Line data Source code
1 : /* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
2 : /*
3 : * Version: MPL 1.1 / GPLv3+ / LGPLv3+
4 : *
5 : * The contents of this file are subject to the Mozilla Public License Version
6 : * 1.1 (the "License"); you may not use this file except in compliance with
7 : * the License or as specified alternatively below. You may obtain a copy of
8 : * the License at http://www.mozilla.org/MPL/
9 : *
10 : * Software distributed under the License is distributed on an "AS IS" basis,
11 : * WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
12 : * for the specific language governing rights and limitations under the
13 : * License.
14 : *
15 : * Major Contributor(s):
16 : * [ Copyright (C) 2012 Red Hat, Inc., Stephan Bergmann <sbergman@redhat.com>
17 : * (initial developer) ]
18 : *
19 : * All Rights Reserved.
20 : *
21 : * For minor contributions see the git repository.
22 : *
23 : * Alternatively, the contents of this file may be used under the terms of
24 : * either the GNU General Public License Version 3 or later (the "GPLv3+"), or
25 : * the GNU Lesser General Public License Version 3 or later (the "LGPLv3+"),
26 : * in which case the provisions of the GPLv3+ or the LGPLv3+ are applicable
27 : * instead of those above.
28 : */
29 :
30 : #include "sal/config.h"
31 :
32 : #include "rtl/textcvt.h"
33 : #include "sal/types.h"
34 :
35 : #include "convertsimple.hxx"
36 : #include "handleundefinedunicodetotextchar.hxx"
37 : #include "tenchelp.hxx"
38 :
39 : #define IMPL_MAX_REPLACECHAR 5
40 :
41 : namespace {
42 :
43 : struct ImplReplaceCharData
44 : {
45 : sal_uInt16 mnUniChar;
46 : sal_uInt16 mnReplaceChar;
47 : };
48 :
49 : static ImplReplaceCharData const aImplRepCharTab[] =
50 : {
51 : { 0x00A0, 0x0020 }, /* NO-BREAK-SPACE */
52 : { 0x00A1, 0x0021 }, /* INVERTED EXCLAMATION MARK */
53 : { 0x00B7, 0x0045 }, /* MIDDLE DOT */
54 : { 0x00BF, 0x003F }, /* INVERTED QUESTION MARK */
55 : { 0x00D7, 0x002A }, /* MULTIPLIKATION SIGN */
56 : { 0x00F7, 0x002F }, /* DIVISION SIGN */
57 : { 0x2000, 0x0020 }, /* EN QUAD */
58 : { 0x2001, 0x0020 }, /* EM QUAD */
59 : { 0x2002, 0x0020 }, /* EN SPACE */
60 : { 0x2003, 0x0020 }, /* EM SPACE */
61 : { 0x2004, 0x0020 }, /* THREE-PER-EM SPACE */
62 : { 0x2005, 0x0020 }, /* FOUR-PER-EM SPACE */
63 : { 0x2006, 0x0020 }, /* SIX-PER-EM SPACE */
64 : { 0x2007, 0x0020 }, /* FIGURE SPACE */
65 : { 0x2008, 0x0020 }, /* PUNCTATION SPACE */
66 : { 0x2009, 0x0020 }, /* THIN SPACE */
67 : { 0x200A, 0x0020 }, /* HAIR SPACE */
68 : { 0x2010, 0x002D }, /* HYPHEN */
69 : { 0x2011, 0x002D }, /* NON-BREAKING HYPHEN */
70 : { 0x2012, 0x002D }, /* FIGURE DASH */
71 : { 0x2013, 0x002D }, /* EN DASH */
72 : { 0x2014, 0x002D }, /* EM DASH */
73 : { 0x2015, 0x002D }, /* HORIZONTAL BAR */
74 : { 0x2018, 0x0027 }, /* LEFT SINGLE QUOTATION MARK */
75 : { 0x2019, 0x0027 }, /* RIGHT SINGLE QUOTATION MARK */
76 : { 0x201A, 0x002C }, /* SINGLE LOW-9 QUOTATION MARK */
77 : { 0x201B, 0x0027 }, /* SINGLE HIGH-RESERVED-9 QUOTATION MARK */
78 : { 0x201C, 0x0022 }, /* LEFT DOUBLE QUOTATION MARK */
79 : { 0x201D, 0x0022 }, /* RIGHT DOUBLE QUOTATION MARK */
80 : { 0x201E, 0x0022 }, /* DOUBLE LOW-9 QUOTATION MARK */
81 : { 0x201F, 0x0022 }, /* DOUBLE HIGH-RESERVED-9 QUOTATION MARK */
82 : { 0x2022, 0x002D }, /* BULLET */
83 : { 0x2023, 0x002D }, /* TRIANGULAR BULLET */
84 : { 0x2024, 0x002D }, /* ONE DOT LEADER */
85 : { 0x2027, 0x002D }, /* HYPHENATION POINT */
86 : { 0x2028, 0x000A }, /* LINE SEPARATOR */
87 : { 0x2029, 0x000D }, /* PARAGRAPH SEPARATOR */
88 : { 0x2032, 0x0027 }, /* PRIME */
89 : { 0x2033, 0x0022 }, /* DOUBLE PRIME */
90 : { 0x2035, 0x0027 }, /* RESERVED PRIME */
91 : { 0x2036, 0x0022 }, /* RESERVED DOUBLE PRIME */
92 : { 0x2039, 0x003C }, /* SINGLE LEFT-POINTING ANGLE QUOTATION MARK */
