chartrans.c

   1 /*
   2  * chartrans.c
   3  * Copyright (C) 1999-2004 A.J. van Os; Released under GNU GPL
   4  *
   5  * Description:
   6  * Translate Word characters to local representation
   7  */
   8
   9 #include <stdlib.h>
  10 #include <string.h>
  11 #include <ctype.h>
  12 #if defined(__STDC_ISO_10646__)
  13 #include <wctype.h>
  14 #endif /* __STDC_ISO_10646__ */
  15 #include "antiword.h"
  16
  17 static const USHORT usCp850[] = {       /* DOS implementation of Latin1 */
  18         0x00c7, 0x00fc, 0x00e9, 0x00e2, 0x00e4, 0x00e0, 0x00e5, 0x00e7,
  19         0x00ea, 0x00eb, 0x00e8, 0x00ef, 0x00ee, 0x00ec, 0x00c4, 0x00c5,
  20         0x00c9, 0x00e6, 0x00c6, 0x00f4, 0x00f6, 0x00f2, 0x00fb, 0x00f9,
  21         0x00ff, 0x00d6, 0x00dc, 0x00f8, 0x00a3, 0x00d8, 0x00d7, 0x0192,
  22         0x00e1, 0x00ed, 0x00f3, 0x00fa, 0x00f1, 0x00d1, 0x00aa, 0x00ba,
  23         0x00bf, 0x00ae, 0x00ac, 0x00bd, 0x00bc, 0x00a1, 0x00ab, 0x00bb,
  24         0x2591, 0x2592, 0x2593, 0x2502, 0x2524, 0x00c1, 0x00c2, 0x00c0,
  25         0x00a9, 0x2563, 0x2551, 0x2557, 0x255d, 0x00a2, 0x00a5, 0x2510,
  26         0x2514, 0x2534, 0x252c, 0x251c, 0x2500, 0x253c, 0x00e3, 0x00c3,
  27         0x255a, 0x2554, 0x2569, 0x2566, 0x2560, 0x2550, 0x256c, 0x00a4,
  28         0x00f0, 0x00d0, 0x00ca, 0x00cb, 0x00c8, 0x0131, 0x00cd, 0x00ce,
  29         0x00cf, 0x2518, 0x250c, 0x2588, 0x2584, 0x00a6, 0x00cc, 0x2580,
  30         0x00d3, 0x00df, 0x00d4, 0x00d2, 0x00f5, 0x00d5, 0x00b5, 0x00fe,
  31         0x00de, 0x00da, 0x00db, 0x00d9, 0x00fd, 0x00dd, 0x00af, 0x00b4,
  32         0x00ad, 0x00b1, 0x2017, 0x00be, 0x00b6, 0x00a7, 0x00f7, 0x00b8,
  33         0x00b0, 0x00a8, 0x00b7, 0x00b9, 0x00b3, 0x00b2, 0x25a0, 0x00a0,
  34 };
  35
  36 static const USHORT usCp1250[] = {      /* Windows implementation of Latin2 */
  37         0x20ac, 0x003f, 0x201a, 0x003f, 0x201e, 0x2026, 0x2020, 0x2021,
  38         0x003f, 0x2030, 0x0160, 0x2039, 0x015a, 0x0164, 0x017d, 0x0179,
  39         0x003f, 0x2018, 0x2019, 0x201c, 0x201d, 0x2022, 0x2013, 0x2014,
  40         0x003f, 0x2122, 0x0161, 0x203a, 0x015b, 0x0165, 0x017e, 0x017a,
  41         0x00a0, 0x02c7, 0x02d8, 0x0141, 0x00a4, 0x0104, 0x00a6, 0x00a7,
  42         0x00a8, 0x00a9, 0x015e, 0x00ab, 0x00ac, 0x00ad, 0x00ae, 0x017b,
  43         0x00b0, 0x00b1, 0x02db, 0x0142, 0x00b4, 0x00b5, 0x00b6, 0x00b7,
  44         0x00b8, 0x0105, 0x015f, 0x00bb, 0x013d, 0x02dd, 0x013e, 0x017c,
  45         0x0154, 0x00c1, 0x00c2, 0x0102, 0x00c4, 0x0139, 0x0106, 0x00c7,
  46         0x010c, 0x00c9, 0x0118, 0x00cb, 0x011a, 0x00cd, 0x00ce, 0x010e,
  47         0x0110, 0x0143, 0x0147, 0x00d3, 0x00d4, 0x0150, 0x00d6, 0x00d7,
  48         0x0158, 0x016e, 0x00da, 0x0170, 0x00dc, 0x00dd, 0x0162, 0x00df,
  49         0x0155, 0x00e1, 0x00e2, 0x0103, 0x00e4, 0x013a, 0x0107, 0x00e7,
  50         0x010d, 0x00e9, 0x0119, 0x00eb, 0x011b, 0x00ed, 0x00ee, 0x010f,
  51         0x0111, 0x0144, 0x0148, 0x00f3, 0x00f4, 0x0151, 0x00f6, 0x00f7,
  52         0x0159, 0x016f, 0x00fa, 0x0171, 0x00fc, 0x00fd, 0x0163, 0x02d9,
  53 };
  54
  55 static const USHORT usCp1251[] = {      /* Windows implementation of Cyrillic */
  56         0x0402, 0x0403, 0x201a, 0x0453, 0x201e, 0x2026, 0x2020, 0x2021,
