1 /*************************************************************************
2  *
3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4  *
5  * Copyright 2000, 2010 Oracle and/or its affiliates.
6  *
7  * OpenOffice.org - a multi-platform office productivity suite
8  *
9  * This file is part of OpenOffice.org.
10  *
11  * OpenOffice.org is free software: you can redistribute it and/or modify
12  * it under the terms of the GNU Lesser General Public License version 3
13  * only, as published by the Free Software Foundation.
14  *
15  * OpenOffice.org is distributed in the hope that it will be useful,
16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18  * GNU Lesser General Public License version 3 for more details
19  * (a copy is included in the LICENSE file that accompanied this code).
20  *
21  * You should have received a copy of the GNU Lesser General Public License
22  * version 3 along with OpenOffice.org.  If not, see
23  * <http://www.openoffice.org/license.html>
24  * for a copy of the LGPLv3 License.
25  *
26  ************************************************************************/
27 
28 // MARKER(update_precomp.py): autogen include statement, do not remove
29 #include "precompiled_i18npool.hxx"
30 #include <assert.h>
31 #include <textconversion.hxx>
32 #include <com/sun/star/i18n/TextConversionType.hpp>
33 #include <com/sun/star/i18n/TextConversionOption.hpp>
34 #include <com/sun/star/linguistic2/ConversionDirection.hpp>
35 #include <com/sun/star/linguistic2/ConversionDictionaryType.hpp>
36 #include <rtl/ustrbuf.hxx>
37 #include <i18nutil/x_rtl_ustring.h>
38 #include <unicode/uchar.h>
39 
40 using namespace com::sun::star::lang;
41 using namespace com::sun::star::i18n;
42 using namespace com::sun::star::linguistic2;
43 using namespace com::sun::star::uno;
44 using namespace rtl;
45 
46 namespace com { namespace sun { namespace star { namespace i18n {
47 
48 #define SCRIPT_OTHERS   0
49 #define SCRIPT_HANJA    1
50 #define SCRIPT_HANGUL   2
51 
52 TextConversion_ko::TextConversion_ko( const Reference < XMultiServiceFactory >& xMSF )
53 {
54     Reference < XInterface > xI;
55 
56     xI = xMSF->createInstance(
57         OUString::createFromAscii("com.sun.star.i18n.ConversionDictionary_ko"));
58 
59     if ( xI.is() )
60         xI->queryInterface( getCppuType((const Reference< XConversionDictionary>*)0) ) >>= xCD;
61 
62     xI = xMSF->createInstance(
63         OUString::createFromAscii( "com.sun.star.linguistic2.ConversionDictionaryList" ));
64 
65     if ( xI.is() )
66         xI->queryInterface( getCppuType((const Reference< XConversionDictionaryList>*)0) ) >>= xCDL;
67 
68     maxLeftLength = maxRightLength = 1;
69 
70     // get maximum length of word in dictionary
71     if (xCDL.is()) {
72         Locale loc(OUString::createFromAscii("ko"),
73                     OUString::createFromAscii("KR"),
74                     OUString());
75         maxLeftLength = xCDL->queryMaxCharCount(loc,
76                         ConversionDictionaryType::HANGUL_HANJA,
77                         ConversionDirection_FROM_LEFT);
78         maxRightLength = xCDL->queryMaxCharCount(loc,
79                         ConversionDictionaryType::HANGUL_HANJA,
80                         ConversionDirection_FROM_RIGHT);
81         if (xCD.is()) {
82             sal_Int32 tmp = xCD->getMaxCharCount(ConversionDirection_FROM_LEFT);
83             if (tmp > maxLeftLength)
84                 maxLeftLength = tmp;
85             tmp = xCD->getMaxCharCount(ConversionDirection_FROM_RIGHT);
86             if (tmp > maxRightLength)
87                 maxRightLength = tmp;
88         }
89     } else if (xCD.is()) {
90         maxLeftLength = xCD->getMaxCharCount(ConversionDirection_FROM_LEFT);
91         maxRightLength = xCD->getMaxCharCount(ConversionDirection_FROM_RIGHT);
92     }
93 
94     implementationName = "com.sun.star.i18n.TextConversion_ko";
95 }
96 
97 sal_Int16 SAL_CALL checkScriptType(sal_Unicode c)
98 {
99     typedef struct {
100         UBlockCode from;
101         UBlockCode to;
102         sal_Int16 script;
103     } UBlock2Script;
104 
105     static UBlock2Script scriptList[] = {