93 : { 0x203A, 0x003E }, /* SINGLE RIGHT-POINTING ANGLE QUOTATION MARK */
94 : { 0x2043, 0x002D }, /* HYPHEN BULLET */
95 : { 0x2044, 0x002F }, /* FRACTION SLASH */
96 : { 0x2160, 0x0049 }, /* ROMAN NUMERAL ONE */
97 : { 0x2164, 0x0056 }, /* ROMAN NUMERAL FIVE */
98 : { 0x2169, 0x0058 }, /* ROMAN NUMERAL TEN */
99 : { 0x216C, 0x004C }, /* ROMAN NUMERAL FIFTY */
100 : { 0x216D, 0x0043 }, /* ROMAN NUMERAL ONE HUNDRED */
101 : { 0x216E, 0x0044 }, /* ROMAN NUMERAL FIVE HUNDRED */
102 : { 0x216F, 0x004D }, /* ROMAN NUMERAL ONE THOUSAND */
103 : { 0x2170, 0x0069 }, /* SMALL ROMAN NUMERAL ONE */
104 : { 0x2174, 0x0076 }, /* SMALL ROMAN NUMERAL FIVE */
105 : { 0x2179, 0x0078 }, /* SMALL ROMAN NUMERAL TEN */
106 : { 0x217C, 0x006C }, /* SMALL ROMAN NUMERAL FIFTY */
107 : { 0x217D, 0x0063 }, /* SMALL ROMAN NUMERAL ONE HUNDRED */
108 : { 0x217E, 0x0064 }, /* SMALL ROMAN NUMERAL FIVE HUNDRED */
109 : { 0x217F, 0x006D }, /* SMALL ROMAN NUMERAL ONE THOUSAND */
110 : { 0x2215, 0x002F }, /* DIVISION SLASH */
111 : { 0x2217, 0x002A }, /* ASTERIX OPERATOR */
112 : { 0xFF00, 0x0020 }, /* FULLWIDTH ASCII FORMS */
113 : { 0xFF01, 0x0021 }, /* FULLWIDTH ASCII FORMS */
114 : { 0xFF02, 0x0022 }, /* FULLWIDTH ASCII FORMS*/
115 : { 0xFF03, 0x0023 }, /* FULLWIDTH ASCII FORMS */
116 : { 0xFF04, 0x0024 }, /* FULLWIDTH ASCII FORMS*/
117 : { 0xFF05, 0x0025 }, /* FULLWIDTH ASCII FORMS */
118 : { 0xFF06, 0x0026 }, /* FULLWIDTH ASCII FORMS*/
119 : { 0xFF07, 0x0027 }, /* FULLWIDTH ASCII FORMS */
120 : { 0xFF08, 0x0028 }, /* FULLWIDTH ASCII FORMS*/
121 : { 0xFF09, 0x0029 }, /* FULLWIDTH ASCII FORMS */
122 : { 0xFF0A, 0x002A }, /* FULLWIDTH ASCII FORMS*/
123 : { 0xFF0B, 0x002B }, /* FULLWIDTH ASCII FORMS */
124 : { 0xFF0C, 0x002C }, /* FULLWIDTH ASCII FORMS*/
125 : { 0xFF0D, 0x002D }, /* FULLWIDTH ASCII FORMS */
126 : { 0xFF0E, 0x002E }, /* FULLWIDTH ASCII FORMS*/
127 : { 0xFF0F, 0x002F }, /* FULLWIDTH ASCII FORMS */
128 : { 0xFF10, 0x0030 }, /* FULLWIDTH ASCII FORMS */
129 : { 0xFF11, 0x0031 }, /* FULLWIDTH ASCII FORMS */
130 : { 0xFF12, 0x0032 }, /* FULLWIDTH ASCII FORMS*/
131 : { 0xFF13, 0x0033 }, /* FULLWIDTH ASCII FORMS */
132 : { 0xFF14, 0x0034 }, /* FULLWIDTH ASCII FORMS*/
133 : { 0xFF15, 0x0035 }, /* FULLWIDTH ASCII FORMS */
134 : { 0xFF16, 0x0036 }, /* FULLWIDTH ASCII FORMS*/
135 : { 0xFF17, 0x0037 }, /* FULLWIDTH ASCII FORMS */
136 : { 0xFF18, 0x0038 }, /* FULLWIDTH ASCII FORMS*/
137 : { 0xFF19, 0x0039 }, /* FULLWIDTH ASCII FORMS */
138 : { 0xFF1A, 0x003A }, /* FULLWIDTH ASCII FORMS*/
139 : { 0xFF1B, 0x003B }, /* FULLWIDTH ASCII FORMS */
140 : { 0xFF1C, 0x003C }, /* FULLWIDTH ASCII FORMS*/
141 : { 0xFF1D, 0x003D }, /* FULLWIDTH ASCII FORMS */
142 : { 0xFF1E, 0x003E }, /* FULLWIDTH ASCII FORMS*/
143 : { 0xFF1F, 0x003F }, /* FULLWIDTH ASCII FORMS */
144 : { 0xFF20, 0x0040 }, /* FULLWIDTH ASCII FORMS */
145 : { 0xFF21, 0x0041 }, /* FULLWIDTH ASCII FORMS */
146 : { 0xFF22, 0x0042 }, /* FULLWIDTH ASCII FORMS*/
147 : { 0xFF23, 0x0043 }, /* FULLWIDTH ASCII FORMS */
148 : { 0xFF24, 0x0044 }, /* FULLWIDTH ASCII FORMS*/
149 : { 0xFF25, 0x0045 }, /* FULLWIDTH ASCII FORMS */
150 : { 0xFF26, 0x0046 }, /* FULLWIDTH ASCII FORMS*/
151 : { 0xFF27, 0x0047 }, /* FULLWIDTH ASCII FORMS */
152 : { 0xFF28, 0x0048 }, /* FULLWIDTH ASCII FORMS*/
153 : { 0xFF29, 0x0049 }, /* FULLWIDTH ASCII FORMS */
154 : { 0xFF2A, 0x004A }, /* FULLWIDTH ASCII FORMS*/
155 : { 0xFF2B, 0x004B }, /* FULLWIDTH ASCII FORMS */
156 : { 0xFF2C, 0x004C }, /* FULLWIDTH ASCII FORMS*/
157 : { 0xFF2D, 0x004D }, /* FULLWIDTH ASCII FORMS */
158 : { 0xFF2E, 0x004E }, /* FULLWIDTH ASCII FORMS*/
159 : { 0xFF2F, 0x004F }, /* FULLWIDTH ASCII FORMS */
160 : { 0xFF30, 0x0050 }, /* FULLWIDTH ASCII FORMS */
161 : { 0xFF31, 0x0051 }, /* FULLWIDTH ASCII FORMS */
