Line data Source code
1 : /* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
2 : /*
3 : * This file is part of the LibreOffice project.
4 : *
5 : * This Source Code Form is subject to the terms of the Mozilla Public
6 : * License, v. 2.0. If a copy of the MPL was not distributed with this
7 : * file, You can obtain one at http://mozilla.org/MPL/2.0/.
8 : *
9 : * This file incorporates work covered by the following license notice:
10 : *
11 : * Licensed to the Apache Software Foundation (ASF) under one or more
12 : * contributor license agreements. See the NOTICE file distributed
13 : * with this work for additional information regarding copyright
14 : * ownership. The ASF licenses this file to you under the Apache
15 : * License, Version 2.0 (the "License"); you may not use this file
16 : * except in compliance with the License. You may obtain a copy of
17 : * the License at http://www.apache.org/licenses/LICENSE-2.0 .
18 : */
19 :
20 : #include <rtl/ustrbuf.hxx>
21 : #include <i18nutil/casefolding.hxx>
22 : #include <i18nutil/unicode.hxx>
23 :
24 : #include <comphelper/processfactory.hxx>
25 : #include <comphelper/string.hxx>
26 : #include <osl/diagnose.h>
27 :
28 : #include <string.h>
29 :
30 : #include "characterclassificationImpl.hxx"
31 : #include "breakiteratorImpl.hxx"
32 :
33 : #define TRANSLITERATION_ALL
34 : #include "transliteration_body.hxx"
35 :
36 : using namespace ::com::sun::star::uno;
37 : using namespace ::com::sun::star::lang;
38 : using namespace ::rtl;
39 :
40 : namespace com { namespace sun { namespace star { namespace i18n {
41 :
42 0 : Transliteration_body::Transliteration_body()
43 : {
44 0 : nMappingType = 0;
45 0 : transliterationName = "Transliteration_body";
46 0 : implementationName = "com.sun.star.i18n.Transliteration.Transliteration_body";
47 0 : }
48 :
49 0 : sal_Int16 SAL_CALL Transliteration_body::getType() throw(RuntimeException, std::exception)
50 : {
51 0 : return TransliterationType::ONE_TO_ONE;
52 : }
53 :
54 0 : sal_Bool SAL_CALL Transliteration_body::equals(
55 : const OUString& /*str1*/, sal_Int32 /*pos1*/, sal_Int32 /*nCount1*/, sal_Int32& /*nMatch1*/,
56 : const OUString& /*str2*/, sal_Int32 /*pos2*/, sal_Int32 /*nCount2*/, sal_Int32& /*nMatch2*/)
57 : throw(RuntimeException, std::exception)
58 : {
59 0 : throw RuntimeException();
60 : }
61 :
62 : Sequence< OUString > SAL_CALL
63 0 : Transliteration_body::transliterateRange( const OUString& str1, const OUString& str2 )
64 : throw( RuntimeException, std::exception)
65 : {
66 0 : Sequence< OUString > ostr(2);
67 0 : ostr[0] = str1;
68 0 : ostr[1] = str2;
69 0 : return ostr;
70 : }
71 :
72 0 : static sal_uInt8 lcl_getMappingTypeForToggleCase( sal_uInt8 nMappingType, sal_Unicode cChar )
73 : {
74 0 : sal_uInt8 nRes = nMappingType;
75 :
76 : // take care of TOGGLE_CASE transliteration:
77 : // nMappingType should not be a combination of flags, thuse we decide now
78 : // which one to use.
79 0 : if (nMappingType == (MappingTypeLowerToUpper | MappingTypeUpperToLower))
80 : {
81 0 : const sal_Int16 nType = unicode::getUnicodeType( cChar );
82 0 : if (nType & 0x02 /* lower case*/)
83 0 : nRes = MappingTypeLowerToUpper;
84 : else
85 : {
86 : // should also work properly for non-upper characters like white spacs, numbers, ...