106         {UBLOCK_HANGUL_JAMO, UBLOCK_HANGUL_JAMO, SCRIPT_HANGUL},
107         {UBLOCK_CJK_RADICALS_SUPPLEMENT, UBLOCK_BOPOMOFO, SCRIPT_HANJA},
108         {UBLOCK_HANGUL_COMPATIBILITY_JAMO, UBLOCK_HANGUL_COMPATIBILITY_JAMO, SCRIPT_HANGUL},
109         {UBLOCK_KANBUN, UBLOCK_YI_RADICALS, SCRIPT_HANJA},
110         {UBLOCK_HANGUL_SYLLABLES, UBLOCK_HANGUL_SYLLABLES, SCRIPT_HANGUL},
111         {UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS, UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS, SCRIPT_HANJA},
112         {UBLOCK_COMBINING_HALF_MARKS, UBLOCK_SMALL_FORM_VARIANTS, SCRIPT_HANJA},
113         {UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS, UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS, SCRIPT_HANJA},
114     };
115 
116 #define scriptListCount sizeof (scriptList) / sizeof (UBlock2Script)
117 
118     UBlockCode block=ublock_getCode((sal_uInt32) c);
119     sal_uInt16 i;
120     for ( i = 0; i < scriptListCount; i++) {
121         if (block <= scriptList[i].to) break;
122     }
123     return (i < scriptListCount && block >= scriptList[i].from) ? scriptList[i].script : SCRIPT_OTHERS;
124 }
125 
126 Sequence< OUString > SAL_CALL
127 TextConversion_ko::getCharConversions(const OUString& aText, sal_Int32 nStartPos, sal_Int32 nLength, sal_Bool toHanja)
128 {
129     sal_Unicode ch;
130     Sequence< OUString > output;
131     const sal_Unicode* (*getHangul2HanjaData)() = (const sal_Unicode* (*)())getFunctionBySymbol("getHangul2HanjaData");
132     const Hangul_Index* (*getHangul2HanjaIndex)() = (const Hangul_Index* (*)()) getFunctionBySymbol("getHangul2HanjaIndex");
133     sal_Int16 (*getHangul2HanjaIndexCount)() = (sal_Int16 (*)()) getFunctionBySymbol("getHangul2HanjaIndexCount");
134     const sal_uInt16* (*getHanja2HangulIndex)() = (const sal_uInt16* (*)()) getFunctionBySymbol("getHanja2HangulIndex");
135     const sal_Unicode* (*getHanja2HangulData)() = (const sal_Unicode* (*)()) getFunctionBySymbol("getHanja2HangulData");
136     if (toHanja && getHangul2HanjaIndex && getHangul2HanjaIndexCount && getHangul2HanjaData) {
137         ch = aText[nStartPos];
138         const Hangul_Index *Hangul_ko = getHangul2HanjaIndex();
139 	    sal_Int16 top =  getHangul2HanjaIndexCount();
140         --top;
141 	    sal_Int16 bottom = 0;
142 
143 	    while (bottom <= top) {
144             sal_Int16 current = (top + bottom) / 2;
145             sal_Unicode current_ch = Hangul_ko[current].code;
146             if (ch < current_ch)
147                 top = current - 1;
148             else if (ch > current_ch)
149                 bottom = current + 1;
150             else {
151                 const sal_Unicode *ptr = getHangul2HanjaData() + Hangul_ko[current].address;
152                 sal_Int16 count = Hangul_ko[current].count;
153                 output.realloc(count);
154                 for (sal_Int16 i = 0; i < count; i++)
155                     output[i] = OUString(ptr + i, 1);
156                 break;
157             }
158 	    }
159     } else if (! toHanja && getHanja2HangulIndex && getHanja2HangulData) {
160         rtl_uString * newStr = x_rtl_uString_new_WithLength( nLength ); // defined in x_rtl_ustring.h
161         sal_Int32 count = 0;
162         while (count < nLength) {
163             ch = aText[nStartPos + count];
164             sal_Unicode address = getHanja2HangulIndex()[ch>>8];
165             if (address != 0xFFFF)
166                 address = getHanja2HangulData()[address + (ch & 0xFF)];
167 
168             if (address != 0xFFFF)
169                 newStr->buffer[count++] = address;
170             else
171                 break;
172         }
173         if (count > 0) {
174             output.realloc(1);
175             output[0] = OUString( newStr->buffer, count);
176         }
177     }
178     return output;
179 }
180 
181 static Sequence< OUString >& operator += (Sequence< OUString > &rSeq1, Sequence< OUString > &rSeq2 )
182 {
183     if (! rSeq1.hasElements() && rSeq2.hasElements())
184         rSeq1 = rSeq2;
185     else if (rSeq2.hasElements()) {
186         sal_Int32 i, j, k, l;
