length<0), str, length));
53 }*/
54
55 U_CDECL_END
56
57 /**
58 * Constructs a transliterator with the default delimiters '{' and
59 * '}'.
60 */
NameUnicodeTransliterator(UnicodeFilter * adoptedFilter)61 NameUnicodeTransliterator::NameUnicodeTransliterator(UnicodeFilter* adoptedFilter) :
62 Transliterator(UNICODE_STRING("Name-Any", 8), adoptedFilter) {
63
64 UnicodeSet *legalPtr = &legal;
65 // Get the legal character set
66 USetAdder sa = {
67 (USet *)legalPtr, // USet* == UnicodeSet*
68 _set_add,
69 NULL, // Don't need _set_addRange
70 NULL, // Don't need _set_addString
71 NULL, // Don't need remove()
72 NULL
73 };
74 uprv_getCharNameCharacters(&sa);
75 }
76
77 /**
78 * Destructor.
79 */
~NameUnicodeTransliterator()80 NameUnicodeTransliterator::~NameUnicodeTransliterator() {}
81
82 /**
83 * Copy constructor.
84 */
NameUnicodeTransliterator(const NameUnicodeTransliterator & o)85 NameUnicodeTransliterator::NameUnicodeTransliterator(const NameUnicodeTransliterator& o) :
86 Transliterator(o), legal(o.legal) {}
87
88 /**
89 * Assignment operator.
90 */
91 /*NameUnicodeTransliterator& NameUnicodeTransliterator::operator=(
92 const NameUnicodeTransliterator& o) {
93 Transliterator::operator=(o);
94 // not necessary: the legal sets should all be the same -- legal=o.legal;
95 return *this;
96 }*/
97
98 /**
99 * Transliterator API.
100 */
clone(void) const101 Transliterator* NameUnicodeTransliterator::clone(void) const {
102 return new NameUnicodeTransliterator(*this);
103 }
104
105 /**
106 * Implements {@link Transliterator#handleTransliterate}.
107 */
handleTransliterate(Replaceable & text,UTransPosition & offsets,UBool isIncremental) const108 void NameUnicodeTransliterator::handleTransliterate(Replaceable& text, UTransPosition& offsets,
109 UBool isIncremental) const {
110 // The failure mode, here and below, is to behave like Any-Null,
111 // if either there is no name data (max len == 0) or there is no
112 // memory (malloc() => NULL).
113
114 int32_t maxLen = uprv_getMaxCharNameLength();
115 if (maxLen == 0) {
116 offsets.start = offsets.limit;
117 return;
118 }
119
120 // Accomodate the longest possible name
121 ++maxLen; // allow for temporary trailing space
122 char* cbuf = (char*) uprv_malloc(maxLen);
123 if (cbuf == NULL) {
124 offsets.start = offsets.limit;
125 return;
126 }
127
128 UnicodeString openPat(TRUE, OPEN, -1);
129 UnicodeString str, name;
130
131 int32_t cursor = offsets.start;
132 int32_t limit = offsets.limit;
133
134 // Modes:
135 // 0 - looking for open delimiter
136 // 1 - after open delimiter
137 int32_t mode = 0;
138 int32_t openPos = -1; // open delim candidate pos
139
140 UChar32 c;
141 while (cursor < limit) {
142 c = text.char32At(cursor);
143
144 switch (mode) {
145 case 0: // looking for open delimiter
146 if (c == OPEN_DELIM) { // quick check first
147 openPos = cursor;
148 int32_t i =
149 ICU_Utility::parsePattern(openPat, text, cursor, limit);
150 if (i >= 0 && i < limit) {
151 mode = 1;
152 name.truncate(0);
153 cursor = i;
154 continue; // *** reprocess char32At(cursor)
155 }
156 }
157 break;
158
159 case 1: // after open delimiter
160 // Look for legal chars. If \s+ is found, convert it
161 // to a single space. If closeDelimiter is found, exit
162 // the loop. If any other character is found, exit the
163 // loop. If the limit is reached, exit the loop.
164
165 // Convert \s+ => SPACE. This assumes there are no
166 // runs of >1 space characters in names.
167 if (PatternProps::isWhiteSpace(c)) {
168 // Ignore leading whitespace
169 if (name.length() > 0 &&
170 name.charAt(name.length()-1) != SPACE) {
171 name.append(SPACE);
172 // If we are too long then abort. maxLen includes
173 // temporary trailing space, so use '>'.
174 if (name.length() > maxLen) {
175 mode = 0;
176 }
177 }
178 break;
179 }
180
181 if (c == CLOSE_DELIM) {
182 int32_t len = name.length();
183
184 // Delete trailing space, if any
185 if (len > 0 &&
186 name.charAt(len-1) == SPACE) {
187 --len;
188 }
189
190 if (uprv_isInvariantUString(name.getBuffer(), len)) {
191 name.extract(0, len, cbuf, maxLen, US_INV);
192
193 UErrorCode status = U_ZERO_ERROR;
194 c = u_charFromName(U_EXTENDED_CHAR_NAME, cbuf, &status);
195 if (U_SUCCESS(status)) {
196 // Lookup succeeded
197
198 // assert(U16_LENGTH(CLOSE_DELIM) == 1);
199 cursor++; // advance over CLOSE_DELIM
200
201 str.truncate(0);
202 str.append(c);
203 text.handleReplaceBetween(openPos, cursor, str);
204
205 // Adjust indices for the change in the length of
206 // the string. Do not assume that str.length() ==
207 // 1, in case of surrogates.
208 int32_t delta = cursor - openPos - str.length();
209 cursor -= delta;
210 limit -= delta;
211 // assert(cursor == openPos + str.length());
212 }
213 }
214 // If the lookup failed, we leave things as-is and
215 // still switch to mode 0 and continue.
216 mode = 0;
217 openPos = -1; // close off candidate
218 continue; // *** reprocess char32At(cursor)
219 }
220
221 // Check if c is a legal char. We assume here that
222 // legal.contains(OPEN_DELIM) is FALSE, so when we abort a
223 // name, we don't have to go back to openPos+1.
224 if (legal.contains(c)) {
225 name.append(c);
226 // If we go past the longest possible name then abort.
227 // maxLen includes temporary trailing space, so use '>='.
228 if (name.length() >= maxLen) {
229 mode = 0;
230 }
231 }
232
233 // Invalid character
234 else {
235 --cursor; // Backup and reprocess this character
236 mode = 0;
237 }
238
239 break;
240 }
241
242 cursor += U16_LENGTH(c);
243 }
244
245 offsets.contextLimit += limit - offsets.limit;
246 offsets.limit = limit;
247 // In incremental mode, only advance the cursor up to the last
248 // open delimiter candidate.
249 offsets.start = (isIncremental && openPos >= 0) ? openPos : cursor;
250
251 uprv_free(cbuf);
252 }
253
254 U_NAMESPACE_END
255
256 #endif /* #if !UCONFIG_NO_TRANSLITERATION */
257