87 0 : nRes = MappingTypeUpperToLower;
88 : }
89 : }
90 :
91 0 : return nRes;
92 : }
93 :
94 : OUString SAL_CALL
95 0 : Transliteration_body::transliterate(
96 : const OUString& inStr, sal_Int32 startPos, sal_Int32 nCount,
97 : Sequence< sal_Int32 >& offset)
98 : throw(RuntimeException, std::exception)
99 : {
100 0 : const sal_Unicode *in = inStr.getStr() + startPos;
101 :
102 : // Two different blocks to eliminate the if(useOffset) condition inside the
103 : // inner k loop. Yes, on massive use even such small things do count.
104 0 : if ( useOffset )
105 : {
106 0 : sal_Int32 nOffCount = 0, i;
107 0 : for (i = 0; i < nCount; i++)
108 : {
109 : // take care of TOGGLE_CASE transliteration:
110 0 : sal_uInt8 nTmpMappingType = nMappingType;
111 0 : if (nMappingType == (MappingTypeLowerToUpper | MappingTypeUpperToLower))
112 0 : nTmpMappingType = lcl_getMappingTypeForToggleCase( nMappingType, in[i] );
113 :
114 0 : const Mapping &map = casefolding::getValue( in, i, nCount, aLocale, nTmpMappingType );
115 0 : nOffCount += map.nmap;
116 : }
117 0 : rtl_uString* pStr = rtl_uString_alloc(nOffCount);
118 0 : sal_Unicode* out = pStr->buffer;
119 :
120 0 : if ( nOffCount != offset.getLength() )
121 0 : offset.realloc( nOffCount );
122 :
123 0 : sal_Int32 j = 0;
124 0 : sal_Int32 * pArr = offset.getArray();
125 0 : for (i = 0; i < nCount; i++)
126 : {
127 : // take care of TOGGLE_CASE transliteration:
128 0 : sal_uInt8 nTmpMappingType = nMappingType;
129 0 : if (nMappingType == (MappingTypeLowerToUpper | MappingTypeUpperToLower))
130 0 : nTmpMappingType = lcl_getMappingTypeForToggleCase( nMappingType, in[i] );
131 :
132 0 : const Mapping &map = casefolding::getValue( in, i, nCount, aLocale, nTmpMappingType );
133 0 : for (sal_Int32 k = 0; k < map.nmap; k++)
134 : {
135 0 : pArr[j] = i + startPos;
136 0 : out[j++] = map.map[k];
137 : }
138 : }
139 0 : out[j] = 0;
140 :
141 0 : return OUString( pStr, SAL_NO_ACQUIRE );
142 : }
143 : else
144 : {
145 : // In the simple case of no offset sequence used we can eliminate the
146 : // first getValue() loop. We could also assume that most calls result
147 : // in identical string lengths, thus using a preallocated
148 : // OUStringBuffer could be an easy way to assemble the return string
149 : // without too much hassle. However, for single characters the
150 : // OUStringBuffer::append() method is quite expensive compared to a
151 : // simple array operation, so it pays here to copy the final result
152 : // instead.
153 :
154 : // Allocate the max possible buffer. Try to use stack instead of heap,
155 : // which would have to be reallocated most times anyways.