162 : { 0xFF32, 0x0052 }, /* FULLWIDTH ASCII FORMS*/
163 : { 0xFF33, 0x0053 }, /* FULLWIDTH ASCII FORMS */
164 : { 0xFF34, 0x0054 }, /* FULLWIDTH ASCII FORMS*/
165 : { 0xFF35, 0x0055 }, /* FULLWIDTH ASCII FORMS */
166 : { 0xFF36, 0x0056 }, /* FULLWIDTH ASCII FORMS*/
167 : { 0xFF37, 0x0057 }, /* FULLWIDTH ASCII FORMS */
168 : { 0xFF38, 0x0058 }, /* FULLWIDTH ASCII FORMS*/
169 : { 0xFF39, 0x0059 }, /* FULLWIDTH ASCII FORMS */
170 : { 0xFF3A, 0x005A }, /* FULLWIDTH ASCII FORMS*/
171 : { 0xFF3B, 0x005B }, /* FULLWIDTH ASCII FORMS */
172 : { 0xFF3C, 0x005C }, /* FULLWIDTH ASCII FORMS*/
173 : { 0xFF3D, 0x005D }, /* FULLWIDTH ASCII FORMS */
174 : { 0xFF3E, 0x005E }, /* FULLWIDTH ASCII FORMS*/
175 : { 0xFF3F, 0x005F }, /* FULLWIDTH ASCII FORMS */
176 : { 0xFF40, 0x0060 }, /* FULLWIDTH ASCII FORMS */
177 : { 0xFF41, 0x0061 }, /* FULLWIDTH ASCII FORMS */
178 : { 0xFF42, 0x0062 }, /* FULLWIDTH ASCII FORMS*/
179 : { 0xFF43, 0x0063 }, /* FULLWIDTH ASCII FORMS */
180 : { 0xFF44, 0x0064 }, /* FULLWIDTH ASCII FORMS*/
181 : { 0xFF45, 0x0065 }, /* FULLWIDTH ASCII FORMS */
182 : { 0xFF46, 0x0066 }, /* FULLWIDTH ASCII FORMS*/
183 : { 0xFF47, 0x0067 }, /* FULLWIDTH ASCII FORMS */
184 : { 0xFF48, 0x0068 }, /* FULLWIDTH ASCII FORMS*/
185 : { 0xFF49, 0x0069 }, /* FULLWIDTH ASCII FORMS */
186 : { 0xFF4A, 0x006A }, /* FULLWIDTH ASCII FORMS*/
187 : { 0xFF4B, 0x006B }, /* FULLWIDTH ASCII FORMS */
188 : { 0xFF4C, 0x006C }, /* FULLWIDTH ASCII FORMS*/
189 : { 0xFF4D, 0x006D }, /* FULLWIDTH ASCII FORMS */
190 : { 0xFF4E, 0x006E }, /* FULLWIDTH ASCII FORMS*/
191 : { 0xFF4F, 0x006F }, /* FULLWIDTH ASCII FORMS */
192 : { 0xFF50, 0x0070 }, /* FULLWIDTH ASCII FORMS */
193 : { 0xFF51, 0x0071 }, /* FULLWIDTH ASCII FORMS */
194 : { 0xFF52, 0x0072 }, /* FULLWIDTH ASCII FORMS*/
195 : { 0xFF53, 0x0073 }, /* FULLWIDTH ASCII FORMS */
196 : { 0xFF54, 0x0074 }, /* FULLWIDTH ASCII FORMS*/
197 : { 0xFF55, 0x0075 }, /* FULLWIDTH ASCII FORMS */
198 : { 0xFF56, 0x0076 }, /* FULLWIDTH ASCII FORMS*/
199 : { 0xFF57, 0x0077 }, /* FULLWIDTH ASCII FORMS */
200 : { 0xFF58, 0x0078 }, /* FULLWIDTH ASCII FORMS*/
201 : { 0xFF59, 0x0079 }, /* FULLWIDTH ASCII FORMS */
202 : { 0xFF5A, 0x007A }, /* FULLWIDTH ASCII FORMS*/
203 : { 0xFF5B, 0x007B }, /* FULLWIDTH ASCII FORMS */
204 : { 0xFF5C, 0x007C }, /* FULLWIDTH ASCII FORMS*/
205 : { 0xFF5D, 0x007D }, /* FULLWIDTH ASCII FORMS */
206 : { 0xFF5E, 0x007E }, /* FULLWIDTH ASCII FORMS*/
207 : { 0xFF5F, 0x007F }, /* FULLWIDTH ASCII FORMS */
208 : { 0xFF61, 0x3002 }, /* HALFWIDTH KATAKANA FORMS */
209 : { 0xFF62, 0x300C }, /* HALFWIDTH KATAKANA FORMS */
210 : { 0xFF63, 0x300D }, /* HALFWIDTH KATAKANA FORMS */
211 : { 0xFF64, 0x3001 }, /* HALFWIDTH KATAKANA FORMS */
212 : { 0xFF65, 0x30FB }, /* HALFWIDTH KATAKANA FORMS */
213 : { 0xFF66, 0x30F2 }, /* HALFWIDTH KATAKANA FORMS */
214 : { 0xFF67, 0x30A1 }, /* HALFWIDTH KATAKANA FORMS */
215 : { 0xFF68, 0x30A3 }, /* HALFWIDTH KATAKANA FORMS */
216 : { 0xFF69, 0x30A5 }, /* HALFWIDTH KATAKANA FORMS */
217 : { 0xFF6A, 0x30A7 }, /* HALFWIDTH KATAKANA FORMS */
218 : { 0xFF6B, 0x30A9 }, /* HALFWIDTH KATAKANA FORMS */
219 : { 0xFF6C, 0x30E3 }, /* HALFWIDTH KATAKANA FORMS */
220 : { 0xFF6D, 0x30E5 }, /* HALFWIDTH KATAKANA FORMS */
221 : { 0xFF6E, 0x30E7 }, /* HALFWIDTH KATAKANA FORMS */
222 : { 0xFF6F, 0x30C3 }, /* HALFWIDTH KATAKANA FORMS */
223 : { 0xFF70, 0x30FC }, /* HALFWIDTH KATAKANA FORMS */
224 : { 0xFF71, 0x30A2 }, /* HALFWIDTH KATAKANA FORMS */
225 : { 0xFF72, 0x30A4 }, /* HALFWIDTH KATAKANA FORMS */
226 : { 0xFF73, 0x30A6 }, /* HALFWIDTH KATAKANA FORMS */
227 : { 0xFF74, 0x30A8 }, /* HALFWIDTH KATAKANA FORMS */
228 : { 0xFF75, 0x30AA }, /* HALFWIDTH KATAKANA FORMS */
229 : { 0xFF76, 0x30AB }, /* HALFWIDTH KATAKANA FORMS */
230 : { 0xFF77, 0x30AD }, /* HALFWIDTH KATAKANA FORMS */
