trfn.c

   1 #include <ctype.h>
   2 #include <stdio.h>
   3 #include <stdlib.h>
   4 #include <string.h>
   5 #include "sbuf.h"
   6 #include "tab.h"
   7 #include "trfn.h"
   8 #include "trfn_agl.h"
   9 #include "trfn_ch.h"
  10
  11 #define WX(w)           (((w) < 0 ? (w) - trfn_div / 20 : (w) + trfn_div / 20) * (long) 10 / trfn_div)
  12 #define LEN(a)          ((sizeof(a) / sizeof((a)[0])))
  13 #define HEXDIGS         "0123456789ABCDEF"
  14 #define NCHAR           8       /* number of characters per glyph */
  15 #define GNLEN           64      /* glyph name length */
  16 #define AGLLEN          8192    /* adobe glyphlist length */
  17
  18 static struct sbuf *sbuf_char;  /* characters */
  19 static struct sbuf *sbuf_kern;  /* kerning pairs */
  20 static int trfn_div;            /* divisor of widths x 10 */
  21 static int trfn_swid;           /* space width */
  22 static int trfn_special;        /* special flag */
  23 static int trfn_kmin;           /* minimum kerning value */
  24 static int trfn_bbox;           /* include bounding box */
  25 static int trfn_noligs;         /* suppress ligatures */
  26 static int trfn_pos;            /* include glyph positions */
  27 static char trfn_ligs[8192];    /* font ligatures */
  28 static char trfn_ligs2[8192];   /* font ligatures, whose length is two */
  29 static char trfn_trname[256];   /* font troff name */
  30 static char trfn_psname[256];   /* font ps name */
  31 static char trfn_path[4096];    /* font path */
  32 /* character type */
  33 static int trfn_asc;            /* minimum height of glyphs with ascender */
  34 static int trfn_desc;           /* minimum depth of glyphs with descender */
  35
  36 /* lookup tables */
  37 static struct tab *tab_agl;     /* adobe glyph list table */
  38 static struct tab *tab_alts;    /* character aliases table */
  39
  40 static int utf8len(int c)
  41 {
  42         if (c > 0 && c <= 0x7f)
  43                 return 1;
  44         if (c >= 0xfc)
  45                 return 6;
  46         if (c >= 0xf8)
  47                 return 5;
  48         if (c >= 0xf0)
  49                 return 4;
  50         if (c >= 0xe0)
  51                 return 3;
  52         if (c >= 0xc0)
  53                 return 2;
  54         return c != 0;
  55 }
  56
  57 static int utf8get(char **src)
  58 {
  59         int result;
  60         int l = 1;
  61         char *s = *src;
  62         if (~((unsigned char) **src) & 0xc0)
  63                 return (unsigned char) *(*src)++;
  64         while (l < 6 && (unsigned char) *s & (0x40 >> l))
  65                 l++;
  66         result = (0x3f >> l) & (unsigned char) *s++;
  67         while (l--)
  68                 result = (result << 6) | ((unsigned char) *s++ & 0x3f);
  69         *src = s;
  70         return result;
  71 }
  72
  73 static void utf8put(char **d, int c)
  74 {
  75         int l = 0;
  76         if (c > 0xffff) {
  77                 *(*d)++ = 0xf0 | (c >> 18);
  78                 l = 3;
  79         } else if (c > 0x7ff) {
  80                 *(*d)++ = 0xe0 | (c >> 12);
  81                 l = 2;
  82         } else if (c > 0x7f) {
  83                 *(*d)++ = 0xc0 | (c >> 6);
  84                 l = 1;
  85         } else {
  86                 *(*d)++ = c > 0 ? c : ' ';
  87         }
  88         while (l--)
  89                 *(*d)++ = 0x80 | ((c >> (l * 6)) & 0x3f);
  90         **d = '\0';
  91 }
  92
  93 static int hexval(char *s, int len)
  94 {
  95         char *digs = HEXDIGS;
  96         int n = 0;
  97         int i;
  98         for (i = 0; i < len; i++) {
  99                 if (s[i] && strchr(digs, s[i]))
 100                         n = n * 16 + (strchr(digs, s[i]) - digs);
 101                 else
 102                         break;
 103         }
 104         return i < 4 ? -1 : n;
 105 }
 106
 107 static int agl_map(char *d, char *s)
 108 {
 109         char *u = tab_get(tab_agl, s);  /* unicode code point like "FB8E" */
 110         if (!u)
 111                 return 1;
 112         while (u && *u) {
 113                 while (*u == ' ')
 114                         u++;
 115                 utf8put(&d, hexval(u, 6));
 116                 u = strchr(u, ' ');
 117         }
 118         *d = '\0';
 119         return 0;
 120 }
 121
 122 static int achar_map(char *name)
 123 {
 124         int i;