156 0 : const sal_Int32 nLocalBuf = 2048;
157 0 : sal_Unicode aLocalBuf[ nLocalBuf * NMAPPINGMAX ], *out = aLocalBuf, *pHeapBuf = NULL;
158 0 : if ( nCount > nLocalBuf )
159 0 : out = pHeapBuf = new sal_Unicode[ nCount * NMAPPINGMAX ];
160 :
161 0 : sal_Int32 j = 0;
162 0 : for ( sal_Int32 i = 0; i < nCount; i++)
163 : {
164 : // take care of TOGGLE_CASE transliteration:
165 0 : sal_uInt8 nTmpMappingType = nMappingType;
166 0 : if (nMappingType == (MappingTypeLowerToUpper | MappingTypeUpperToLower))
167 0 : nTmpMappingType = lcl_getMappingTypeForToggleCase( nMappingType, in[i] );
168 :
169 0 : const Mapping &map = casefolding::getValue( in, i, nCount, aLocale, nTmpMappingType );
170 0 : for (sal_Int32 k = 0; k < map.nmap; k++)
171 : {
172 0 : out[j++] = map.map[k];
173 : }
174 : }
175 :
176 0 : OUString aRet( out, j );
177 0 : if ( pHeapBuf )
178 0 : delete [] pHeapBuf;
179 0 : return aRet;
180 : }
181 : }
182 :
183 : OUString SAL_CALL
184 0 : Transliteration_body::transliterateChar2String( sal_Unicode inChar ) throw(RuntimeException, std::exception)
185 : {
186 0 : const Mapping &map = casefolding::getValue(&inChar, 0, 1, aLocale, nMappingType);
187 0 : rtl_uString* pStr = rtl_uString_alloc(map.nmap);
188 0 : sal_Unicode* out = pStr->buffer;
189 : sal_Int32 i;
190 :
191 0 : for (i = 0; i < map.nmap; i++)
192 0 : out[i] = map.map[i];
193 0 : out[i] = 0;
194 :
195 0 : return OUString( pStr, SAL_NO_ACQUIRE );
196 : }
197 :
198 : sal_Unicode SAL_CALL
199 0 : Transliteration_body::transliterateChar2Char( sal_Unicode inChar ) throw(MultipleCharsOutputException, RuntimeException, std::exception)
200 : {
201 0 : const Mapping &map = casefolding::getValue(&inChar, 0, 1, aLocale, nMappingType);
202 0 : if (map.nmap > 1)
203 0 : throw MultipleCharsOutputException();
204 0 : return map.map[0];
205 : }
206 :
207 : OUString SAL_CALL
208 0 : Transliteration_body::folding( const OUString& inStr, sal_Int32 startPos, sal_Int32 nCount,
209 : Sequence< sal_Int32 >& offset) throw(RuntimeException, std::exception)
210 : {
211 0 : return this->transliterate(inStr, startPos, nCount, offset);
212 : }
213 :
214 0 : Transliteration_casemapping::Transliteration_casemapping()
215 : {
216 0 : nMappingType = 0;
217 0 : transliterationName = "casemapping(generic)";
218 0 : implementationName = "com.sun.star.i18n.Transliteration.Transliteration_casemapping";
219 0 : }
220 :
221 : void SAL_CALL
222 0 : Transliteration_casemapping::setMappingType( const sal_uInt8 rMappingType, const Locale& rLocale )
223 : {
224 0 : nMappingType = rMappingType;
225 0 : aLocale = rLocale;
226 0 : }
227 :
228 0 : Transliteration_u2l::Transliteration_u2l()
229 : {
230 0 : nMappingType = MappingTypeUpperToLower;
231 0 : transliterationName = "upper_to_lower(generic)";
232 0 : implementationName = "com.sun.star.i18n.Transliteration.Transliteration_u2l";
233 0 : }
234 :
235 0 : Transliteration_l2u::Transliteration_l2u()
236 : {
237 0 : nMappingType = MappingTypeLowerToUpper;
238 0 : transliterationName = "lower_to_upper(generic)";
239 0 : implementationName = "com.sun.star.i18n.Transliteration.Transliteration_l2u";
240 0 : }
241 :
242 0 : Transliteration_togglecase::Transliteration_togglecase()
243 : {
244 : // usually nMappingType must NOT be a combiantion of different flages here,
245 : // but we take care of that problem in Transliteration_body::transliterate above
246 : // before that value is used. There we will decide which of both is to be used on
247 : // a per character basis.
248 0 : nMappingType = MappingTypeLowerToUpper | MappingTypeUpperToLower;
249 0 : transliterationName = "toggle(generic)";
250 0 : implementationName = "com.sun.star.i18n.Transliteration.Transliteration_togglecase";
251 0 : }
252 :
253 0 : Transliteration_titlecase::Transliteration_titlecase()
254 : {
255 0 : nMappingType = MappingTypeToTitle;
256 0 : transliterationName = "title(generic)";
257 0 : implementationName = "com.sun.star.i18n.Transliteration.Transliteration_titlecase";
258 0 : }
259 :
260 0 : static OUString transliterate_titlecase_Impl(
261 : const OUString& inStr, sal_Int32 startPos, sal_Int32 nCount,
262 : const Locale &rLocale,
263 : Sequence< sal_Int32 >& offset )
264 : throw(RuntimeException)
265 : {
266 0 : const OUString aText( inStr.copy( startPos, nCount ) );
267 :
268 0 : OUString aRes;
269 0 : if (!aText.isEmpty())
270 : {
271 0 : Reference< XComponentContext > xContext = ::comphelper::getProcessComponentContext();
272 0 : CharacterClassificationImpl aCharClassImpl( xContext );
273 :
274 : // because aCharClassImpl.toTitle does not handle ligatures or Beta but will raise
275 : // an exception we need to handle the first chara manually...