231 : { 0xFF78, 0x30AF }, /* HALFWIDTH KATAKANA FORMS */
232 : { 0xFF79, 0x30B1 }, /* HALFWIDTH KATAKANA FORMS */
233 : { 0xFF7A, 0x30B3 }, /* HALFWIDTH KATAKANA FORMS */
234 : { 0xFF7B, 0x30B5 }, /* HALFWIDTH KATAKANA FORMS */
235 : { 0xFF7C, 0x30B7 }, /* HALFWIDTH KATAKANA FORMS */
236 : { 0xFF7D, 0x30B9 }, /* HALFWIDTH KATAKANA FORMS */
237 : { 0xFF7E, 0x30BB }, /* HALFWIDTH KATAKANA FORMS */
238 : { 0xFF7F, 0x30BD }, /* HALFWIDTH KATAKANA FORMS */
239 : { 0xFF80, 0x30BF }, /* HALFWIDTH KATAKANA FORMS */
240 : { 0xFF81, 0x30C1 }, /* HALFWIDTH KATAKANA FORMS */
241 : { 0xFF82, 0x30C4 }, /* HALFWIDTH KATAKANA FORMS */
242 : { 0xFF83, 0x30C6 }, /* HALFWIDTH KATAKANA FORMS */
243 : { 0xFF84, 0x30C8 }, /* HALFWIDTH KATAKANA FORMS */
244 : { 0xFF85, 0x30CA }, /* HALFWIDTH KATAKANA FORMS */
245 : { 0xFF86, 0x30CB }, /* HALFWIDTH KATAKANA FORMS */
246 : { 0xFF87, 0x30CC }, /* HALFWIDTH KATAKANA FORMS */
247 : { 0xFF88, 0x30CD }, /* HALFWIDTH KATAKANA FORMS */
248 : { 0xFF89, 0x30CE }, /* HALFWIDTH KATAKANA FORMS */
249 : { 0xFF8A, 0x30CF }, /* HALFWIDTH KATAKANA FORMS */
250 : { 0xFF8B, 0x30D2 }, /* HALFWIDTH KATAKANA FORMS */
251 : { 0xFF8C, 0x30D5 }, /* HALFWIDTH KATAKANA FORMS */
252 : { 0xFF8D, 0x30D8 }, /* HALFWIDTH KATAKANA FORMS */
253 : { 0xFF8E, 0x30DB }, /* HALFWIDTH KATAKANA FORMS */
254 : { 0xFF8F, 0x30DE }, /* HALFWIDTH KATAKANA FORMS */
255 : { 0xFF90, 0x30DF }, /* HALFWIDTH KATAKANA FORMS */
256 : { 0xFF91, 0x30E0 }, /* HALFWIDTH KATAKANA FORMS */
257 : { 0xFF92, 0x30E1 }, /* HALFWIDTH KATAKANA FORMS */
258 : { 0xFF93, 0x30E2 }, /* HALFWIDTH KATAKANA FORMS */
259 : { 0xFF94, 0x30E4 }, /* HALFWIDTH KATAKANA FORMS */
260 : { 0xFF95, 0x30E6 }, /* HALFWIDTH KATAKANA FORMS */
261 : { 0xFF96, 0x30E8 }, /* HALFWIDTH KATAKANA FORMS */
262 : { 0xFF97, 0x30E9 }, /* HALFWIDTH KATAKANA FORMS */
263 : { 0xFF98, 0x30EA }, /* HALFWIDTH KATAKANA FORMS */
264 : { 0xFF99, 0x30EB }, /* HALFWIDTH KATAKANA FORMS */
265 : { 0xFF9A, 0x30EC }, /* HALFWIDTH KATAKANA FORMS */
266 : { 0xFF9B, 0x30ED }, /* HALFWIDTH KATAKANA FORMS */
267 : { 0xFF9C, 0x30EF }, /* HALFWIDTH KATAKANA FORMS */
268 : { 0xFF9D, 0x30F3 }, /* HALFWIDTH KATAKANA FORMS */
269 : { 0xFF9E, 0x309B }, /* HALFWIDTH KATAKANA FORMS */
270 : { 0xFF9F, 0x309C }, /* HALFWIDTH KATAKANA FORMS */
271 : { 0xFFA0, 0x3164 }, /* HALFWIDTH HANGUL FORMS */
272 : { 0xFFA1, 0x3131 }, /* HALFWIDTH HANGUL FORMS */
273 : { 0xFFA2, 0x3132 }, /* HALFWIDTH HANGUL FORMS */
274 : { 0xFFA3, 0x3133 }, /* HALFWIDTH HANGUL FORMS */
275 : { 0xFFA4, 0x3134 }, /* HALFWIDTH HANGUL FORMS */
276 : { 0xFFA5, 0x3135 }, /* HALFWIDTH HANGUL FORMS */
277 : { 0xFFA6, 0x3136 }, /* HALFWIDTH HANGUL FORMS */
278 : { 0xFFA7, 0x3137 }, /* HALFWIDTH HANGUL FORMS */
279 : { 0xFFA8, 0x3138 }, /* HALFWIDTH HANGUL FORMS */
280 : { 0xFFA9, 0x3139 }, /* HALFWIDTH HANGUL FORMS */
281 : { 0xFFAA, 0x313A }, /* HALFWIDTH HANGUL FORMS */
282 : { 0xFFAB, 0x313B }, /* HALFWIDTH HANGUL FORMS */
283 : { 0xFFAC, 0x313C }, /* HALFWIDTH HANGUL FORMS */
284 : { 0xFFAD, 0x313D }, /* HALFWIDTH HANGUL FORMS */
285 : { 0xFFAE, 0x313E }, /* HALFWIDTH HANGUL FORMS */
286 : { 0xFFAF, 0x313F }, /* HALFWIDTH HANGUL FORMS */
287 : { 0xFFB0, 0x3140 }, /* HALFWIDTH HANGUL FORMS */
288 : { 0xFFB1, 0x3141 }, /* HALFWIDTH HANGUL FORMS */
289 : { 0xFFB2, 0x3142 }, /* HALFWIDTH HANGUL FORMS */
290 : { 0xFFB3, 0x3143 }, /* HALFWIDTH HANGUL FORMS */
291 : { 0xFFB4, 0x3144 }, /* HALFWIDTH HANGUL FORMS */
292 : { 0xFFB5, 0x3145 }, /* HALFWIDTH HANGUL FORMS */
293 : { 0xFFB6, 0x3146 }, /* HALFWIDTH HANGUL FORMS */
294 : { 0xFFB7, 0x3147 }, /* HALFWIDTH HANGUL FORMS */
295 : { 0xFFB8, 0x3148 }, /* HALFWIDTH HANGUL FORMS */
296 : { 0xFFB9, 0x3149 }, /* HALFWIDTH HANGUL FORMS */
297 : { 0xFFBA, 0x314A }, /* HALFWIDTH HANGUL FORMS */
