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 #include "tenchelp.h" 29 #include "rtl/textcvt.h" 30 31 /* ======================================================================= */ 32 33 #define IMPL_MAX_REPLACECHAR 5 34 35 sal_uInt16 ImplGetReplaceChar(sal_Unicode c); 36 37 sal_uInt16 const * ImplGetReplaceString(sal_Unicode c); 38 39 /* ----------------------------------------------------------------------- */ 40 41 typedef struct 42 { 43 sal_uInt16 mnUniChar; 44 sal_uInt16 mnReplaceChar; 45 } ImplReplaceCharData; 46 47 static ImplReplaceCharData const aImplRepCharTab[] = 48 { 49 { 0x00A0, 0x0020 }, /* NO-BREAK-SPACE */ 50 { 0x00A1, 0x0021 }, /* INVERTED EXCLAMATION MARK */ 51 { 0x00B7, 0x0045 }, /* MIDDLE DOT */ 52 { 0x00BF, 0x003F }, /* INVERTED QUESTION MARK */ 53 { 0x00D7, 0x002A }, /* MULTIPLIKATION SIGN */ 54 { 0x00F7, 0x002F }, /* DIVISION SIGN */ 55 { 0x2000, 0x0020 }, /* EN QUAD */ 56 { 0x2001, 0x0020 }, /* EM QUAD */ 57 { 0x2002, 0x0020 }, /* EN SPACE */ 58 { 0x2003, 0x0020 }, /* EM SPACE */ 59 { 0x2004, 0x0020 }, /* THREE-PER-EM SPACE */ 60 { 0x2005, 0x0020 }, /* FOUR-PER-EM SPACE */ 61 { 0x2006, 0x0020 }, /* SIX-PER-EM SPACE */ 62 { 0x2007, 0x0020 }, /* FIGURE SPACE */ 63 { 0x2008, 0x0020 }, /* PUNCTATION SPACE */ 64 { 0x2009, 0x0020 }, /* THIN SPACE */ 65 { 0x200A, 0x0020 }, /* HAIR SPACE */ 66 { 0x2010, 0x002D }, /* HYPHEN */ 67 { 0x2011, 0x002D }, /* NON-BREAKING HYPHEN */ 68 { 0x2012, 0x002D }, /* FIGURE DASH */ 69 { 0x2013, 0x002D }, /* EN DASH */ 70 { 0x2014, 0x002D }, /* EM DASH */ 71 { 0x2015, 0x002D }, /* HORIZONTAL BAR */ 72 { 0x2018, 0x0027 }, /* LEFT SINGLE QUOTATION MARK */ 73 { 0x2019, 0x0027 }, /* RIGHT SINGLE QUOTATION MARK */ 74 { 0x201A, 0x002C }, /* SINGLE LOW-9 QUOTATION MARK */ 75 { 0x201B, 0x0027 }, /* SINGLE HIGH-RESERVED-9 QUOTATION MARK */ 76 { 0x201C, 0x0022 }, /* LEFT DOUBLE QUOTATION MARK */ 77 { 0x201D, 0x0022 }, /* RIGHT DOUBLE QUOTATION MARK */ 78 { 0x201E, 0x0022 }, /* DOUBLE LOW-9 QUOTATION MARK */ 79 { 0x201F, 0x0022 }, /* DOUBLE HIGH-RESERVED-9 QUOTATION MARK */ 80 { 0x2022, 0x002D }, /* BULLET */ 81 { 0x2023, 0x002D }, /* TRIANGULAR BULLET */ 82 { 0x2024, 0x002D }, /* ONE DOT LEADER */ 83 { 0x2027, 0x002D }, /* HYPHENATION POINT */ 84 { 0x2028, 0x000A }, /* LINE SEPARATOR */ 85 { 0x2029, 0x000D }, /* PARAGRAPH SEPARATOR */ 86 { 0x2032, 0x0027 }, /* PRIME */ 87 { 0x2033, 0x0022 }, /* DOUBLE PRIME */ 88 { 0x2035, 0x0027 }, /* RESERVED PRIME */ 89 { 0x2036, 0x0022 }, /* RESERVED DOUBLE PRIME */ 90 { 0x2039, 0x003C }, /* SINGLE LEFT-POINTING ANGLE QUOTATION MARK */ 91 { 0x203A, 0x003E }, /* SINGLE RIGHT-POINTING ANGLE QUOTATION MARK */ 92 { 0x2043, 0x002D }, /* HYPHEN BULLET */ 93 { 0x2044, 0x002F }, /* FRACTION SLASH */ 94 { 0x2160, 0x0049 }, /* ROMAN NUMERAL ONE */ 95 { 0x2164, 0x0056 }, /* ROMAN NUMERAL FIVE */ 96 { 0x2169, 0x0058 }, /* ROMAN NUMERAL TEN */ 97 { 0x216C, 0x004C }, /* ROMAN NUMERAL FIFTY */ 98 { 0x216D, 0x0043 }, /* ROMAN NUMERAL ONE HUNDRED */ 99 { 0x216E, 0x0044 }, /* ROMAN NUMERAL FIVE HUNDRED */ 100 { 0x216F, 0x004D }, /* ROMAN NUMERAL ONE THOUSAND */ 101 { 0x2170, 0x0069 }, /* SMALL ROMAN NUMERAL ONE */ 102 { 0x2174, 0x0076 }, /* SMALL ROMAN NUMERAL FIVE */ 103 { 0x2179, 0x0078 }, /* SMALL