276 :
277 : // we don't want to change surrogates by accident, thuse we use proper code point iteration
278 0 : sal_Int32 nPos = 0;
279 0 : sal_uInt32 cFirstChar = aText.iterateCodePoints( &nPos );
280 0 : OUString aResolvedLigature( &cFirstChar, 1 );
281 : // toUpper can be used to properly resolve ligatures and characters like Beta
282 0 : aResolvedLigature = aCharClassImpl.toUpper( aResolvedLigature, 0, aResolvedLigature.getLength(), rLocale );
283 : // since toTitle will leave all-uppercase text unchanged we first need to
284 : // use toLower to bring possible 2nd and following charas in lowercase
285 0 : aResolvedLigature = aCharClassImpl.toLower( aResolvedLigature, 0, aResolvedLigature.getLength(), rLocale );
286 0 : sal_Int32 nResolvedLen = aResolvedLigature.getLength();
287 :
288 : // now we can properly use toTitle to get the expected result for the resolved string.
289 : // The rest of the text should just become lowercase.
290 0 : aRes = aCharClassImpl.toTitle( aResolvedLigature, 0, nResolvedLen, rLocale );
291 0 : aRes += aCharClassImpl.toLower( aText, 1, aText.getLength() - 1, rLocale );
292 0 : offset.realloc( aRes.getLength() );
293 :
294 0 : sal_Int32 *pOffset = offset.getArray();
295 0 : sal_Int32 nLen = offset.getLength();
296 0 : for (sal_Int32 i = 0; i < nLen; ++i)
297 : {
298 0 : sal_Int32 nIdx = 0;
299 0 : if (i >= nResolvedLen)
300 0 : nIdx = i - nResolvedLen + 1;
301 0 : pOffset[i] = nIdx;
302 0 : }
303 : }
304 : #if OSL_DEBUG_LEVEL > 1
305 : const sal_Int32 *pCOffset = offset.getConstArray();
306 : (void) pCOffset;
307 : #endif
308 :
309 0 : return aRes;
310 : }
311 :
312 : // this function expects to be called on a word-by-word basis,
313 : // namely that startPos points to the first char of the word
314 0 : OUString SAL_CALL Transliteration_titlecase::transliterate(
315 : const OUString& inStr, sal_Int32 startPos, sal_Int32 nCount,
316 : Sequence< sal_Int32 >& offset )
317 : throw(RuntimeException, std::exception)
318 : {
319 0 : return transliterate_titlecase_Impl( inStr, startPos, nCount, aLocale, offset );
320 : }
321 :
322 0 : Transliteration_sentencecase::Transliteration_sentencecase()
323 : {
324 0 : nMappingType = MappingTypeToTitle; // though only to be applied to the first word...
325 0 : transliterationName = "sentence(generic)";
326 0 : implementationName = "com.sun.star.i18n.Transliteration.Transliteration_sentencecase";
327 0 : }
328 :
329 : // this function expects to be called on a sentence-by-sentence basis,
330 : // namely that startPos points to the first word (NOT first char!) in the sentence
331 0 : OUString SAL_CALL Transliteration_sentencecase::transliterate(
332 : const OUString& inStr, sal_Int32 startPos, sal_Int32 nCount,
333 : Sequence< sal_Int32 >& offset )
334 : throw(RuntimeException, std::exception)
335 : {
336 0 : return transliterate_titlecase_Impl( inStr, startPos, nCount, aLocale, offset );
337 : }
338 :
339 : } } } }
340 :
341 : /* vim:set shiftwidth=4 softtabstop=4 expandtab: */
|