298 : { 0xFFBB, 0x314B }, /* HALFWIDTH HANGUL FORMS */
299 : { 0xFFBC, 0x314C }, /* HALFWIDTH HANGUL FORMS */
300 : { 0xFFBD, 0x314D }, /* HALFWIDTH HANGUL FORMS */
301 : { 0xFFBE, 0x314E }, /* HALFWIDTH HANGUL FORMS */
302 : { 0xFFC2, 0x314F }, /* HALFWIDTH HANGUL FORMS */
303 : { 0xFFC3, 0x3150 }, /* HALFWIDTH HANGUL FORMS */
304 : { 0xFFC4, 0x3151 }, /* HALFWIDTH HANGUL FORMS */
305 : { 0xFFC5, 0x3152 }, /* HALFWIDTH HANGUL FORMS */
306 : { 0xFFC6, 0x3153 }, /* HALFWIDTH HANGUL FORMS */
307 : { 0xFFC7, 0x3154 }, /* HALFWIDTH HANGUL FORMS */
308 : { 0xFFCA, 0x3155 }, /* HALFWIDTH HANGUL FORMS */
309 : { 0xFFCB, 0x3156 }, /* HALFWIDTH HANGUL FORMS */
310 : { 0xFFCC, 0x3157 }, /* HALFWIDTH HANGUL FORMS */
311 : { 0xFFCD, 0x3158 }, /* HALFWIDTH HANGUL FORMS */
312 : { 0xFFCE, 0x3159 }, /* HALFWIDTH HANGUL FORMS */
313 : { 0xFFCF, 0x315A }, /* HALFWIDTH HANGUL FORMS */
314 : { 0xFFD2, 0x315B }, /* HALFWIDTH HANGUL FORMS */
315 : { 0xFFD3, 0x315C }, /* HALFWIDTH HANGUL FORMS */
316 : { 0xFFD4, 0x315D }, /* HALFWIDTH HANGUL FORMS */
317 : { 0xFFD5, 0x315E }, /* HALFWIDTH HANGUL FORMS */
318 : { 0xFFD6, 0x315F }, /* HALFWIDTH HANGUL FORMS */
319 : { 0xFFD7, 0x3160 }, /* HALFWIDTH HANGUL FORMS */
320 : { 0xFFDA, 0x3161 }, /* HALFWIDTH HANGUL FORMS */
321 : { 0xFFDB, 0x3162 }, /* HALFWIDTH HANGUL FORMS */
322 : { 0xFFDC, 0x3163 }, /* HALFWIDTH HANGUL FORMS */
323 : { 0xFFE0, 0x00A2 }, /* FULLWIDTH CENT SIGN */
324 : { 0xFFE1, 0x00A3 }, /* FULLWIDTH POUND SIGN */
325 : { 0xFFE2, 0x00AC }, /* FULLWIDTH NOT SIGN */
326 : { 0xFFE3, 0x00AF }, /* FULLWIDTH MACRON */
327 : { 0xFFE4, 0x00A6 }, /* FULLWIDTH BROKEN BAR */
328 : { 0xFFE5, 0x00A5 }, /* FULLWIDTH YEN SIGN */
329 : { 0xFFE6, 0x20A9 }, /* FULLWIDTH WON SIGN */
330 : { 0xFFE8, 0x2502 }, /* HALFWIDTH FORMS LIGHT VERTICAL */
331 : { 0xFFE9, 0x2190 }, /* HALFWIDTH LEFTWARDS ARROW */
332 : { 0xFFEA, 0x2191 }, /* HALFWIDTH UPWARDS ARROW */
333 : { 0xFFEB, 0x2192 }, /* HALFWIDTH RIGHTWARDS ARROW */
334 : { 0xFFEC, 0x2193 }, /* HALFWIDTH DOWNWARDS ARROW */
335 : { 0xFFED, 0x25A0 }, /* HALFWIDTH BLACK SQUARE */
336 : { 0xFFEE, 0x25CB }, /* HALFWIDTH WHITE CIRCLE */
337 : { 0xFFFD, 0x003F } /* REPLACEMENT CHARACTER */
338 : };
339 :
340 4 : sal_uInt16 ImplGetReplaceChar( sal_Unicode c )
341 : {
342 : sal_uInt16 nLow;
343 : sal_uInt16 nHigh;
344 : sal_uInt16 nMid;
345 : sal_uInt16 nCompareChar;
346 : const ImplReplaceCharData* pCharData;
347 :
348 4 : nLow = 0;
349 4 : nHigh = (sizeof( aImplRepCharTab )/sizeof( ImplReplaceCharData ))-1;
350 34 : do
351 : {
352 34 : nMid = (nLow+nHigh)/2;
353 34 : pCharData = aImplRepCharTab+nMid;
354 34 : nCompareChar = pCharData->mnUniChar;
355 34 : if ( c < nCompareChar )
356 : {
357 20 : if ( !nMid )
358 0 : break;
359 20 : nHigh = nMid-1;
360 : }
361 : else
362 : {
363 14 : if ( c > nCompareChar )
364 14 : nLow = nMid+1;
365 : else
366 0 : return pCharData->mnReplaceChar;
367 : }
368 : }
369 : while ( nLow <= nHigh );
370 :
371 4 : return 0;
372 : }
373 :
374 : struct ImplReplaceCharStrData
375 : {
376 : sal_uInt16 mnUniChar;
377 : sal_uInt16 maReplaceChars[IMPL_MAX_REPLACECHAR];
378 : };
379 :
380 : static ImplReplaceCharStrData const aImplRepCharStrTab[] =
381 : {
382 : { 0x00A9, { 0x0028, 0x0063, 0x0029, 0x0000, 0x0000 } }, /* COPYRIGHT SIGN */
383 : { 0x00AB, { 0x003C, 0x003C, 0x0000, 0x0000, 0x0000 } }, /* LEFT-POINTING-DOUBLE ANGLE QUOTATION MARK */
384 : { 0x0AE0, { 0x0028, 0x0072, 0x0029, 0x0000, 0x0000 } }, /* REGISTERED SIGN */
385 : { 0x00BB, { 0x003E, 0x003E, 0x0000, 0x0000, 0x0000 } }, /* RIGHT-POINTING-DOUBLE ANGLE QUOTATION MARK */
386 : { 0x00BC, { 0x0031, 0x002F, 0x0034, 0x0000, 0x0000 } }, /* VULGAR FRACTION ONE QUARTER */
387 : { 0x00BD, { 0x0031, 0x002F, 0x0032, 0x0000, 0x0000 } }, /* VULGAR FRACTION ONE HALF */
