/* trf-charmap.c - Unnecessarily complex routines for mapping special characters in the non-ASCII range (128..255) onto troff equivalents. This file may require modification for local troff versions. Tries to handle the four different character sets allowed in the RTF specification (ANSI, Macintosh, PC and PCA). That part in itself is easy, but the complexity stems from trying to allow for local variations in troff versions (not all troff's understand the same set of special characters) and for particular macro packages (-me, -me, -ms). The last one is silly at this point since no other part of rtf2troff knows anything about writing macro-package-specific output. Maybe someday. SelectFormatterMaps () - selects character maps specific to the given formatter, e.g., xroff, pstroff, eroff. SelectMacPackMaps () - select override maps for the given macro package. SelectCharSetMaps () - select maps for particular character set. CharMapping () - return troff equivalent for a special character. */ # include # include # include "rtf.h" # include "rtf2troff.h" /* Structure for mapping character values >= 128 to troff strings for different character sets. The reason char value/string pairs are used instead of a 128-entry table is that while the latter would be faster, it's much easier to build an incomplete table with the former, and the entries don't need to be in any particular order. */ typedef struct CharMap CharMap; struct CharMap { int charVal; char *charStr; }; /* Structure for associating default and override maps with particular versions of troff. */ # define ansiMap 0 # define macMap 1 # define pcMap 2 # define pcaMap 3 # define mapTypes 4 typedef struct CharMapTable CharMapTable; struct CharMapTable { char *formatter; CharMap *defCharMap[mapTypes]; CharMap *meCharMap[mapTypes]; CharMap *mmCharMap[mapTypes]; CharMap *msCharMap[mapTypes]; }; /* ANSI char map is taken from Windows 3.0 and Word for Windows 1.1 manuals. Note that these conflict, so there are some guesses. PC char map is taken from Windows 3.0 manual. */ static CharMap ansiCharMap [] = { 0x91, "`", /* left curly single quote */ 0x92, "'", /* right curly single quote */ 0x93, "``", /* left curly double quote */ 0x94, "''", /* right curly double quote */ 0x95, "o", /* ?? */ 0xa2, "\\(ct", /* cent */ 0xa6, "|", /* vert. bar */ 0xa7, "\\(sc", /* section */ 0xa9, "\\(co", /* copyright */ 0xab, "<<", /* alternate quote */ 0xac, "\\(no", /* not */ 0xad, "\\(hy", /* short dash */ 0xae, "\\(rg", /* registered */ 0xaf, "\\(em", /* long dash */ 0xb0, "\\(de", /* degree */ 0xb1, "\\(+-", /* plus or minus */ 0xb4, "\\(aa", /* acute accent */ 0xb5, "\\(*m", /* micro (use greek mu) */ 0xbb, ">>", /* alternate end-quote*/ 0xbc, "\\(14", /* one-fourth */ 0xbd, "\\(12", /* one-half */ 0xbe, "\\(34", /* three-fourths */ 0xc6, "AE", /* joined A-E */ 0xd7, "\\(mu", /* math multiply */ 0xdf, "\\(*b", /* German B? (fake with greek beta) */ 0xe6, "ae", /* joined a-e */ 0xf7, "\\(di", /* divide */ 0, NULL }; static CharMap macCharMap [] = { 0xa0, "\\(dg", /* dagger */ 0xa1, "\\(de", /* degree */ 0xa2, "\\(ct", /* cent */ 0xa4, "\\(sc", /* section */ 0xa5, "\\(bu", /* bullet */ 0xa7, "\\(*b", /* German B? (fake with greek beta) */ 0xa8, "\\(rg", /* registered */ 0xa9, "\\(co", /* copyright */ 0xaa, "(TM)", /* trademark */ 0xab, "\\(aa", /* acute accent */ 0xad, "\\(!=", /* not equal */ 0xae, "AE", /* joined A-E */ 0xb0, "\\(if", /* infinity */ 0xb1, "\\(+-", /* plus or minus */ 0xb2, "\\(<=", /* less than or equal */ 0xb3, "\\(>=", /* greater than or equal */ 0xb5, "\\(*m", /* micro (use greek mu) */ 0xb6, "\\(*d", /* delta */ 0xb7, "\\(*S", /* summation (Sigma) */ 0xb8, "\\(*P", /* Pi */ 0xb9, "\\(*p", /* pi */ 0xba, "\\(is", /* integral sign */ 0xbd, "\\(*W", /* ohm-sign (greek Omega) */ 0xbe, "ae", /* joined a-e */ 0xc2, "\\(no", /* not */ 0xc3, "\\(sr", /* square root */ 0xc5, "\\(ap", /* approximately */ 0xc6, "\\(*D", /* Delta */ 0xc7, "<<", /* alternate quote */ 0xc8, ">>", /* alternate end-quote*/ 0xc9, "...", /* ellipsis */ 0xca, "\\ ", /* unbreakable