 125         for (i = 0; i < LEN(achars); i++) {
 126                 struct achar *a = &achars[i];
 127                 if (!strncmp(a->name, name, strlen(a->name))) {
 128                         char *postfix = name + strlen(a->name);
 129                         if (!*postfix)
 130                                 return a->c;
 131                         if (!strcmp("isolated", postfix))
 132                                 return a->s ? a->s : a->c;
 133                         if (!strcmp("initial", postfix))
 134                                 return a->i ? a->i : a->c;
 135                         if (!strcmp("medial", postfix))
 136                                 return a->m ? a->m : a->c;
 137                         if (!strcmp("final", postfix))
 138                                 return a->f ? a->f : a->c;
 139                 }
 140         }
 141         return 0;
 142 }
 143
 144 static int achar_shape(int c, int pjoin, int njoin)
 145 {
 146         int i;
 147         for (i = 0; i < LEN(achars); i++) {
 148                 struct achar *a = &achars[i];
 149                 if (a->c == c) {
 150                         if (!pjoin && !njoin)
 151                                 return a->c;
 152                         if (!pjoin && njoin)
 153                                 return a->i ? a->i : a->c;
 154                         if (pjoin && njoin)
 155                                 return a->m ? a->m : a->c;
 156                         if (pjoin && !njoin)
 157                                 return a->f ? a->f : a->c;
 158                 }
 159         }
 160         return c;
 161 }
 162
 163 static void ashape(char *str, char *ext)
 164 {
 165         int s[NCHAR];
 166         char *src = str;
 167         int i, l;
 168         int bjoin = !strcmp(".medi", ext) || !strcmp(".fina", ext);
 169         int ejoin = !strcmp(".medi", ext) || !strcmp(".init", ext);
 170         for (l = 0; l < NCHAR && *src; l++)
 171                 s[l] = utf8get(&src);
 172         for (i = 0; i < l; i++)
 173                 s[i] = achar_shape(s[i], i > 0 || bjoin, i < l - 1 || ejoin);
 174         for (i = 0; i < l; i++)
 175                 utf8put(&str, s[i]);
 176 }
 177
 178 /* find the utf-8 name of src with the given unicode codepoint */
 179 static int trfn_name(char *dst, char *src, int codepoint)
 180 {
 181         char ch[GNLEN];
 182         char *d = dst;
 183         char *s;
 184         int i;
 185         if (codepoint) {
 186                 utf8put(&dst, codepoint);
 187                 return 0;
 188         }
 189         if (!src || src[0] == '.')
 190                 return 1;
 191         while (*src && *src != '.') {
 192                 s = ch;
 193                 if (src[0] == '_')
 194                         src++;
 195                 while (*src && *src != '_' && *src != '.')
 196                         *s++ = *src++;
 197                 *s = '\0';
 198                 if (!agl_map(d, ch)) {
 199                         d = strchr(d, '\0');
 200                 } else if (ch[0] == 'u' && ch[1] == 'n' &&
 201                                 ch[2] == 'i' && hexval(ch + 3, 4) > 0) {
 202                         for (i = 0; strlen(ch + 3 + 4 * i) >= 4; i++)
 203                                 utf8put(&d, hexval(ch + 3 + 4 * i, 4));
 204                 } else if (ch[0] == 'u' && hexval(ch + 1, 4) > 0) {
 205                         utf8put(&d, hexval(ch + 1, 6));
 206                 } else if (achar_map(ch)) {
 207                         utf8put(&d, achar_map(ch));
 208                 } else {
 209                         return 1;
 210                 }
 211         }
 212         ashape(dst, src);
 213         return *src && strcmp(src, ".medi") && strcmp(src, ".fina") &&
 214                         strcmp(src, ".init") && strcmp(src, ".isol");
 215 }
 216
 217 static void trfn_aglexceptions(char *dst)
 218 {
 219         int i;
 220         for (i = 0; i < LEN(agl_exceptions); i++)
 221                 if (!strcmp(agl_exceptions[i][0], dst))
 222                         strcpy(dst, agl_exceptions[i][1]);
 223 }
 224
 225 static void trfn_ligput(char *c)
 226 {
 227         char *dst = strlen(c) == 2 ? trfn_ligs2 : trfn_ligs;
 228         sprintf(strchr(dst, '\0'), "%s ", c);
 229 }
 230
 231 static void trfn_lig(char *c)
 232 {
 233         int i;