388 : { 0x00BE, { 0x0033, 0x002F, 0x0034, 0x0000, 0x0000 } }, /* VULGAR FRACTION THREE QUARTERS */
389 : { 0x00C6, { 0x0041, 0x0045, 0x0000, 0x0000, 0x0000 } }, /* LATIN CAPITAL LETTER AE */
390 : { 0x00E6, { 0x0061, 0x0065, 0x0000, 0x0000, 0x0000 } }, /* LATIN SMALL LETTER AE */
391 : { 0x0152, { 0x004F, 0x0045, 0x0000, 0x0000, 0x0000 } }, /* LATIN CAPITAL LIGATURE OE */
392 : { 0x0153, { 0x006F, 0x0065, 0x0000, 0x0000, 0x0000 } }, /* LATIN SMALL LIGATURE OE */
393 : { 0x2025, { 0x002E, 0x002E, 0x0000, 0x0000, 0x0000 } }, /* TWO DOT LEADER */
394 : { 0x2026, { 0x002E, 0x002E, 0x002E, 0x0000, 0x0000 } }, /* HORIZONTAL ELLIPSES */
395 : { 0x2034, { 0x0027, 0x0027, 0x0027, 0x0000, 0x0000 } }, /* TRIPPLE PRIME */
396 : { 0x2037, { 0x0027, 0x0027, 0x0027, 0x0000, 0x0000 } }, /* RESERVED TRIPPLE PRIME */
397 : { 0x20AC, { 0x0045, 0x0055, 0x0052, 0x0000, 0x0000 } }, /* EURO SIGN */
398 : { 0x2122, { 0x0028, 0x0074, 0x006D, 0x0029, 0x0000 } }, /* TRADE MARK SIGN */
399 : { 0x2153, { 0x0031, 0x002F, 0x0033, 0x0000, 0x0000 } }, /* VULGAR FRACTION ONE THIRD */
400 : { 0x2154, { 0x0032, 0x002F, 0x0033, 0x0000, 0x0000 } }, /* VULGAR FRACTION TWO THIRD */
401 : { 0x2155, { 0x0031, 0x002F, 0x0035, 0x0000, 0x0000 } }, /* VULGAR FRACTION ONE FIFTH */
402 : { 0x2156, { 0x0032, 0x002F, 0x0035, 0x0000, 0x0000 } }, /* VULGAR FRACTION TWO FIFTH */
403 : { 0x2157, { 0x0033, 0x002F, 0x0035, 0x0000, 0x0000 } }, /* VULGAR FRACTION THREE FIFTH */
404 : { 0x2158, { 0x0034, 0x002F, 0x0035, 0x0000, 0x0000 } }, /* VULGAR FRACTION FOUR FIFTH */
405 : { 0x2159, { 0x0031, 0x002F, 0x0036, 0x0000, 0x0000 } }, /* VULGAR FRACTION ONE SIXTH */
406 : { 0x215A, { 0x0035, 0x002F, 0x0036, 0x0000, 0x0000 } }, /* VULGAR FRACTION FIVE SIXTH */
407 : { 0x215B, { 0x0031, 0x002F, 0x0038, 0x0000, 0x0000 } }, /* VULGAR FRACTION ONE EIGHTH */
408 : { 0x215C, { 0x0033, 0x002F, 0x0038, 0x0000, 0x0000 } }, /* VULGAR FRACTION THREE EIGHTH */
409 : { 0x215D, { 0x0035, 0x002F, 0x0038, 0x0000, 0x0000 } }, /* VULGAR FRACTION FIVE EIGHTH */
410 : { 0x215E, { 0x0037, 0x002F, 0x0038, 0x0000, 0x0000 } }, /* VULGAR FRACTION SEVEN EIGHTH */
411 : { 0x215F, { 0x0031, 0x002F, 0x0000, 0x0000, 0x0000 } }, /* FRACTION NUMERATOR ONE */
412 : { 0x2161, { 0x0049, 0x0049, 0x0000, 0x0000, 0x0000 } }, /* ROMAN NUMERAL TWO */
413 : { 0x2162, { 0x0049, 0x0049, 0x0049, 0x0000, 0x0000 } }, /* ROMAN NUMERAL THREE */
414 : { 0x2163, { 0x0049, 0x0056, 0x0000, 0x0000, 0x0000 } }, /* ROMAN NUMERAL FOUR */
415 : { 0x2165, { 0x0056, 0x0049, 0x0000, 0x0000, 0x0000 } }, /* ROMAN NUMERAL SIX */
416 : { 0x2166, { 0x0056, 0x0049, 0x0049, 0x0000, 0x0000 } }, /* ROMAN NUMERAL SEVEN */
417 : { 0x2168, { 0x0056, 0x0049, 0x0049, 0x0049, 0x0000 } }, /* ROMAN NUMERAL EIGHT */
418 : { 0x2169, { 0x0049, 0x0058, 0x0000, 0x0000, 0x0000 } }, /* ROMAN NUMERAL NINE */
419 : { 0x216A, { 0x0058, 0x0049, 0x0000, 0x0000, 0x0000 } }, /* ROMAN NUMERAL ELEVEN */
420 : { 0x216B, { 0x0058, 0x0049, 0x0049, 0x0000, 0x0000 } }, /* ROMAN NUMERAL TWELVE */
421 : { 0x2171, { 0x0069, 0x0069, 0x0000, 0x0000, 0x0000 } }, /* SMALL ROMAN NUMERAL TWO */
422 : { 0x2172, { 0x0069, 0x0069, 0x0069, 0x0000, 0x0000 } }, /* SMALL ROMAN NUMERAL THREE */
423 : { 0x2173, { 0x0069, 0x0076, 0x0000, 0x0000, 0x0000 } }, /* SMALL ROMAN NUMERAL FOUR */
424 : { 0x2175, { 0x0076, 0x0069, 0x0000, 0x0000, 0x0000 } }, /* SMALL ROMAN NUMERAL SIX */
425 : { 0x2176, { 0x0076, 0x0069, 0x0069, 0x0000, 0x0000 } }, /* SMALL ROMAN NUMERAL SEVEN */
426 : { 0x2178, { 0x0076, 0x0069, 0x0069, 0x0069, 0x0000 } }, /* SMALL ROMAN NUMERAL EIGHT */
427 : { 0x2179, { 0x0069, 0x0078, 0x0000, 0x0000, 0x0000 } }, /* SMALL ROMAN NUMERAL NINE */
428 : { 0x217A, { 0x0078, 0x0069, 0x0000, 0x0000, 0x0000 } }, /* SMALL ROMAN NUMERAL ELEVEN */