187         k = l = rSeq1.getLength();
188         rSeq1.realloc(l + rSeq2.getLength());
189 
190         for (i = 0; i < rSeq2.getLength(); i++) {
191             for (j = 0; j < l; j++)
192                 if (rSeq1[j] == rSeq2[i])
193                     break;
194             if (j == l)
195                 rSeq1[k++] = rSeq2[i];
196         }
197         if (rSeq1.getLength() > k)
198             rSeq1.realloc(k);
199     }
200     return rSeq1;
201 }
202 
203 TextConversionResult SAL_CALL
204 TextConversion_ko::getConversions( const OUString& aText, sal_Int32 nStartPos, sal_Int32 nLength,
205     const Locale& aLocale, sal_Int16 nConversionType, sal_Int32 nConversionOptions)
206     throw(  RuntimeException, IllegalArgumentException, NoSupportException )
207 {
208     TextConversionResult result;
209     Sequence <OUString> candidates;
210     result.Boundary.startPos = result.Boundary.endPos = 0;
211 
212     // do conversion only when there are right conversion type and dictionary services.
213     if (nConversionType == TextConversionType::TO_HANGUL ||
214             nConversionType == TextConversionType::TO_HANJA) {
215         sal_Int32 start, end, length = aText.getLength() - nStartPos;
216 
217         if (length < 0 || nStartPos < 0)
218             length = 0;
219         else if (length > nLength)
220             length = nLength;
221 
222         sal_Int16 scriptType = SCRIPT_OTHERS;
223         sal_Int32 len = 1;
224         sal_Bool toHanja = (nConversionType == TextConversionType::TO_HANJA);
225         // FROM_LEFT:  Hangul -> Hanja
226         // FROM_RIGHT: Hanja  -> Hangul
227         ConversionDirection eDirection = toHanja ? ConversionDirection_FROM_LEFT : ConversionDirection_FROM_RIGHT;
228         sal_Int32 maxLength = toHanja ? maxLeftLength : maxRightLength;
229         if (maxLength == 0) maxLength = 1;
230 
231         // search for a max length of convertible text
232         for (start = 0, end = 0; start < length; start++) {
233             if (end <= start) {
234                 scriptType = checkScriptType(aText[nStartPos + start]);
235                 if (nConversionType == TextConversionType::TO_HANJA) {
236                     if (scriptType != SCRIPT_HANGUL) // skip non-Hangul characters
237                         continue;
238                 } else {
239                     if (scriptType != SCRIPT_HANJA) // skip non-Hanja characters
240                         continue;
241                 }
242                 end = start + 1;
243             }
244             if (nConversionOptions & TextConversionOption::CHARACTER_BY_CHARACTER) {
245                 result.Candidates = getCharConversions(aText, nStartPos + start, len, toHanja); // char2char conversion
246             } else {
247                 for (; end < length && end - start < maxLength; end++)
248                     if (checkScriptType(aText[nStartPos + end]) != scriptType)
249                         break;
250 
251                 for (len = end - start; len > 0; len--) {
252                     if (len > 1) {
253                         try {
254                             if (xCDL.is())
255                                 result.Candidates = xCDL->queryConversions(aText, start + nStartPos, len,
256                                     aLocale, ConversionDictionaryType::HANGUL_HANJA, eDirection, nConversionOptions); // user dictionary
257                         }
258                         catch ( NoSupportException & ) {
259                             // clear reference (when there is no user dictionary) in order
260                             // to not always have to catch this exception again
261                             // in further calls. (save time)
262                             xCDL = 0;
263                         }
264                         catch (...) {
265                             // catch all other exceptions to allow
266                             // querying the system dictionary in the next line
267                         }
268                         if (xCD.is() && toHanja) { // System dictionary would not do Hanja_to_Hangul conversion.