  57         0x20ac, 0x2030, 0x0409, 0x2039, 0x040a, 0x040c, 0x040b, 0x040f,
  58         0x0452, 0x2018, 0x2019, 0x201c, 0x201d, 0x2022, 0x2013, 0x2014,
  59         0x00f3, 0x2122, 0x0459, 0x203a, 0x045a, 0x045c, 0x045b, 0x045f,
  60         0x00a0, 0x040e, 0x045e, 0x0408, 0x00a4, 0x0490, 0x00a6, 0x00a7,
  61         0x0401, 0x00a9, 0x0404, 0x00ab, 0x00ac, 0x00ad, 0x00ae, 0x0407,
  62         0x00b0, 0x00b1, 0x0406, 0x0456, 0x0491, 0x00b5, 0x00b6, 0x00b7,
  63         0x0451, 0x2116, 0x0454, 0x00bb, 0x0458, 0x0405, 0x0455, 0x0457,
  64         0x0410, 0x0411, 0x0412, 0x0413, 0x0414, 0x0415, 0x0416, 0x0417,
  65         0x0418, 0x0419, 0x041a, 0x041b, 0x041c, 0x041d, 0x041e, 0x041f,
  66         0x0420, 0x0421, 0x0422, 0x0423, 0x0424, 0x0425, 0x0426, 0x0427,
  67         0x0428, 0x0429, 0x042a, 0x042b, 0x042c, 0x042d, 0x042e, 0x042f,
  68         0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437,
  69         0x0438, 0x0439, 0x043a, 0x043b, 0x043c, 0x043d, 0x043e, 0x043f,
  70         0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447,
  71         0x0448, 0x0449, 0x044a, 0x044b, 0x044c, 0x044d, 0x044e, 0x044f,
  72 };
  73
  74 static const USHORT usCp1252[] = {      /* Windows implementation of Latin1 */
  75         0x20ac, 0x003f, 0x201a, 0x0192, 0x201e, 0x2026, 0x2020, 0x2021,
  76         0x02c6, 0x2030, 0x0160, 0x2039, 0x0152, 0x003f, 0x017d, 0x003f,
  77         0x003f, 0x2018, 0x2019, 0x201c, 0x201d, 0x2022, 0x2013, 0x2014,
  78         0x02dc, 0x2122, 0x0161, 0x203a, 0x0153, 0x003f, 0x017e, 0x0178,
  79         0x00a0, 0x00a1, 0x00a2, 0x00a3, 0x00a4, 0x00a5, 0x00a6, 0x00a7,
  80         0x00a8, 0x00a9, 0x00aa, 0x00ab, 0x00ac, 0x00ad, 0x00ae, 0x00af,
  81         0x00b0, 0x00b1, 0x00b2, 0x00b3, 0x00b4, 0x00b5, 0x00b6, 0x00b7,
  82         0x00b8, 0x00b9, 0x00ba, 0x00bb, 0x00bc, 0x00bd, 0x00be, 0x00bf,
  83         0x00c0, 0x00c1, 0x00c2, 0x00c3, 0x00c4, 0x00c5, 0x00c6, 0x00c7,
  84         0x00c8, 0x00c9, 0x00ca, 0x00cb, 0x00cc, 0x00cd, 0x00ce, 0x00cf,
  85         0x00d0, 0x00d1, 0x00d2, 0x00d3, 0x00d4, 0x00d5, 0x00d6, 0x00d7,
  86         0x00d8, 0x00d9, 0x00da, 0x00db, 0x00dc, 0x00dd, 0x00de, 0x00df,
  87         0x00e0, 0x00e1, 0x00e2, 0x00e3, 0x00e4, 0x00e5, 0x00e6, 0x00e7,
  88         0x00e8, 0x00e9, 0x00ea, 0x00eb, 0x00ec, 0x00ed, 0x00ee, 0x00ef,
  89         0x00f0, 0x00f1, 0x00f2, 0x00f3, 0x00f4, 0x00f5, 0x00f6, 0x00f7,
  90         0x00f8, 0x00f9, 0x00fa, 0x00fb, 0x00fc, 0x00fd, 0x00fe, 0x00ff,
  91 };
  92
  93 static const USHORT usMacRoman[] = {    /* Apple implementation of Latin1 */
  94         0x00c4, 0x00c5, 0x00c7, 0x00c9, 0x00d1, 0x00d6, 0x00dc, 0x00e1,
  95         0x00e0, 0x00e2, 0x00e4, 0x00e3, 0x00e5, 0x00e7, 0x00e9, 0x00e8,
  96         0x00ea, 0x00eb, 0x00ed, 0x00ec, 0x00ee, 0x00ef, 0x00f1, 0x00f3,
  97         0x00f2, 0x00f4, 0x00f6, 0x00f5, 0x00fa, 0x00f9, 0x00fb, 0x00fc,
  98         0x2020, 0x00b0, 0x00a2, 0x00a3, 0x00a7, 0x2022, 0x00b6, 0x00df,
  99         0x00ae, 0x00a9, 0x2122, 0x00b4, 0x00a8, 0x2260, 0x00c6, 0x00d8,
 100         0x221e, 0x00b1, 0x2264, 0x2265, 0x00a5, 0x00b5, 0x2202, 0x2211,
 101         0x220f, 0x03c0, 0x222b, 0x00aa, 0x00ba, 0x2126, 0x00e6, 0x00f8,