429 : { 0x217B, { 0x0058, 0x0069, 0x0069, 0x0000, 0x0000 } } /* SMALL ROMAN NUMERAL TWELVE */
430 : };
431 :
432 0 : const sal_uInt16* ImplGetReplaceString( sal_Unicode c )
433 : {
434 : sal_uInt16 nLow;
435 : sal_uInt16 nHigh;
436 : sal_uInt16 nMid;
437 : sal_uInt16 nCompareChar;
438 : const ImplReplaceCharStrData* pCharData;
439 :
440 0 : nLow = 0;
441 0 : nHigh = (sizeof( aImplRepCharStrTab )/sizeof( ImplReplaceCharStrData ))-1;
442 0 : do
443 : {
444 0 : nMid = (nLow+nHigh)/2;
445 0 : pCharData = aImplRepCharStrTab+nMid;
446 0 : nCompareChar = pCharData->mnUniChar;
447 0 : if ( c < nCompareChar )
448 : {
449 0 : if ( !nMid )
450 0 : break;
451 0 : nHigh = nMid-1;
452 : }
453 : else
454 : {
455 0 : if ( c > nCompareChar )
456 0 : nLow = nMid+1;
457 : else
458 0 : return pCharData->maReplaceChars;
459 : }
460 : }
461 : while ( nLow <= nHigh );
462 :
463 0 : return 0;
464 : }
465 :
466 : // Writes 0--2 characters to dest:
467 257070 : static int ImplConvertUnicodeCharToChar(
468 : const ImplByteConvertData* pConvertData, sal_Unicode c, char * dest )
469 : {
470 : const ImplUniCharTabData* pToCharExTab;
471 :
472 257070 : if ( c < 0x80 )
473 : {
474 0 : dest[0] = static_cast< char >(c);
475 0 : return 1;
476 : }
477 257070 : if ( (c >= pConvertData->mnToCharStart1) && (c <= pConvertData->mnToCharEnd1) )
478 : {
479 2996 : dest[0] = static_cast< char >(pConvertData->mpToCharTab1[c-pConvertData->mnToCharStart1]);
480 3127 : if ( dest[0] != 0 )
481 2865 : return 1;
482 : }
483 254074 : else if ( (c >= pConvertData->mnToCharStart2) && (c <= pConvertData->mnToCharEnd2) )
484 : {
485 384 : dest[0] = static_cast< char >(pConvertData->mpToCharTab2[c-pConvertData->mnToCharStart2]);
486 384 : if ( dest[0] != 0 )
487 384 : return 1;
488 : }
489 253821 : pToCharExTab = pConvertData->mpToCharExTab;
490 253821 : if ( pToCharExTab )
491 : {
492 : sal_uInt16 nLow;
493 : sal_uInt16 nHigh;
494 : sal_uInt16 nMid;
495 : sal_uInt16 nCompareChar;
496 : const ImplUniCharTabData* pCharExData;
497 :
498 253815 : nLow = 0;
499 253815 : nHigh = pConvertData->mnToCharExCount-1;
500 1467543 : do
501 : {
502 1470262 : nMid = (nLow+nHigh)/2;
503 1470262 : pCharExData = pToCharExTab+nMid;
504 1470262 : nCompareChar = pCharExData->mnUniChar;
505 1470262 : if ( c < nCompareChar )
506 : {
507 186125 : if ( !nMid )
508 1111 : break;
509 185014 : nHigh = nMid-1;
510 : }
511 : else
512 : {
513 1284137 : if ( c > nCompareChar )
514 1282529 : nLow = nMid+1;
515 : else
516 : {
517 1608 : dest[0] = static_cast< char >(pCharExData->mnChar);
518 1608 : if ( pCharExData->mnChar2 == 0 )
519 1505 : return 1;
520 : else
521 : {
522 103 : dest[1] = static_cast< char >(pCharExData->mnChar2);
523 103 : return 2;
524 : }
525 : }
526 : }
527 : }
528 : while ( nLow <= nHigh );
529 : }
530 252213 : return 0;
531 : }
532 :
533 : }
534 :
535 221676 : sal_Size sal::detail::textenc::convertCharToUnicode(
536 : void const * pData, SAL_UNUSED_PARAMETER void *, char const * pSrcBuf,
537 : sal_Size nSrcBytes, sal_Unicode * pDestBuf, sal_Size nDestChars,
538 : sal_uInt32 nFlags, sal_uInt32 * pInfo, sal_Size * pSrcCvtBytes)
539 : {
540 : sal_uChar c;
541 : sal_Unicode cConv;
542 221676 : const ImplByteConvertData* pConvertData = (const ImplByteConvertData*)pData;
543 : sal_Unicode* pEndDestBuf;
544 : const char* pEndSrcBuf;
545 :
546 221676 : *pInfo = 0;
547 221676 : pEndDestBuf = pDestBuf+nDestChars;
548 221676 : pEndSrcBuf = pSrcBuf+nSrcBytes;
549 9662257 : while ( pSrcBuf < pEndSrcBuf )
550 : {
551 9219116 : c = (sal_uChar)*pSrcBuf;
552 9219116 : if ( c < 0x80 )
553 9208915 : cConv = c;
554 : else
555 : {
556 10201 : if ( (c >= pConvertData->mnToUniStart1) && (c <= pConvertData->mnToUniEnd1) )
557 9276 : cConv = pConvertData->mpToUniTab1[c-pConvertData->mnToUniStart1];