ROMAN NUMERAL TEN */ 104 { 0x217C, 0x006C }, /* SMALL ROMAN NUMERAL FIFTY */ 105 { 0x217D, 0x0063 }, /* SMALL ROMAN NUMERAL ONE HUNDRED */ 106 { 0x217E, 0x0064 }, /* SMALL ROMAN NUMERAL FIVE HUNDRED */ 107 { 0x217F, 0x006D }, /* SMALL ROMAN NUMERAL ONE THOUSAND */ 108 { 0x2215, 0x002F }, /* DIVISION SLASH */ 109 { 0x2217, 0x002A }, /* ASTERIX OPERATOR */ 110 { 0xFF00, 0x0020 }, /* FULLWIDTH ASCII FORMS */ 111 { 0xFF01, 0x0021 }, /* FULLWIDTH ASCII FORMS */ 112 { 0xFF02, 0x0022 }, /* FULLWIDTH ASCII FORMS*/ 113 { 0xFF03, 0x0023 }, /* FULLWIDTH ASCII FORMS */ 114 { 0xFF04, 0x0024 }, /* FULLWIDTH ASCII FORMS*/ 115 { 0xFF05, 0x0025 }, /* FULLWIDTH ASCII FORMS */ 116 { 0xFF06, 0x0026 }, /* FULLWIDTH ASCII FORMS*/ 117 { 0xFF07, 0x0027 }, /* FULLWIDTH ASCII FORMS */ 118 { 0xFF08, 0x0028 }, /* FULLWIDTH ASCII FORMS*/ 119 { 0xFF09, 0x0029 }, /* FULLWIDTH ASCII FORMS */ 120 { 0xFF0A, 0x002A }, /* FULLWIDTH ASCII FORMS*/ 121 { 0xFF0B, 0x002B }, /* FULLWIDTH ASCII FORMS */ 122 { 0xFF0C, 0x002C }, /* FULLWIDTH ASCII FORMS*/ 123 { 0xFF0D, 0x002D }, /* FULLWIDTH ASCII FORMS */ 124 { 0xFF0E, 0x002E }, /* FULLWIDTH ASCII FORMS*/ 125 { 0xFF0F, 0x002F }, /* FULLWIDTH ASCII FORMS */ 126 { 0xFF10, 0x0030 }, /* FULLWIDTH ASCII FORMS */ 127 { 0xFF11, 0x0031 }, /* FULLWIDTH ASCII FORMS */ 128 { 0xFF12, 0x0032 }, /* FULLWIDTH ASCII FORMS*/ 129 { 0xFF13, 0x0033 }, /* FULLWIDTH ASCII FORMS */ 130 { 0xFF14, 0x0034 }, /* FULLWIDTH ASCII FORMS*/ 131 { 0xFF15, 0x0035 }, /* FULLWIDTH ASCII FORMS */ 132 { 0xFF16, 0x0036 }, /* FULLWIDTH ASCII FORMS*/ 133 { 0xFF17, 0x0037 }, /* FULLWIDTH ASCII FORMS */ 134 { 0xFF18, 0x0038 }, /* FULLWIDTH ASCII FORMS*/ 135 { 0xFF19, 0x0039 }, /* FULLWIDTH ASCII FORMS */ 136 { 0xFF1A, 0x003A }, /* FULLWIDTH ASCII FORMS*/ 137 { 0xFF1B, 0x003B }, /* FULLWIDTH ASCII FORMS */ 138 { 0xFF1C, 0x003C }, /* FULLWIDTH ASCII FORMS*/ 139 { 0xFF1D, 0x003D }, /* FULLWIDTH ASCII FORMS */ 140 { 0xFF1E, 0x003E }, /* FULLWIDTH ASCII FORMS*/ 141 { 0xFF1F, 0x003F }, /* FULLWIDTH ASCII FORMS */ 142 { 0xFF20, 0x0040 }, /* FULLWIDTH ASCII FORMS */ 143 { 0xFF21, 0x0041 }, /* FULLWIDTH ASCII FORMS */ 144 { 0xFF22, 0x0042 }, /* FULLWIDTH ASCII FORMS*/ 145 { 0xFF23, 0x0043 }, /* FULLWIDTH ASCII FORMS */ 146 { 0xFF24, 0x0044 }, /* FULLWIDTH ASCII FORMS*/ 147 { 0xFF25, 0x0045 }, /* FULLWIDTH ASCII FORMS */ 148 { 0xFF26, 0x0046 }, /* FULLWIDTH ASCII FORMS*/ 149 { 0xFF27, 0x0047 }, /* FULLWIDTH ASCII FORMS */ 150 { 0xFF28, 0x0048 }, /* FULLWIDTH ASCII FORMS*/ 151 { 0xFF29, 0x0049 }, /* FULLWIDTH ASCII FORMS */ 152 { 0xFF2A, 0x004A }, /* FULLWIDTH ASCII FORMS*/ 153 { 0xFF2B, 0x004B }, /* FULLWIDTH ASCII FORMS */ 154 { 0xFF2C, 0x004C }, /* FULLWIDTH ASCII FORMS*/ 155 { 0xFF2D, 0x004D }, /* FULLWIDTH ASCII FORMS */ 156 { 0xFF2E, 0x004E }, /* FULLWIDTH ASCII FORMS*/ 157 { 0xFF2F, 0x004F }, /* FULLWIDTH ASCII FORMS */ 158 { 0xFF30, 0x0050 }, /* FULLWIDTH ASCII FORMS */ 159 { 0xFF31, 0x0051 }, /* FULLWIDTH ASCII FORMS */ 160 { 0xFF32, 0x0052 }, /* FULLWIDTH ASCII FORMS*/ 161 { 0xFF33, 0x0053 }, /* FULLWIDTH ASCII FORMS */ 162 { 0xFF34, 0x0054 }, /* FULLWIDTH ASCII FORMS*/ 163 { 0xFF35, 0x0055 }, /* FULLWIDTH ASCII FORMS */ 164 { 0xFF36, 0x0056 }, /* FULLWIDTH ASCII FORMS*/ 165 { 0xFF37, 0x0057 }, /* FULLWIDTH ASCII FORMS */ 166 { 0xFF38, 0x0058 }, /* FULLWIDTH ASCII FORMS*/ 167 { 0xFF39, 0x0059 }, /* FULLWIDTH ASCII FORMS */ 168 { 0xFF3A, 0x005A }, /* FULLWIDTH ASCII FORMS*/ 169 { 0xFF3B, 0x005B }, /* FULLWIDTH ASCII FORMS */ 170 { 0xFF3C, 0x005C }, /* FULLWIDTH ASCII FORMS*/ 171 { 0xFF3D, 0x005D }, /* FULLWIDTH ASCII FORMS */ 172 { 0xFF3E, 0x005E }, /* FULLWIDTH ASCII FORMS*/ 173 { 0xFF3F, 0x005F }, /* FULLWIDTH ASCII FORMS */ 174 { 0xFF40, 0x0060 }, /* FULLWIDTH ASCII FORMS */ 175 { 0xFF41, 0x0061 }, /* FULLWIDTH ASCII FORMS */ 176 { 0xFF42, 0x0062 }, /* FULLWIDTH ASCII FORMS*/ 177 { 0xFF43, 0x0063 }, /* FULLWIDTH ASCII FORMS */ 178 { 0xFF44, 0x0064 }, /* FULLWIDTH ASCII FORMS*/ 179 { 0xFF45, 0x0065 }, /* FULLWIDTH ASCII FORMS */ 180 { 0xFF46, 0x0066 }, /* FULLWIDTH ASCII FORMS*/ 181 { 0xFF47, 0x0067 }, /* FULLWIDTH ASCII FORMS */ 182 { 0xFF48, 0x0068 }, /* FULLWIDTH ASCII FORMS*/ 183 { 0xFF49, 0x0069 }, /* FULLWIDTH ASCII FORMS */ 184 { 0xFF4A, 0x006A }, /* FULLWIDTH ASCII FORMS*/ 185 { 0xFF4B, 0x006B }, /* FULLWIDTH ASCII FORMS */ 186 { 0xFF4C, 0x006C }, /* FULLWIDTH ASCII FORMS*/ 187 { 0xFF4D, 0x006D }, /* FULLWIDTH ASCII FORMS */ 188 { 0xFF4E, 0x006E }, /* FULLWIDTH ASCII FORMS*/ 189 { 0xFF4F, 0x006F }, /* FULLWIDTH ASCII FORMS */ 190 { 0xFF50, 0x0070 }, /* FULLWIDTH ASCII FORMS */ 191 { 0xFF51, 0x0071 }, /* FULLWIDTH ASCII FORMS */ 192 { 0xFF52, 0x0072 }, /* FULLWIDTH ASCII FORMS*/ 193 { 0xFF53, 0x0073 }, /* FULLWIDTH ASCII FORMS */ 194 { 0xFF54, 0x0074 }, /* FULLWIDTH ASCII FORMS*/ 195 { 0xFF55, 0x0075 }, /* FULLWIDTH ASCII FORMS */ 196 { 0xFF56, 0x0076 }, /* FULLWIDTH ASCII FORMS*/ 197 { 0xFF57, 0x0077 }, /* FULLWIDTH ASCII FORMS */ 198 { 0xFF58, 0x0078 }, /* FULLWIDTH ASCII FORMS*/ 199 { 0xFF59, 0x0079 }, /* FULLWIDTH ASCII FORMS */ 200 { 0xFF5A, 0x007A }, /* FULLWIDTH ASCII FORMS*/ 201 { 0xFF5B, 0x007B }, /* FULLWIDTH ASCII FORMS */ 202 { 0xFF5C, 0x007C }, /* FULLWIDTH ASCII FORMS*/ 203 { 0xFF5D, 0x007D }, /* FULLWIDTH ASCII FORMS */ 204 { 0xFF5E, 0x007E }, /* FULLWIDTH ASCII FORMS*/ 205 { 0xFF5F, 0x007F }, /* FULLWIDTH ASCII FORMS */ 206 { 0xFF61, 0x3002 }, /* HALFWIDTH KATAKANA FORMS */ 207 { 0xFF62, 0x300C }, /* HALFWIDTH KATAKANA FORMS */ 208 { 0xFF63, 0x300D }, /* HALFWIDTH KATAKANA FORMS */ 209 { 0xFF64, 0x3001 }, /* HALFWIDTH KATAKANA FORMS */ 210 { 0xFF65, 0x30FB }, /* HALFWIDTH KATAKANA FORMS */ 211 { 0xFF66, 0x30F2 }, /* HALFWIDTH KATAKANA FORMS */ 212 { 0xFF67, 0x30A1 }, /* HALFWIDTH KATAKANA FORMS */ 213 { 0xFF68, 0x30A3 }, /* HALFWIDTH KATAKANA FORMS */ 214 { 0xFF69, 0x30A5 }, /* HALFWIDTH KATAKANA FORMS */ 215 { 0xFF6A, 0x30A7 }, /* HALFWIDTH KATAKANA FORMS */ 216 { 0xFF6B, 0x30A9 }, /* HALFWIDTH KATAKANA FORMS */ 217 { 0xFF6C, 0x30E3 }, /* HALFWIDTH KATAKANA FORMS */ 218 { 0xFF6D, 0x30E5 }, /* HALFWIDTH KATAKANA FORMS */ 219 { 0xFF6E, 0x30E7 }, /* HALFWIDTH KATAKANA FORMS */ 220 { 0xFF6F, 0x30C3 }, /* HALFWIDTH KATAKANA FORMS */ 221 { 0xFF70, 0x30FC }, /* HALFWIDTH KATAKANA FORMS */ 222 { 0xFF71, 0x30A2 }, /* HALFWIDTH KATAKANA FORMS */ 223 { 0xFF72, 0x30A4 }, /* HALFWIDTH KATAKANA FORMS */ 224 { 0xFF73, 0x30A6 }, /* HALFWIDTH KATAKANA FORMS */ 225 { 0xFF74, 0x30A8 }, /* HALFWIDTH KATAKANA FORMS */ 226 { 0xFF75, 0x30AA }, /* HALFWIDTH KATAKANA FORMS */ 227 { 0xFF76, 0x30AB }, /* HALFWIDTH KATAKANA FORMS */ 228 { 0xFF77, 0x30AD }, /* HALFWIDTH KATAKANA FORMS */ 229 { 0xFF78, 0x30AF }, /* HALFWIDTH KATAKANA FORMS */ 230 { 0xFF79, 0x30B1 }, /* HALFWIDTH KATAKANA FORMS */ 231 { 0xFF7A, 0x30B3 }, /* HALFWIDTH KATAKANA FORMS */ 232 { 0xFF7B, 0x30B5 }, /* HALFWIDTH KATAKANA FORMS */ 233 { 0xFF7C, 0x30B7 }, /* HALFWIDTH KATAKANA FORMS */ 234 { 0xFF7D, 0x30B9 }, /* HALFWIDTH KATAKANA FORMS */ 235 { 0xFF7E, 0x30BB }, /* HALFWIDTH KATAKANA FORMS */ 236 { 0xFF7F, 0x30BD }, /* HALFWIDTH KATAKANA FORMS */ 237 { 0xFF80, 0x30BF }, /* HALFWIDTH KATAKANA FORMS */ 238 { 0xFF81, 0x30C1 }, /* HALFWIDTH KATAKANA FORMS */ 239 { 0xFF82, 0x30C4 }, /* HALFWIDTH KATAKANA FORMS */ 240 { 0xFF83, 0x30C6 }, /* HALFWIDTH KATAKANA FORMS */ 241 { 0xFF84, 0x30C8 }, /* HALFWIDTH KATAKANA FORMS */ 242 { 0xFF85, 0x30CA }, /* HALFWIDTH KATAKANA FORMS */ 243 { 0xFF86, 0x30CB }, /* HALFWIDTH KATAKANA FORMS */ 244 { 0xFF87, 0x30CC }, /* HALFWIDTH KATAKANA FORMS */ 245 { 0xFF88, 0x30CD }, /* HALFWIDTH KATAKANA FORMS */ 246 { 0xFF89, 0x30CE }, /* HALFWIDTH KATAKANA FORMS */ 247 { 0xFF8A, 0x30CF }, /* HALFWIDTH KATAKANA FORMS */ 248 { 0xFF8B, 0x30D2 }, /* HALFWIDTH KATAKANA FORMS */ 249 { 0xFF8C, 0x30D5 }, /* HALFWIDTH KATAKANA FORMS */ 250 { 0xFF8D, 0x30D8 }, /* HALFWIDTH KATAKANA FORMS */ 251 { 0xFF8E, 0x30DB }, /* HALFWIDTH KATAKANA FORMS */ 252 { 0xFF8F, 0x30DE }, /* HALFWIDTH KATAKANA FORMS */ 253 { 0xFF90, 0x30DF }, /* HALFWIDTH KATAKANA FORMS */ 254 { 0xFF91, 0x30E0 }, /* HALFWIDTH KATAKANA FORMS */ 255 { 0xFF92, 0x30E1 }, /* HALFWIDTH KATAKANA FORMS */ 256 { 0xFF93, 0x30E2 }, /* HALFWIDTH KATAKANA FORMS */ 257 { 0xFF94, 0x30E4 }, /* HALFWIDTH KATAKANA FORMS */ 258 { 0xFF95, 0x30E6 }, /* HALFWIDTH KATAKANA FORMS */ 259 { 0xFF96, 0x30E8 }, /* HALFWIDTH KATAKANA FORMS */ 260 { 0xFF97, 0x30E9 }, /* HALFWIDTH KATAKANA FORMS */ 261 { 0xFF98, 0x30EA }, /* HALFWIDTH KATAKANA FORMS */ 262 { 0xFF99, 0x30EB }, /* HALFWIDTH KATAKANA FORMS */ 263 { 0xFF9A, 0x30EC }, /* HALFWIDTH KATAKANA FORMS */ 264 { 0xFF9B, 0x30ED }, /* HALFWIDTH KATAKANA FORMS */ 265 { 0xFF9C, 0x30EF }, /* HALFWIDTH KATAKANA FORMS */ 266 { 0xFF9D, 0x30F3 }, /* HALFWIDTH KATAKANA FORMS */ 267 { 0xFF9E, 0x309B }, /* HALFWIDTH KATAKANA FORMS */ 268 { 0xFF9F, 0x309C }, /* HALFWIDTH KATAKANA FORMS */ 269 { 0xFFA0, 0x3164 }, /* HALFWIDTH HANGUL FORMS */ 270 { 0xFFA1, 0x3131 }, /* HALFWIDTH HANGUL FORMS */ 271 { 0xFFA2, 0x3132 }, /* HALFWIDTH HANGUL FORMS */ 272 { 0xFFA3, 0x3133 }, /* HALFWIDTH HANGUL FORMS */ 273 { 0xFFA4, 0x3134 }, /* HALFWIDTH HANGUL FORMS */ 274 { 0xFFA5, 0x3135 }, /* HALFWIDTH HANGUL FORMS */ 275 { 0xFFA6, 0x3136 }, /* HALFWIDTH HANGUL FORMS */ 276 { 0xFFA7, 0x3137 }, /* HALFWIDTH HANGUL FORMS */ 277 { 0xFFA8, 0x3138 }, /* HALFWIDTH HANGUL FORMS */ 278 { 0xFFA9, 0x3139 }, /* HALFWIDTH HANGUL FORMS */ 279 { 0xFFAA, 0x313A }, /* HALFWIDTH HANGUL FORMS */ 280 { 0xFFAB, 0x313B }, /* HALFWIDTH HANGUL FORMS */ 281 { 0xFFAC, 0x313C }, /* HALFWIDTH HANGUL FORMS */ 282 { 0xFFAD, 0x313D }, /* HALFWIDTH HANGUL FORMS */ 283 { 0xFFAE, 0x313E }, /* HALFWIDTH HANGUL FORMS */ 284 { 0xFFAF, 0x313F }, /* HALFWIDTH HANGUL FORMS */ 285 { 0xFFB0, 0x3140 }, /* HALFWIDTH HANGUL FORMS */ 286 { 0xFFB1, 0x3141 }, /* HALFWIDTH HANGUL FORMS */ 287 { 0xFFB2, 0x3142 }, /* HALFWIDTH HANGUL FORMS */ 288 { 0xFFB3, 0x3143 }, /* HALFWIDTH HANGUL FORMS */ 289 { 0xFFB4, 0x3144 }, /* HALFWIDTH HANGUL FORMS */ 290 { 0xFFB5, 0x3145 }, /* HALFWIDTH HANGUL FORMS */ 291 { 0xFFB6, 0x3146 }, /* HALFWIDTH HANGUL FORMS */ 292 { 0xFFB7, 0x3147 }, /* HALFWIDTH HANGUL FORMS */ 293 { 0xFFB8, 0x3148 }, /* HALFWIDTH HANGUL FORMS */ 294 { 0xFFB9, 0x3149 }, /* HALFWIDTH HANGUL FORMS */ 295 { 0xFFBA, 0x314A }, /* HALFWIDTH HANGUL FORMS */ 296 { 0xFFBB, 0x314B }, /* HALFWIDTH HANGUL FORMS */ 297 { 0xFFBC, 0x314C }, /* HALFWIDTH HANGUL FORMS */ 298 { 0xFFBD, 0x314D }, /* HALFWIDTH HANGUL FORMS */ 299 { 0xFFBE, 0x314E }, /* HALFWIDTH HANGUL FORMS */ 300 { 0xFFC2, 0x314F }, /* HALFWIDTH HANGUL FORMS */ 301 { 0xFFC3, 0x3150 }, /* HALFWIDTH HANGUL FORMS */ 302 { 0xFFC4, 0x3151 }, /* HALFWIDTH HANGUL FORMS */ 303 { 0xFFC5, 0x3152 }, /* HALFWIDTH HANGUL FORMS */ 304 { 0xFFC6, 0x3153 }, /* HALFWIDTH HANGUL FORMS */ 305 { 0xFFC7, 0x3154 }, /* HALFWIDTH HANGUL FORMS */ 306 { 0xFFCA, 0x3155 }, /* HALFWIDTH HANGUL FORMS */ 307 { 0xFFCB, 0x3156 }, /* HALFWIDTH HANGUL FORMS */ 308 { 0xFFCC, 0x3157 }, /* HALFWIDTH HANGUL FORMS */ 309 { 0xFFCD, 0x3158 }, /* HALFWIDTH HANGUL FORMS */ 310 { 0xFFCE, 0x3159 }, /* HALFWIDTH HANGUL FORMS */ 311 { 0xFFCF, 0x315A }, /* HALFWIDTH HANGUL FORMS */ 312 { 0xFFD2, 0x315B }, /* HALFWIDTH HANGUL FORMS */ 313 { 0xFFD3, 0x315C }, /* HALFWIDTH HANGUL FORMS */ 314 { 0xFFD4, 0x315D }, /* HALFWIDTH HANGUL FORMS */ 315 { 0xFFD5, 0x315E }, /* HALFWIDTH HANGUL FORMS */ 316 { 0xFFD6, 0x315F }, /* HALFWIDTH HANGUL FORMS */ 317 { 0xFFD7, 0x3160 }, /* HALFWIDTH HANGUL FORMS */ 318 { 0xFFDA, 0x3161 }, /* HALFWIDTH HANGUL FORMS */ 319 { 0xFFDB, 0x3162 }, /* HALFWIDTH HANGUL FORMS */ 320 { 0xFFDC, 0x3163 }, /* HALFWIDTH HANGUL FORMS */ 321 { 0xFFE0, 0x00A2 }, /* FULLWIDTH CENT SIGN */ 322 { 0xFFE1, 0x00A3 }, /* FULLWIDTH POUND SIGN */ 323 { 0xFFE2, 0x00AC }, /* FULLWIDTH NOT SIGN */ 324 { 0xFFE3, 0x00AF }, /* FULLWIDTH MACRON */ 325 { 0xFFE4, 0x00A6 }, /* FULLWIDTH BROKEN BAR */ 326 { 0xFFE5, 0x00A5 }, /* FULLWIDTH YEN SIGN */ 327 { 0xFFE6, 0x20A9 }, /* FULLWIDTH WON SIGN */ 328 { 0xFFE8, 0x2502 }, /* HALFWIDTH FORMS LIGHT VERTICAL */ 329 { 0xFFE9, 0x2190 }, /* HALFWIDTH LEFTWARDS ARROW */ 330 { 0xFFEA, 0x2191 }, /* HALFWIDTH UPWARDS ARROW */ 331 { 0xFFEB, 0x2192 }, /* HALFWIDTH RIGHTWARDS ARROW */ 332 { 0xFFEC, 0x2193 }, /* HALFWIDTH DOWNWARDS ARROW */ 333 { 0xFFED, 0x25A0 }, /* HALFWIDTH BLACK SQUARE */ 334 { 0xFFEE, 0x25CB }, /* HALFWIDTH WHITE CIRCLE */ 335 { 0xFFFD, 0x003F } /* REPLACEMENT CHARACTER */ 336 }; 337 338 sal_uInt16 ImplGetReplaceChar( sal_Unicode c ) 339 { 340 sal_uInt16 nLow; 341 sal_uInt16 nHigh; 342 sal_uInt16 nMid; 343 sal_uInt16 nCompareChar; 344 const ImplReplaceCharData* pCharData; 345 346 nLow = 0; 347 nHigh = (sizeof( aImplRepCharTab )/sizeof( ImplReplaceCharData ))-1; 348 do 349 { 350 nMid = (nLow+nHigh)/2; 351 pCharData = aImplRepCharTab+nMid; 352 nCompareChar = pCharData->mnUniChar; 353 if ( c < nCompareChar ) 354 { 355 if ( !nMid ) 356 break; 357 nHigh = nMid-1; 358 } 359 else 360 { 361 if ( c > nCompareChar ) 362 