 102         0x00bf, 0x00a1, 0x00ac, 0x221a, 0x0192, 0x2248, 0x2206, 0x00ab,
 103         0x00bb, 0x2026, 0x00a0, 0x00c0, 0x00c3, 0x00d5, 0x0152, 0x0153,
 104         0x2013, 0x2014, 0x201c, 0x201d, 0x2018, 0x2019, 0x00f7, 0x25ca,
 105         0x00ff, 0x0178, 0x2044, 0x00a4, 0x2039, 0x203a, 0xfb01, 0xfb02,
 106         0x2021, 0x00b7, 0x201a, 0x201e, 0x2030, 0x00c2, 0x00ca, 0x00c1,
 107         0x00cb, 0x00c8, 0x00cd, 0x00ce, 0x00cf, 0x00cc, 0x00d3, 0x00d4,
 108         0x003f, 0x00d2, 0x00da, 0x00db, 0x00d9, 0x0131, 0x02c6, 0x02dc,
 109         0x00af, 0x02d8, 0x02d9, 0x02da, 0x00b8, 0x02dd, 0x02db, 0x02c7,
 110 };
 111
 112 static const USHORT usPrivateArea[] = {
 113         0x0020, 0x0021, 0x2200, 0x0023, 0x2203, 0x0025, 0x0026, 0x220d,
 114         0x0028, 0x0029, 0x2217, 0x002b, 0x002c, 0x2212, 0x002e, 0x002f,
 115         0x0030, 0x0031, 0x0032, 0x0033, 0x0034, 0x0035, 0x0036, 0x0037,
 116         0x0038, 0x0039, 0x003a, 0x003b, 0x003c, 0x2019, 0x003e, 0x003f,
 117         0x201d, 0x201c, 0x0392, 0x03a7, 0x0394, 0x0395, 0x03a6, 0x0393,
 118         0x0397, 0x0399, 0x03d1, 0x039a, 0x039b, 0x039c, 0x039d, 0x039f,
 119         0x03a0, 0x0398, 0x03a1, 0x03a3, 0x03a4, 0x03a5, 0x03c2, 0x03a9,
 120         0x039e, 0x03a8, 0x0396, 0x005b, 0x2234, 0x005d, 0x22a5, 0x005f,
 121         0x003f, 0x03b1, 0x03b2, 0x03c7, 0x03b4, 0x03b5, 0x03c6, 0x03b3,
 122         0x03b7, 0x03b9, 0x03d5, 0x03ba, 0x03bb, 0x03bc, 0x03bd, 0x03bf,
 123         0x03c0, 0x03b8, 0x03c1, 0x03c3, 0x03c4, 0x03c5, 0x03d6, 0x03c9,
 124         0x03be, 0x03c8, 0x03b6, 0x007b, 0x007c, 0x007d, 0x223c, 0x003f,
 125         0x003f, 0x003f, 0x003f, 0x003f, 0x003f, 0x003f, 0x003f, 0x003f,
 126         0x003f, 0x003f, 0x003f, 0x003f, 0x003f, 0x003f, 0x003f, 0x003f,
 127         0x003f, 0x003f, 0x003f, 0x2022, 0x003f, 0x003f, 0x003f, 0x003f,
 128         0x003f, 0x003f, 0x003f, 0x003f, 0x003f, 0x003f, 0x003f, 0x003f,
 129         0x20ac, 0x03d2, 0x2032, 0x2264, 0x2044, 0x221e, 0x0192, 0x2663,
 130         0x2666, 0x2665, 0x2660, 0x2194, 0x2190, 0x2191, 0x2192, 0x2193,
 131         0x00b0, 0x00b1, 0x2033, 0x2265, 0x00d7, 0x221d, 0x2202, 0x2022,
 132         0x00f7, 0x2260, 0x2261, 0x2248, 0x2026, 0x007c, 0x23af, 0x21b5,
 133         0x2135, 0x2111, 0x211c, 0x2118, 0x2297, 0x2295, 0x2205, 0x2229,
 134         0x222a, 0x2283, 0x2287, 0x2284, 0x2282, 0x2286, 0x2208, 0x2209,
 135         0x2220, 0x2207, 0x00ae, 0x00a9, 0x2122, 0x220f, 0x221a, 0x22c5,
 136         0x00ac, 0x2227, 0x2228, 0x21d4, 0x21d0, 0x21d1, 0x21d2, 0x21d3,
 137         0x22c4, 0x3008, 0x00ae, 0x00a9, 0x2122, 0x2211, 0x239b, 0x239c,
 138         0x239d, 0x23a1, 0x23a2, 0x23a3, 0x23a7, 0x23a8, 0x23a9, 0x23aa,
 139         0x003f, 0x3009, 0x222b, 0x2320, 0x23ae, 0x2321, 0x239e, 0x239f,
 140         0x23a0, 0x23a4, 0x23a5, 0x23a6, 0x23ab, 0x23ac, 0x23ad, 0x003f,
 141 };
 142
 143 typedef struct char_table_tag {
 144         UCHAR   ucLocal;
 145         USHORT  usUnicode;
 146 } char_table_type;
 147
 148 static char_table_type  atCharTable[256];
 149 static size_t           tNextPosFree = 0;
 150
 151
 152 /*
 153  * iCompare - compare two records
 154  *
 155  * Compares two records. For use by qsort(3C) and bsearch(3C).