558 925 : else if ( (c >= pConvertData->mnToUniStart2) && (c <= pConvertData->mnToUniEnd2) )
559 899 : cConv = pConvertData->mpToUniTab2[c-pConvertData->mnToUniStart2];
560 : else
561 26 : cConv = 0;
562 10201 : if ( !cConv )
563 : {
564 405 : *pInfo |= RTL_TEXTTOUNICODE_INFO_UNDEFINED;
565 405 : if ( (nFlags & RTL_TEXTTOUNICODE_FLAGS_UNDEFINED_MASK) == RTL_TEXTTOUNICODE_FLAGS_UNDEFINED_ERROR )
566 : {
567 211 : *pInfo |= RTL_TEXTTOUNICODE_INFO_ERROR;
568 211 : break;
569 : }
570 194 : else if ( (nFlags & RTL_TEXTTOUNICODE_FLAGS_UNDEFINED_MASK) == RTL_TEXTTOUNICODE_FLAGS_UNDEFINED_IGNORE )
571 : {
572 0 : pSrcBuf++;
573 0 : continue;
574 : }
575 : else
576 194 : cConv = ImplGetUndefinedUnicodeChar(c, nFlags);
577 : }
578 : }
579 :
580 9218905 : if ( pDestBuf == pEndDestBuf )
581 : {
582 0 : *pInfo |= RTL_TEXTTOUNICODE_INFO_ERROR | RTL_TEXTTOUNICODE_INFO_DESTBUFFERTOSMALL;
583 0 : break;
584 : }
585 :
586 9218905 : *pDestBuf = cConv;
587 9218905 : pDestBuf++;
588 9218905 : pSrcBuf++;
589 : }
590 :
591 221676 : *pSrcCvtBytes = nSrcBytes - (pEndSrcBuf-pSrcBuf);
592 221676 : return (nDestChars - (pEndDestBuf-pDestBuf));
593 : }
594 :
595 523950 : sal_Size sal::detail::textenc::convertUnicodeToChar(
596 : void const * pData, SAL_UNUSED_PARAMETER void *,
597 : sal_Unicode const * pSrcBuf, sal_Size nSrcChars, char * pDestBuf,
598 : sal_Size nDestBytes, sal_uInt32 nFlags, sal_uInt32 * pInfo,
599 : sal_Size * pSrcCvtChars)
600 : {
601 : sal_Unicode c;
602 523950 : const ImplByteConvertData* pConvertData = (const ImplByteConvertData*)pData;
603 : char* pEndDestBuf;
604 : const sal_Unicode* pEndSrcBuf;
605 : int i;
606 : int n;
607 : sal_uInt16 cTemp;
608 : char aTempBuf[IMPL_MAX_REPLACECHAR+2];
609 : const sal_uInt16* pReplace;
610 :
611 523950 : *pInfo = 0;
612 523950 : pEndDestBuf = pDestBuf+nDestBytes;
613 523950 : pEndSrcBuf = pSrcBuf+nSrcChars;
614 15720932 : while ( pSrcBuf < pEndSrcBuf )
615 : {
616 14924921 : c = *pSrcBuf;
617 14924921 : if ( c < 0x80 )
618 : {
619 14667851 : aTempBuf[0] = static_cast< char >(c);
620 14667851 : n = 1;
621 : }
622 : else
623 : {
624 257070 : n = ImplConvertUnicodeCharToChar( pConvertData, c, aTempBuf );
625 :
626 257070 : if ( n == 0 )
627 : {
628 252213 : if ( nFlags & RTL_UNICODETOTEXT_FLAGS_UNDEFINED_REPLACE )
629 : {
630 4 : cTemp = ImplGetReplaceChar( c );
631 4 : if ( cTemp )
632 : n = ImplConvertUnicodeCharToChar(
633 0 : pConvertData, cTemp, aTempBuf );
634 : }
635 :
636 252213 : if ( n == 0 )
637 : {
638 252213 : if ( nFlags & RTL_UNICODETOTEXT_FLAGS_UNDEFINED_REPLACESTR )
639 : {
640 0 : pReplace = ImplGetReplaceString( c );
641 0 : if ( pReplace )
642 : {
643 0 : while ( *pReplace && (n < IMPL_MAX_REPLACECHAR) )
644 : {
645 : i = ImplConvertUnicodeCharToChar(
646 0 : pConvertData, *pReplace, aTempBuf + n );
647 0 : if ( i == 0 )
648 : {
649 0 : n = 0;
650 0 : break;
651 : }
652 0 : pReplace++;
653 0 : n += i;
654 : }
655 : }
656 : }
657 :
658 : /* Handle undefined and surrogates characters */
659 : /* (all surrogates characters are undefined) */
660 252213 : if ( n == 0 )
661 : {
662 252213 : if (sal::detail::textenc::handleUndefinedUnicodeToTextChar(
663 : &pSrcBuf, pEndSrcBuf, &pDestBuf, pEndDestBuf,
664 : nFlags, pInfo))
665 324 : continue;
666 : else
667 251889 : break;
668 : }
669 : }
670 : }
671 : }
672 :
673 14672708 : if ( pEndDestBuf - pDestBuf < n )
674 : {
675 0 : *pInfo |= RTL_UNICODETOTEXT_INFO_ERROR | RTL_UNICODETOTEXT_INFO_DESTBUFFERTOSMALL;
676 0 : break;
677 : }
678 :
679 29345519 : for ( i = 0; i < n; ++i )
680 14672811 : *pDestBuf++ = aTempBuf[i];
681 14672708 : pSrcBuf++;
682 : }
683 :
684 523950 : *pSrcCvtChars = nSrcChars - (pEndSrcBuf-pSrcBuf);
685 523950 : return (nDestBytes - (pEndDestBuf-pDestBuf));
686 : }
|