nLow = nMid+1; 363 else 364 return pCharData->mnReplaceChar; 365 } 366 } 367 while ( nLow <= nHigh ); 368 369 return 0; 370 } 371 372 /* ----------------------------------------------------------------------- */ 373 374 typedef struct 375 { 376 sal_uInt16 mnUniChar; 377 sal_uInt16 maReplaceChars[IMPL_MAX_REPLACECHAR]; 378 } ImplReplaceCharStrData; 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 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 nLow = 0; 441 nHigh = (sizeof( aImplRepCharStrTab )/sizeof( ImplReplaceCharStrData ))-1; 442 do 443 { 444 nMid = (nLow+nHigh)/2; 445 pCharData = aImplRepCharStrTab+nMid; 446 nCompareChar = pCharData->mnUniChar; 447 if ( c < nCompareChar ) 448 { 449 if ( !nMid ) 450 break; 451 nHigh = nMid-1; 452 } 453 else 454 { 455 if ( c > nCompareChar ) 456 nLow = nMid+1; 457 else 458 return pCharData->maReplaceChars; 459 } 460 } 461 while ( nLow <= nHigh ); 462 463 return 0; 464 } 465 466 /* ======================================================================= */ 467 468 sal_Size ImplSymbolToUnicode( const ImplTextConverterData* pData, 469 void* pContext, 470 const sal_Char* pSrcBuf, sal_Size nSrcBytes, 471 sal_Unicode* pDestBuf, sal_Size nDestChars, 472 sal_uInt32 nFlags, sal_uInt32* pInfo, 473 sal_Size* pSrcCvtBytes ) 474 { 475 sal_uChar c; 476 sal_Unicode* pEndDestBuf; 477 const sal_Char* pEndSrcBuf; 478 479 (void) pData; /* unused */ 480 (void) pContext; /* unused */ 481 (void) nFlags; /* unused */ 482 483 *pInfo = 0; 484 pEndDestBuf = pDestBuf+nDestChars; 485 pEndSrcBuf = pSrcBuf+nSrcBytes; 486 while ( pSrcBuf < pEndSrcBuf ) 487 { 488 if ( pDestBuf == pEndDestBuf ) 489 { 490 *pInfo |= RTL_TEXTTOUNICODE_INFO_ERROR | RTL_TEXTTOUNICODE_INFO_DESTBUFFERTOSMALL; 491 break; 492 } 493 494 /* 0-31 (all Control-Character get the same Unicode value) */ 495 c = (sal_uChar)*pSrcBuf; 496 if ( c <= 0x1F ) 497 *pDestBuf = (sal_Unicode)c; 498 else 499 *pDestBuf = ((sal_Unicode)c)+0xF000; 500 pDestBuf++; 501 pSrcBuf++; 502 } 503 504 *pSrcCvtBytes = nSrcBytes - (pEndSrcBuf-pSrcBuf); 505 return (nDestChars - (pEndDestBuf-pDestBuf)); 506 } 507 508 /* ----------------------------------------------------------------------- */ 509 510 sal_Size ImplUnicodeToSymbol( const ImplTextConverterData* pData, 511 void* pContext, 512 const sal_Unicode* pSrcBuf, sal_Size nSrcChars, 513 sal_Char* pDestBuf, sal_Size nDestBytes, 514 sal_uInt32 nFlags, sal_uInt32* pInfo, 515 sal_Size* pSrcCvtChars ) 516 { 517 sal_Unicode c; 518 sal_Char* pEndDestBuf; 519 const sal_Unicode* pEndSrcBuf; 520 521 (void) pContext; /* unused */ 522 523 *pInfo = 0; 524 pEndDestBuf = pDestBuf+nDestBytes; 525 pEndSrcBuf = pSrcBuf+nSrcChars; 526 while ( pSrcBuf < pEndSrcBuf ) 527 { 528 if ( pDestBuf == pEndDestBuf ) 529 { 530 *pInfo |= RTL_UNICODETOTEXT_INFO_ERROR | RTL_UNICODETOTEXT_INFO_DESTBUFFERTOSMALL; 531 break; 532 } 533 534 c = *pSrcBuf; 535 if ( (c >= 0xF000) && (c <= 0xF0FF) ) 536 { 537 *pDestBuf = (sal_Char)(sal_uChar)(c-0xF000); 538 pDestBuf++; 539 pSrcBuf++; 540 } 541 // Normally 0x001F, but in many cases also symbol characters 542 // are stored in the first 256 bytes, so that we don't change 543 // these values 544 else if ( c <= 0x00FF ) 545 { 546 *pDestBuf = (sal_Char)(sal_uChar)c; 547 pDestBuf++; 548 pSrcBuf++; 549 } 550 else 551 { 552 if ( nFlags & RTL_UNICODETOTEXT_FLAGS_UNDEFINED_REPLACE ) 553 { 554 /* !!! */ 555 /* Only ascii characters < 0x1F */ 556 } 557 558 /* Handle undefined and surrogates characters */ 559 /* (all surrogates characters are undefined) */ 560 if (!ImplHandleUndefinedUnicodeToTextChar(pData, 561 &pSrcBuf, 562 pEndSrcBuf, 563 &pDestBuf, 564 pEndDestBuf, 565 nFlags, 566 pInfo)) 567 break; 568 } 569 } 570 571 *pSrcCvtChars = nSrcChars - (pEndSrcBuf-pSrcBuf); 572 return (nDestBytes - (pEndDestBuf-pDestBuf)); 573 } 574 575 /* ======================================================================= */ 576 577 sal_Size ImplCharToUnicode( const ImplTextConverterData* pData, 578 void* pContext, 579 const sal_Char* pSrcBuf, sal_Size nSrcBytes, 580 sal_Unicode* pDestBuf, sal_Size nDestChars, 581 sal_uInt32 nFlags, sal_uInt32* pInfo, 582 sal_Size* pSrcCvtBytes ) 583 { 584 sal_uChar c; 585 sal_Unicode cConv; 586 const ImplByteConvertData* pConvertData = (const ImplByteConvertData*)pData; 587 sal_Unicode* pEndDestBuf; 588 const sal_Char* pEndSrcBuf; 589 590 (void) pContext; /* unused */ 591 592 *pInfo = 0; 593 pEndDestBuf = pDestBuf+nDestChars; 594 pEndSrcBuf = pSrcBuf+nSrcBytes; 595 while ( pSrcBuf < pEndSrcBuf ) 596 { 597 c = (sal_uChar)*pSrcBuf; 598 if ( c < 0x80 ) 599 cConv = c; 600 else 601 { 602 if ( (c >= pConvertData->mnToUniStart1) && (c <= pConvertData->mnToUniEnd1) ) 603 cConv = pConvertData->mpToUniTab1[c-pConvertData->mnToUniStart1]; 604 else if ( (c >= pConvertData->mnToUniStart2) && (c <= pConvertData->mnToUniEnd2) ) 605 cConv = pConvertData->mpToUniTab2[c-pConvertData->mnToUniStart2]; 606 else 607 cConv = 0; 608 if ( !cConv ) 609 { 610 *pInfo |= RTL_TEXTTOUNICODE_INFO_UNDEFINED; 611 if ( (nFlags & RTL_TEXTTOUNICODE_FLAGS_UNDEFINED_MASK) == RTL_TEXTTOUNICODE_FLAGS_UNDEFINED_ERROR ) 612 { 613 *pInfo |= RTL_TEXTTOUNICODE_INFO_ERROR; 614 break; 615 } 616 else if ( (nFlags & RTL_TEXTTOUNICODE_FLAGS_UNDEFINED_MASK) == RTL_TEXTTOUNICODE_FLAGS_UNDEFINED_IGNORE ) 617 { 618 pSrcBuf++; 619 continue; 620 } 621 else 622 cConv = ImplGetUndefinedUnicodeChar(c, nFlags); 623 } 624 } 625 626 if ( pDestBuf == pEndDestBuf ) 627 { 628 *pInfo |= RTL_TEXTTOUNICODE_INFO_ERROR | RTL_TEXTTOUNICODE_INFO_DESTBUFFERTOSMALL; 629 break; 630 } 631 632 *pDestBuf = cConv; 633 pDestBuf++; 634 pSrcBuf++; 635 } 636 637 *pSrcCvtBytes = nSrcBytes - (pEndSrcBuf-pSrcBuf); 638 return (nDestChars - (pEndDestBuf-pDestBuf)); 639 } 640 641 /* ----------------------------------------------------------------------- */ 642 643 // Writes 0--2 characters to dest: 644 static int ImplConvertUnicodeCharToChar( 645 const ImplByteConvertData* pConvertData, sal_Unicode c, sal_Char * dest ) 646 { 647 const ImplUniCharTabData* pToCharExTab; 648 649 if ( c < 0x80 ) 650 { 651 dest[0] = (sal_Char)c; 652 return 1; 653 } 654 if ( (c >= pConvertData->mnToCharStart1) && (c <= pConvertData->mnToCharEnd1) ) 655 { 656 dest[0] = (sal_Char)pConvertData->mpToCharTab1[c-pConvertData->mnToCharStart1]; 657 if ( dest[0] != 0 ) 658 return 1; 659 } 660 else if ( (c >= pConvertData->mnToCharStart2) && (c <= pConvertData->mnToCharEnd2) ) 661 { 662 