 156  *
 157  * returns -1 if rec1 < rec2, 0 if rec1 == rec2, 1 if rec1 > rec2
 158  */
 159 static int
 160 iCompare(const void *pvRecord1, const void *pvRecord2)
 161 {
 162         USHORT  usUnicode1, usUnicode2;
 163
 164         usUnicode1 = ((char_table_type *)pvRecord1)->usUnicode;
 165         usUnicode2 = ((char_table_type *)pvRecord2)->usUnicode;
 166
 167         if (usUnicode1 < usUnicode2) {
 168                 return -1;
 169         }
 170         if (usUnicode1 > usUnicode2) {
 171                 return 1;
 172         }
 173         return 0;
 174 } /* end of iCompare */
 175
 176 /*
 177  * pGetCharTableRecord - get the character table record
 178  *
 179  * returns a pointer to the record when found, otherwise NULL
 180  */
 181 static const char_table_type *
 182 pGetCharTableRecord(USHORT usUnicode)
 183 {
 184         char_table_type tKey;
 185
 186         if (tNextPosFree == 0) {
 187                 return NULL;
 188         }
 189         tKey.usUnicode = usUnicode;
 190         tKey.ucLocal = 0;
 191         return (char_table_type *)bsearch(&tKey,
 192                         atCharTable,
 193                         tNextPosFree, sizeof(atCharTable[0]),
 194                         iCompare);
 195 } /* end of pGetCharTableRecord */
 196
 197 /*
 198  * ucGetBulletCharacter - get the local representation of the bullet
 199  */
 200 UCHAR
 201 ucGetBulletCharacter(conversion_type eConversionType, encoding_type eEncoding)
 202 {
 203 #if defined(__riscos)
 204         return 0x8f;
 205 #else
 206         const char_table_type   *pRec;
 207
 208         fail(eEncoding == encoding_utf_8);
 209
 210         if (eEncoding == encoding_latin_1 &&
 211             (eConversionType == conversion_ps ||
 212              eConversionType == conversion_pdf)) {
 213                 /* Ugly, but it makes the PostScript and PDF look better */
 214                 return (UCHAR)143;
 215         }
 216         if (eConversionType != conversion_text &&
 217             eConversionType != conversion_fmt_text) {
 218                 pRec = pGetCharTableRecord(UNICODE_BULLET);
 219                 if (pRec != NULL) {
 220                         return pRec->ucLocal;
 221                 }
 222                 pRec = pGetCharTableRecord(UNICODE_BULLET_OPERATOR);
 223                 if (pRec != NULL) {
 224                         return pRec->ucLocal;
 225                 }
 226                 pRec = pGetCharTableRecord(UNICODE_MIDDLE_DOT);
 227                 if (pRec != NULL) {
 228                         return pRec->ucLocal;
 229                 }
 230         }
 231         return (UCHAR)'.';
 232 #endif /* __riscos */
 233 } /* end of ucGetBulletCharacter */
 234
 235 /*
 236  * ucGetNbspCharacter - get the local representation of the non-breaking space
 237  */
 238 UCHAR
 239 ucGetNbspCharacter(void)
 240 {
 241         const char_table_type   *pRec;
 242
 243         pRec = pGetCharTableRecord(0x00a0);     /* Unicode non-breaking space */
 244         if (pRec == NULL) {
 245                 DBG_MSG("Non-breaking space record not found");
 246                 /* No value found, use the best guess */
 247                 return (UCHAR)0xa0;
 248         }
 249         return pRec->ucLocal;
 250 } /* end of ucGetNbspCharacter */
 251
 252 /*
 253  * bReadCharacterMappingTable - read the mapping table
 254  *
 255  * Read the character mapping table from file and have the contents sorted
 256  *
 257  * returns TRUE if successful, otherwise FALSE
 258  */
 259 BOOL
 260 bReadCharacterMappingTable(FILE *pFile)
 261 {
 262         char    *pcTmp;
 263         ULONG   ulUnicode;
 264         UINT    uiLocal;
 265         int     iFields;
 266         char    szLine[81];
 267
 268         if (pFile == NULL) {
 269                 return FALSE;
 270         }
 271
 272         /* Clean the table first */
 273         (void)memset(atCharTable, 0, sizeof(atCharTable));
 274
 275         /* Fill the table */
 276         while (fgets(szLine, (int)sizeof(szLine), pFile)) {
 277                 if (szLine[0] == '#' ||
 278                     szLine[0] == '\r' ||
 279                     szLine[0] == '\n') {
 280                         /* Comment or empty line */
 281                         continue;
 282                 }
 283                 iFields = sscanf(szLine, "%x %lx %*s", &uiLocal, &ulUnicode);
 284                 if (iFields != 2) {
 285                         pcTmp = strchr(szLine, '\r');
 286                         if (pcTmp != NULL) {