269                             candidates = xCD->getConversions(aText, start + nStartPos, len, eDirection, nConversionOptions);
270                             result.Candidates += candidates;
271                         }
272                     } else if (! toHanja) { // do whole word character 2 character conversion for Hanja to Hangul conversion
273                         result.Candidates = getCharConversions(aText, nStartPos + start, length - start, toHanja);
274                         if (result.Candidates.hasElements())
275                             len = result.Candidates[0].getLength();
276                     }
277                     if (result.Candidates.hasElements())
278                         break;
279                 }
280             }
281             // found match
282             if (result.Candidates.hasElements()) {
283                 result.Boundary.startPos = start + nStartPos;;
284                 result.Boundary.endPos = start + len + nStartPos;
285                 return result;
286             }
287         }
288     } else
289         throw NoSupportException(); // Conversion type is not supported in this service.
290     return result;
291 }
292 
293 OUString SAL_CALL
294 TextConversion_ko::getConversion( const OUString& aText, sal_Int32 nStartPos, sal_Int32 nLength,
295     const Locale& aLocale, sal_Int16 nConversionType, sal_Int32 nConversionOptions)
296     throw(  RuntimeException, IllegalArgumentException, NoSupportException )
297 {
298     sal_Int32 length = aText.getLength() - nStartPos;
299 
300     if (length <= 0 || nStartPos < 0)
301         return OUString();
302     else if (length > nLength)
303         length = nLength;
304 
305     OUStringBuffer aBuf(length + 1);
306     TextConversionResult result;
307     const sal_Unicode *str = aText.getStr();
308 
309     for (sal_Int32 start = nStartPos; length + nStartPos > start; start = result.Boundary.endPos) {
310 
311         result = getConversions(aText, start, length + nStartPos - start, aLocale, nConversionType, nConversionOptions);
312 
313         if (result.Boundary.endPos > 0) {
314             if (result.Boundary.startPos > start)
315                 aBuf.append(str + start, result.Boundary.startPos - start); // append skip portion
316             aBuf.append(result.Candidates[0]); // append converted portion
317         } else {
318             if (length + nStartPos > start)
319                 aBuf.append(str + start, length + nStartPos - start); // append last portion
320             break;
321         }
322     }
323 
324     return aBuf.makeStringAndClear();
325 }
326 
327 OUString SAL_CALL
328 TextConversion_ko::getConversionWithOffset( const OUString& aText, sal_Int32 nStartPos, sal_Int32 nLength,
329     const Locale& rLocale, sal_Int16 nConversionType, sal_Int32 nConversionOptions, Sequence<sal_Int32>& offset)
330     throw(  RuntimeException, IllegalArgumentException, NoSupportException )
331 {
332     offset.realloc(0);
333     return getConversion(aText, nStartPos, nLength, rLocale, nConversionType, nConversionOptions);
334 }
335 
336 sal_Bool SAL_CALL
337 TextConversion_ko::interactiveConversion( const Locale& /*rLocale*/, sal_Int16 /*nTextConversionType*/, sal_Int32 /*nTextConversionOptions*/ )
338     throw(  RuntimeException, IllegalArgumentException, NoSupportException )
339 {
340     return sal_True;
341 }
342 
343 } } } }
344