dest[0] = (sal_Char)pConvertData->mpToCharTab2[c-pConvertData->mnToCharStart2]; 663 if ( dest[0] != 0 ) 664 return 1; 665 } 666 pToCharExTab = pConvertData->mpToCharExTab; 667 if ( pToCharExTab ) 668 { 669 sal_uInt16 nLow; 670 sal_uInt16 nHigh; 671 sal_uInt16 nMid; 672 sal_uInt16 nCompareChar; 673 const ImplUniCharTabData* pCharExData; 674 675 nLow = 0; 676 nHigh = pConvertData->mnToCharExCount-1; 677 do 678 { 679 nMid = (nLow+nHigh)/2; 680 pCharExData = pToCharExTab+nMid; 681 nCompareChar = pCharExData->mnUniChar; 682 if ( c < nCompareChar ) 683 { 684 if ( !nMid ) 685 break; 686 nHigh = nMid-1; 687 } 688 else 689 { 690 if ( c > nCompareChar ) 691 nLow = nMid+1; 692 else 693 { 694 dest[0] = (sal_Char)pCharExData->mnChar; 695 if ( pCharExData->mnChar2 == 0 ) 696 return 1; 697 else 698 { 699 dest[1] = (sal_Char)pCharExData->mnChar2; 700 return 2; 701 } 702 } 703 } 704 } 705 while ( nLow <= nHigh ); 706 } 707 return 0; 708 } 709 710 /* ----------------------------------------------------------------------- */ 711 712 sal_Size ImplUnicodeToChar( const ImplTextConverterData* pData, 713 void* pContext, 714 const sal_Unicode* pSrcBuf, sal_Size nSrcChars, 715 sal_Char* pDestBuf, sal_Size nDestBytes, 716 sal_uInt32 nFlags, sal_uInt32* pInfo, 717 sal_Size* pSrcCvtChars ) 718 { 719 sal_Unicode c; 720 const ImplByteConvertData* pConvertData = (const ImplByteConvertData*)pData; 721 sal_Char* pEndDestBuf; 722 const sal_Unicode* pEndSrcBuf; 723 int i; 724 int n; 725 sal_uInt16 cTemp; 726 sal_Char aTempBuf[IMPL_MAX_REPLACECHAR+2]; 727 const sal_uInt16* pReplace; 728 729 (void) pContext; /* unused */ 730 731 *pInfo = 0; 732 pEndDestBuf = pDestBuf+nDestBytes; 733 pEndSrcBuf = pSrcBuf+nSrcChars; 734 while ( pSrcBuf < pEndSrcBuf ) 735 { 736 c = *pSrcBuf; 737 if ( c < 0x80 ) 738 { 739 aTempBuf[0] = (sal_Char)c; 740 n = 1; 741 } 742 else 743 { 744 n = ImplConvertUnicodeCharToChar( pConvertData, c, aTempBuf ); 745 746 if ( n == 0 ) 747 { 748 if ( nFlags & RTL_UNICODETOTEXT_FLAGS_UNDEFINED_REPLACE ) 749 { 750 cTemp = ImplGetReplaceChar( c ); 751 if ( cTemp ) 752 n = ImplConvertUnicodeCharToChar( 753 pConvertData, cTemp, aTempBuf ); 754 } 755 756 if ( n == 0 ) 757 { 758 if ( nFlags & RTL_UNICODETOTEXT_FLAGS_UNDEFINED_REPLACESTR ) 759 { 760 pReplace = ImplGetReplaceString( c ); 761 if ( pReplace ) 762 { 763 while ( *pReplace && (n < IMPL_MAX_REPLACECHAR) ) 764 { 765 i = ImplConvertUnicodeCharToChar( 766 pConvertData, *pReplace, aTempBuf + n ); 767 if ( i == 0 ) 768 { 769 n = 0; 770 break; 771 } 772 pReplace++; 773 n += i; 774 } 775 } 776 } 777 778 /* Handle undefined and surrogates characters */ 779 /* (all surrogates characters are undefined) */ 780 if ( n == 0 ) 781 { 782 if (ImplHandleUndefinedUnicodeToTextChar(pData, 783 &pSrcBuf, 784 pEndSrcBuf, 785 &pDestBuf, 786 pEndDestBuf, 787 nFlags, 788 pInfo)) 789 continue; 790 else 791 break; 792 } 793 } 794 } 795 } 796 797 if ( pEndDestBuf - pDestBuf < n ) 798 { 799 *pInfo |= RTL_UNICODETOTEXT_INFO_ERROR | RTL_UNICODETOTEXT_INFO_DESTBUFFERTOSMALL; 800 break; 801 } 802 803 for ( i = 0; i < n; ++i ) 804 *pDestBuf++ = aTempBuf[i]; 805 pSrcBuf++; 806 } 807 808 *pSrcCvtChars = nSrcChars - (pEndSrcBuf-pSrcBuf); 809 return (nDestBytes - (pEndDestBuf-pDestBuf)); 810 } 811