 287                                 *pcTmp = '\0';
 288                         }
 289                         pcTmp = strchr(szLine, '\n');
 290                         if (pcTmp != NULL) {
 291                                 *pcTmp = '\0';
 292                         }
 293                         werr(0, "Syntax error in: '%s'", szLine);
 294                         continue;
 295                 }
 296                 if (uiLocal > 0xff || ulUnicode > 0xffff) {
 297                         werr(0, "Syntax error in: '%02x %04lx'",
 298                                         uiLocal, ulUnicode);
 299                         continue;
 300                 }
 301                 /* Store only the relevant entries */
 302                 if (uiLocal != ulUnicode || uiLocal >= 0x80) {
 303                         atCharTable[tNextPosFree].ucLocal = (UCHAR)uiLocal;
 304                         atCharTable[tNextPosFree].usUnicode = (USHORT)ulUnicode;
 305                         tNextPosFree++;
 306                 }
 307                 if (tNextPosFree >= elementsof(atCharTable)) {
 308                         werr(0, "Too many entries in the character mapping "
 309                                 "file. Ignoring the rest.");
 310                         break;
 311                 }
 312         }
 313
 314         if (tNextPosFree != 0) {
 315                 DBG_HEX(atCharTable[0].usUnicode);
 316                 DBG_HEX(atCharTable[tNextPosFree - 1].usUnicode);
 317
 318                 qsort(atCharTable,
 319                         tNextPosFree, sizeof(atCharTable[0]),
 320                         iCompare);
 321
 322                 DBG_HEX(atCharTable[0].usUnicode);
 323                 DBG_HEX(atCharTable[tNextPosFree - 1].usUnicode);
 324         }
 325
 326         return TRUE;
 327 } /* end of bReadCharacterMappingTable */
 328
 329 /*
 330  * ulTranslateCharacters - Translate characters to local representation
 331  *
 332  * Translate all characters to local representation
 333  *
 334  * returns the translated character
 335  */
 336 ULONG
 337 ulTranslateCharacters(USHORT usChar, ULONG ulFileOffset, int iWordVersion,
 338         conversion_type eConversionType, encoding_type eEncoding,
 339         BOOL bUseMacCharSet)
 340 {
 341         const char_table_type   *pTmp;
 342         const USHORT    *usCharSet;
 343
 344         usCharSet = NULL;
 345         if (bUseMacCharSet) {
 346                 /* Macintosh character set */
 347                 usCharSet = usMacRoman;
 348         } else if (iWordVersion == 0) {
 349                 /* DOS character set */
 350                 usCharSet = usCp850;
 351         } else {
 352                 /* Windows character set */
 353                 switch (eEncoding) {
 354                 case encoding_latin_2:
 355                         usCharSet = usCp1250;
 356                         break;
 357                 case encoding_cyrillic:
 358                         usCharSet = usCp1251;
 359                         break;
 360                 case encoding_latin_1:
 361                 default:
 362                         usCharSet = usCp1252;
 363                         break;
 364                 }
 365         }
 366         fail(usCharSet == NULL);
 367         if (usChar >= 0x80 && usChar <= 0x9f) {
 368                 /* Translate implementation defined characters */
 369                 usChar = usCharSet[usChar - 0x80];
 370         } else if (iWordVersion < 8 && usChar >= 0xa0 && usChar <= 0xff) {
 371                 /* Translate old character set to Unixcode */
 372                 usChar = usCharSet[usChar - 0x80];
 373         }
 374
 375         /* Microsoft Unicode to real Unicode */
 376         if (usChar >= 0xf020 && usChar <= 0xf0ff) {
 377                 DBG_HEX_C(usPrivateArea[usChar - 0xf020] == 0x003f, usChar);
 378                 usChar = usPrivateArea[usChar - 0xf020];
 379         }
 380
 381         /* Characters with a special meaning in Word */
 382         switch (usChar) {
 383         case IGNORE_CHARACTER:
 384         case FOOTNOTE_SEPARATOR:
 385         case FOOTNOTE_CONTINUATION:
 386         case ANNOTATION:
 387         case FRAME:
 388         case LINE_FEED:
 389         case WORD_SOFT_HYPHEN:
 390         case UNICODE_HYPHENATION_POINT:
 391                 return IGNORE_CHARACTER;
 392         case PICTURE:
 393         case TABLE_SEPARATOR:
 394         case TAB:
 395         case HARD_RETURN:
 396         case PAGE_BREAK:
 397         case PAR_END:
 398         case COLUMN_FEED:
 399                 return (ULONG)usChar;
 400         case FOOTNOTE_OR_ENDNOTE:
 401                 NO_DBG_HEX(ulFileOffset);
 402                 switch (eGetNotetype(ulFileOffset)) {
 403                 case notetype_is_footnote:
 404                         return FOOTNOTE_CHAR;
 405                 case notetype_is_endnote:
 406                         return ENDNOTE_CHAR;
 407                 default:
 408                         return UNKNOWN_NOTE_CHAR;
 409                 }
 410         case WORD_UNBREAKABLE_JOIN:
 411                 return (ULONG)OUR_UNBREAKABLE_JOIN;
 412         default:
 413                 break;
 414         }
 415
 416         if (eEncoding != encoding_utf_8) {
 417                 /* Latin characters in an oriental text */
 418                 if (usChar >= 0xff01 && usChar <= 0xff5e) {
 419                         usChar -= 0xfee0;
 420                 }
 421         }
 422
 423         if (eEncoding == encoding_latin_1 &&
 424             (eConversionType == conversion_ps ||
 425              eConversionType == conversion_pdf)) {
 426                 /* Ugly, but it makes the PostScript and PDF look better */
 427                 switch (usChar) {
 428                 case UNICODE_ELLIPSIS:
 429                         return 140;
 430                 case UNICODE_TRADEMARK_SIGN:
 431                         return 141;
 432                 case UNICODE_PER_MILLE_SIGN:
 433                         return 142;
 434                 case UNICODE_BULLET:
 435                 case UNICODE_BULLET_OPERATOR:
 436                 case UNICODE_BLACK_CLUB_SUIT:
 437                         return 143;
 438                 case UNICODE_LEFT_SINGLE_QMARK:
 439                         return 144;
 440                 case UNICODE_RIGHT_SINGLE_QMARK:
 441                         return 145;
 442                 case UNICODE_SINGLE_LEFT_ANGLE_QMARK:
 443                         return 146;
 444                 case UNICODE_SINGLE_RIGHT_ANGLE_QMARK:
 445                         return 147;
 446                 case UNICODE_LEFT_DOUBLE_QMARK:
 447                         return 148;
 448                 case UNICODE_RIGHT_DOUBLE_QMARK:
 449                         return 149;
 450                 case UNICODE_DOUBLE_LOW_9_QMARK:
 451                         return 150;
 452                 case UNICODE_EN_DASH:
 453                         return 151;
 454                 case UNICODE_EM_DASH:
 455                         return 152;
 456                 case UNICODE_MINUS_SIGN:
 457                         return 153;
 458                 case UNICODE_CAPITAL_LIGATURE_OE:
 459                         return 154;
 460                 case UNICODE_SMALL_LIGATURE_OE:
 461                         return 155;
 462                 case UNICODE_DAGGER:
 463                         return 156;
 464                 case UNICODE_DOUBLE_DAGGER:
 465                         return 157;
 466                 case UNICODE_SMALL_LIGATURE_FI:
 467                         return 158;
 468                 case UNICODE_SMALL_LIGATURE_FL:
 469                         return 159;
 470                 default:
 471                         break;
 472                 }
 473         }
 474
 475         if (eConversionType == conversion_pdf) {
 476                 if (eEncoding == encoding_latin_1) {
 477                         switch (usChar) {
 478                         case UNICODE_EURO_SIGN:
 479                                 return 128;
 480                         default:
 481                                 break;
 482                         }
 483                 } else if (eEncoding == encoding_latin_2) {
 484                         switch (usChar) {
 485                         case UNICODE_CAPITAL_D_WITH_STROKE:
 486                         case UNICODE_SMALL_D_WITH_STROKE:
 487                                 return 0x3f;
 488                         default:
 489                                 break;
 490                         }
 491                 }
 492         }
 493
 494         if (usChar < 0x80) {
 495                 /* US ASCII */
 496                 if (usChar < 0x20 || usChar == 0x7f) {
 497                         /* Ignore control characters */
 498                         DBG_HEX(usChar);
 499                         DBG_FIXME();
 500                         return IGNORE_CHARACTER;
 501                 }
 502                 return (ULONG)usChar;
 503         }