space */ 0xd0, "\\(hy", /* short dash */ 0xd1, "\\(em", /* long dash */ 0xd2, "``", /* left curly double quote */ 0xd3, "''", /* right curly double quote */ 0xd4, "`", /* left curly single quote */ 0xd5, "'", /* right curly single quote */ 0xd6, "\\(di", /* divide */ 0, NULL }; static CharMap pcCharMap [] = { 0x91, "ae", /* joined a-e */ 0x9b, "\\(ct", /* cent */ 0xa9, "\\(em", /* long dash */ 0xaa, "\\(no", /* not */ 0xab, "\\(12", /* one-half */ 0xac, "\\(14", /* one-fourth */ 0xae, "<<", /* alternate quote */ 0xaf, ">>", /* alternate end-quote*/ 0xf1, "\\(+-", /* plus or minus */ 0xf8, "\\(de", /* degree */ 0, NULL }; static CharMap pcaCharMap [] = { 0, NULL }; /* These -me maps are just here for example */ static CharMap meAnsiCharMap [] = { 0, NULL }; static CharMap meMacCharMap [] = { 0, NULL }; static CharMap mePcCharMap [] = { 0, NULL }; static CharMap mePcaCharMap [] = { 0, NULL }; static CharMapTable charMapTable [] = { { "troff", { ansiCharMap, macCharMap, pcCharMap, pcaCharMap }, /* defaults */ { NULL, NULL, NULL, NULL }, /* -me overrides */ { NULL, NULL, NULL, NULL }, /* -mm overrides */ { NULL, NULL, NULL, NULL } /* -ms overrides */ }, { "xroff", { ansiCharMap, macCharMap, pcCharMap, pcaCharMap }, /* defaults */ { NULL, NULL, NULL, NULL }, /* -me overrides */ { NULL, NULL, NULL, NULL }, /* -mm overrides */ { NULL, NULL, NULL, NULL } /* -ms overrides */ }, { "pstroff", { ansiCharMap, macCharMap, pcCharMap, pcaCharMap }, /* defaults */ { NULL, NULL, NULL, NULL }, /* -me overrides */ { NULL, NULL, NULL, NULL }, /* -mm overrides */ { NULL, NULL, NULL, NULL } /* -ms overrides */ }, { /* This should be the last entry. Do not change it. Put new entries before this one. */ NULL } }; /* curCharMaps points to current entry in character map table. macroMaps points to current set of override maps (default NULL until SelectMacPackMaps() is called). defaultMap points to the default map within curCharMaps (default is the ANSI map). overrideMap points to the map within the macro package override maps (default is the ANSI map after SelectMacPackMaps() is called). */ static CharMapTable *curCharMaps = &charMapTable[0]; static CharMap **macroMaps = NULL; static CharMap *defaultMap = ansiCharMap; static CharMap *overrideMap = NULL; void SelectFormatterMaps (name) char *name; { CharMapTable *cmtp; for (cmtp = charMapTable; cmtp->formatter != NULL; cmtp++) { if (strcmp (name, cmtp->formatter) == 0) { curCharMaps = cmtp; macroMaps = NULL; defaultMap = curCharMaps->defCharMap[ansiMap]; overrideMap = NULL; return; } } fprintf (stderr, "Unknown formatter: %s\n", name); exit (1); } void SelectMacPackMaps (name) char *name; { if (strcmp (name, "-me") == 0) macroMaps = curCharMaps->meCharMap; else if (strcmp (name, "-mm") == 0) macroMaps = curCharMaps->mmCharMap; else if (strcmp (name, "-ms") == 0) macroMaps = curCharMaps->msCharMap; else { fprintf (stderr, "Unknown macro package: %s\n", name); exit (1); } if (macroMaps != NULL) overrideMap = macroMaps[ansiMap]; } /* mapSet should be an rtfCharSet minor number */ void SelectCharSetMaps (mapSet) int mapSet; { int mapIndex; switch (mapSet) { case rtfAnsiCharSet: mapIndex = ansiMap; break; case rtfMacCharSet: mapIndex = macMap; break; case rtfPcCharSet: mapIndex = pcMap; break; case rtfPcaCharSet: mapIndex = pcaMap; break; default: fprintf (stderr, "Unknown charset: %d\n", mapSet); exit (1); } defaultMap = curCharMaps->defCharMap[mapIndex]; if (macroMaps != NULL) overrideMap = macroMaps[mapIndex]; } /* Figure out what string a special (non-ASCII) character should map to. (Determine troff equivalent using "normal" characters.) */ static char *_CharMapping (c, map) int c; CharMap *map; { CharMap *cmp; char *p = NULL; for (cmp = map; cmp->charStr != NULL; cmp++) { if (c == cmp->charVal) { p = cmp->charStr; break; } } return (p); } char *CharMapping (c) int c; { CharMap *cmp; char *p = NULL; if (overrideMap != NULL) p = _CharMapping (c, overrideMap); if (p == NULL && defaultMap != NULL) p = _CharMapping (c, defaultMap); return (p != NULL ? p : "<>"); }