 504
 505         if (eEncoding == encoding_utf_8) {
 506                 /* No need to convert Unicode characters */
 507                 return (ULONG)usChar;
 508         }
 509
 510         /* Unicode to local representation */
 511         pTmp = pGetCharTableRecord(usChar);
 512         if (pTmp != NULL) {
 513                 DBG_HEX_C(usChar >= 0x7f && usChar <= 0x9f, usChar);
 514                 return (ULONG)pTmp->ucLocal;
 515         }
 516
 517         /* Fancy characters to simple US ASCII */
 518         switch (usChar) {
 519         case UNICODE_SMALL_F_HOOK:
 520                 return (ULONG)'f';
 521         case UNICODE_GREEK_CAPITAL_CHI:
 522                 return (ULONG)'X';
 523         case UNICODE_GREEK_SMALL_UPSILON:
 524                 return (ULONG)'v';
 525         case UNICODE_MODIFIER_CIRCUMFLEX:
 526         case UNICODE_UPWARDS_ARROW:
 527                 return (ULONG)'^';
 528         case UNICODE_SMALL_TILDE:
 529         case UNICODE_TILDE_OPERATOR:
 530                 return (ULONG)'~';
 531         case UNICODE_EN_QUAD:
 532         case UNICODE_EM_QUAD:
 533         case UNICODE_EN_SPACE:
 534         case UNICODE_EM_SPACE:
 535         case UNICODE_THREE_PER_EM_SPACE:
 536         case UNICODE_FOUR_PER_EM_SPACE:
 537         case UNICODE_SIX_PER_EM_SPACE:
 538         case UNICODE_FIGURE_SPACE:
 539         case UNICODE_PUNCTUATION_SPACE:
 540         case UNICODE_THIN_SPACE:
 541         case UNICODE_NARROW_NO_BREAK_SPACE:
 542         case UNICODE_LIGHT_SHADE:
 543         case UNICODE_MEDIUM_SHADE:
 544         case UNICODE_DARK_SHADE:
 545                 return (ULONG)' ';
 546         case UNICODE_LEFT_DOUBLE_QMARK:
 547         case UNICODE_RIGHT_DOUBLE_QMARK:
 548         case UNICODE_DOUBLE_LOW_9_QMARK:
 549         case UNICODE_DOUBLE_HIGH_REV_9_QMARK:
 550         case UNICODE_DOUBLE_PRIME:
 551                 return (ULONG)'"';
 552         case UNICODE_LEFT_SINGLE_QMARK:
 553         case UNICODE_RIGHT_SINGLE_QMARK:
 554         case UNICODE_SINGLE_LOW_9_QMARK:
 555         case UNICODE_SINGLE_HIGH_REV_9_QMARK:
 556         case UNICODE_PRIME:
 557                 return (ULONG)'\'';
 558         case UNICODE_HYPHEN:
 559         case UNICODE_NON_BREAKING_HYPHEN:
 560         case UNICODE_FIGURE_DASH:
 561         case UNICODE_EN_DASH:
 562         case UNICODE_EM_DASH:
 563         case UNICODE_HORIZONTAL_BAR:
 564         case UNICODE_MINUS_SIGN:
 565         case UNICODE_BD_LIGHT_HORIZONTAL:
 566         case UNICODE_BD_DOUBLE_HORIZONTAL:
 567                 return (ULONG)'-';
 568         case UNICODE_DOUBLE_VERTICAL_LINE:
 569         case UNICODE_BD_LIGHT_VERTICAL:
 570         case UNICODE_BD_DOUBLE_VERTICAL:
 571                 return (ULONG)'|';
 572         case UNICODE_DOUBLE_LOW_LINE:
 573                 return (ULONG)'_';
 574         case UNICODE_DAGGER:
 575                 return (ULONG)'+';
 576         case UNICODE_DOUBLE_DAGGER:
 577                 return (ULONG)'#';
 578         case UNICODE_BULLET:
 579         case UNICODE_BULLET_OPERATOR:
 580         case UNICODE_BLACK_CLUB_SUIT:
 581                 return (ULONG)ucGetBulletCharacter(eConversionType, eEncoding);
 582         case UNICODE_ONE_DOT_LEADER:
 583         case UNICODE_TWO_DOT_LEADER:
 584                 return (ULONG)'.';
 585         case UNICODE_ELLIPSIS:
 586 #if defined(__riscos)
 587                 return (ULONG)OUR_ELLIPSIS;
 588 #else
 589                 if (ulFileOffset == 0) {
 590                         return (ULONG)OUR_ELLIPSIS;
 591                 }
 592                 return UNICODE_ELLIPSIS;
 593 #endif /* __riscos */
 594         case UNICODE_DOUBLE_LEFT_ANGLE_QMARK:
 595         case UNICODE_TRIANGULAR_BULLET:
 596         case UNICODE_SINGLE_LEFT_ANGLE_QMARK:
 597         case UNICODE_LEFTWARDS_ARROW:
 598                 return (ULONG)'<';
 599         case UNICODE_DOUBLE_RIGHT_ANGLE_QMARK:
 600         case UNICODE_SINGLE_RIGHT_ANGLE_QMARK:
 601         case UNICODE_RIGHTWARDS_ARROW:
 602                 return (ULONG)'>';
 603         case UNICODE_UNDERTIE:
 604                 return (ULONG)'-';
 605         case UNICODE_N_ARY_SUMMATION:
 606                 return (ULONG)'S';
 607         case UNICODE_EURO_SIGN:
 608                 return (ULONG)'E';
 609         case UNICODE_CIRCLE:
 610         case UNICODE_SQUARE:
 611                 return (ULONG)'O';
 612         case UNICODE_DIAMOND:
 613                 return (ULONG)OUR_DIAMOND;
 614         case UNICODE_NUMERO_SIGN:
 615                 return (ULONG)'N';
 616         case UNICODE_KELVIN_SIGN:
 617                 return (ULONG)'K';
 618         case UNICODE_DOWNWARDS_ARROW:
 619                 return (ULONG)'v';
 620         case UNICODE_FRACTION_SLASH:
 621         case UNICODE_DIVISION_SLASH:
 622                 return (ULONG)'/';
 623         case UNICODE_ASTERISK_OPERATOR:
 624                 return (ULONG)'*';
 625         case UNICODE_RATIO:
 626                 return (ULONG)':';
 627         case UNICODE_BD_LIGHT_DOWN_RIGHT:
 628         case UNICODE_BD_LIGHT_DOWN_AND_LEFT:
 629         case UNICODE_BD_LIGHT_UP_AND_RIGHT:
 630         case UNICODE_BD_LIGHT_UP_AND_LEFT:
 631         case UNICODE_BD_LIGHT_VERTICAL_AND_RIGHT:
 632         case UNICODE_BD_LIGHT_VERTICAL_AND_LEFT:
 633         case UNICODE_BD_LIGHT_DOWN_AND_HORIZONTAL:
 634         case UNICODE_BD_LIGHT_UP_AND_HORIZONTAL:
 635         case UNICODE_BD_LIGHT_VERTICAL_AND_HORIZONTAL:
 636         case UNICODE_BD_DOUBLE_DOWN_AND_RIGHT:
 637         case UNICODE_BD_DOUBLE_DOWN_AND_LEFT:
 638         case UNICODE_BD_DOUBLE_UP_AND_RIGHT:
 639         case UNICODE_BD_DOUBLE_UP_AND_LEFT:
 640         case UNICODE_BD_DOUBLE_VERTICAL_AND_RIGHT:
 641         case UNICODE_BD_DOUBLE_VERTICAL_AND_LEFT:
 642         case UNICODE_BD_DOUBLE_DOWN_AND_HORIZONTAL:
 643         case UNICODE_BD_DOUBLE_UP_AND_HORIZONTAL:
 644         case UNICODE_BD_DOUBLE_VERTICAL_AND_HORIZONTAL:
 645         case UNICODE_BLACK_SQUARE:
 646                 return (ULONG)'+';
 647         case UNICODE_HAIR_SPACE:
 648         case UNICODE_ZERO_WIDTH_SPACE:
 649         case UNICODE_ZERO_WIDTH_NON_JOINER:
 650         case UNICODE_ZERO_WIDTH_JOINER:
 651         case UNICODE_LEFT_TO_RIGHT_MARK:
 652         case UNICODE_RIGHT_TO_LEFT_MARK:
 653         case UNICODE_LEFT_TO_RIGHT_EMBEDDING:
 654         case UNICODE_RIGHT_TO_LEFT_EMBEDDING:
 655         case UNICODE_POP_DIRECTIONAL_FORMATTING:
 656         case UNICODE_LEFT_TO_RIGHT_OVERRIDE:
 657         case UNICODE_RIGHT_TO_LEFT_OVERRIDE:
 658         case UNICODE_ZERO_WIDTH_NO_BREAK_SPACE:
 659                 return IGNORE_CHARACTER;
 660         default:
 661                 break;
 662         }
 663
 664         if (usChar == UNICODE_TRADEMARK_SIGN) {
 665                 /*
 666                  * No local representation, it doesn't look like anything in
 667                  * US-ASCII and a question mark does more harm than good.
 668                  */
 669                 return IGNORE_CHARACTER;
 670         }
 671
 672         if (usChar >= 0xa0 && usChar <= 0xff) {
 673                 /* Before Word 97, Word did't use Unicode */
 674                 return (ULONG)usChar;
 675         }
 676
 677         DBG_HEX_C(usChar < 0x3000 || usChar >= 0xd800, ulFileOffset);
 678         DBG_HEX_C(usChar < 0x3000 || usChar >= 0xd800, usChar);
 679         DBG_MSG_C(usChar >= 0xe000 && usChar < 0xf900, "Private Use Area");
 680
 681         /* Untranslated Unicode character */
 682         return 0x3f;
 683 } /* end of ulTranslateCharacters */
 684
 685 /*
 686  * ulToUpper - convert letter to upper case
 687  *
 688  * This function converts a letter to upper case. Unlike toupper(3) this
 689  * function is independent from the settings of locale. This comes in handy
 690  * for people who have to read Word documents in more than one language or
 691  * contain more than one language.
 692  *
 693  * returns the converted letter, or ulChar if the conversion was not possible.
 694  */
 695 ULONG
 696 ulToUpper(ULONG ulChar)
 697 {
 698         if (ulChar < 0x80) {
 699                 /* US ASCII: use standard function */
 700                 return (ULONG)toupper((int)ulChar);
 701         }
 702         if (ulChar >= 0xe0 && ulChar <= 0xfe && ulChar != 0xf7) {
 703                 /*
 704                  * Lower case accented characters
 705                  * 0xf7 is Division sign; 0xd7 is Multiplication sign
 706                  * 0xff is y with diaeresis; 0xdf is Sharp s
 707                  */
 708                 return ulChar & ~0x20;
 709         }
 710 #if defined(__STDC_ISO_10646__)
 711         /*
 712          * If this is ISO C99 and all locales have wchar_t = ISO 10646
 713          * (e.g., glibc 2.2 or newer), then use standard function
 714          */
 715         if (ulChar > 0xff) {
 716                 return (ULONG)towupper((wint_t)ulChar);
 717         }
 718 #endif /* __STDC_ISO_10646__ */
 719         return ulChar;
 720 } /* end of ulToUpper */