 234         for (i = 0; i < LEN(agl_exceptions); i++)
 235                 if (!strcmp(agl_exceptions[i][1], c))
 236                         return;
 237         if (c[0] && c[1] && strlen(c) > utf8len((unsigned char) c[0])) {
 238                 trfn_ligput(c);
 239         } else {
 240                 for (i = 0; i < LEN(ligs_utf8); i++)
 241                         if (!strcmp(ligs_utf8[i][0], c))
 242                                 trfn_ligput(ligs_utf8[i][1]);
 243         }
 244 }
 245
 246 static int trfn_type(char *s, int lly, int ury)
 247 {
 248         int typ = 0;
 249         int c = !s[0] || s[1] ? 0 : (unsigned char) *s;
 250         if (c == 't' && !trfn_asc)
 251                 trfn_asc = ury;
 252         if ((c == 'g' || c == 'j' || c == 'p' || c == 'q' || c == 'y') &&
 253                         (!trfn_desc || trfn_desc < lly))
 254                 trfn_desc = lly;
 255         if (!trfn_desc || !trfn_asc) {
 256                 if (c > 0 && c < 128)
 257                         return ctype_ascii[c];
 258                 return 3;
 259         }
 260         if (!trfn_desc || lly <= trfn_desc)
 261                 typ |= 1;
 262         if (!trfn_asc || ury >= trfn_asc)
 263                 typ |= 2;
 264         return typ;
 265 }
 266
 267 /* n is the position and u is the unicode codepoint */
 268 void trfn_char(char *psname, int n, int u, int wid,
 269                 int llx, int lly, int urx, int ury)
 270 {
 271         char uc[GNLEN];                 /* mapping unicode character */
 272         char **a_tr;                    /* troff character names */
 273         char pos[GNLEN] = "";           /* postscript character position/name */
 274         int typ;                        /* character type */
 275         /* initializing character attributes */
 276         if (trfn_name(uc, psname, u))
 277                 strcpy(uc, "---");
 278         trfn_aglexceptions(uc);
 279         if (trfn_pos && n >= 0 && n < 256)
 280                 sprintf(pos, "%d", n);
 281         if (trfn_pos && n < 0 && !uc[1] && uc[0] >= 32 && uc[0] <= 125)
 282                 if (!strchr(psname, '.'))
 283                         sprintf(pos, "%d", uc[0]);
 284         typ = trfn_type(!strchr(psname, '.') ? uc : "", lly, ury);
 285         if (!trfn_swid && (!strcmp(" ", uc) || !strcmp(" ", uc)))
 286                 trfn_swid = WX(wid);
 287         /* printing troff charset */
 288         if (strchr(uc, ' '))    /* space not allowed in char names */
 289                 strcpy(uc, "---");
 290         if (strcmp("---", uc))
 291                 trfn_lig(uc);
 292         sbuf_printf(sbuf_char, "char %s\t%d", uc, WX(wid));
 293         if (trfn_bbox && (llx || lly || urx || ury))
 294                 sbuf_printf(sbuf_char, ",%d,%d,%d,%d",
 295                         WX(llx), WX(lly), WX(urx), WX(ury));
 296         sbuf_printf(sbuf_char, "\t%d\t%s\t%s\n", typ, psname, pos);
 297         a_tr = tab_get(tab_alts, uc);
 298         while (a_tr && *a_tr)
 299                 sbuf_printf(sbuf_char, "char %s\t\"\n", *a_tr++);
 300 }
 301
 302 void trfn_kern(char *c1, char *c2, int x)
 303 {
 304         if (WX(x) && abs(WX(x)) >= trfn_kmin)
 305                 sbuf_printf(sbuf_kern, "kern %s\t%s\t%d\n", c1, c2, WX(x));
 306 }
 307
 308 void trfn_trfont(char *name)
 309 {
 310         if (!trfn_trname[0])
 311                 strcpy(trfn_trname, name);
 312 }
 313
 314 void trfn_psfont(char *name)
 315 {
 316         if (!trfn_psname[0])
 317                 strcpy(trfn_psname, name);
 318 }
 319
 320 void trfn_pspath(char *name)
 321 {
 322         if (!trfn_path[0])
 323                 strcpy(trfn_path, name);
 324 }
 325
 326 void trfn_print(void)
 327 {
 328         if (trfn_trname[0])
 329                 printf("name %s\n", trfn_trname);
 330         if (trfn_psname[0])
 331                 printf("fontname %s\n", trfn_psname);
 332         if (trfn_path[0])
 333                 printf("fontpath %s\n", trfn_path);
 334         printf("spacewidth %d\n", trfn_swid);
 335         if (!trfn_noligs)
 336                 printf("ligatures %s%s0\n", trfn_ligs, trfn_ligs2);
 337         if (trfn_special)
 338                 printf("special\n");
 339         fputs(sbuf_buf(sbuf_char), stdout);
 340         fputs(sbuf_buf(sbuf_kern), stdout);
 341 }
 342
 343 void trfn_init(int res, int spc, int kmin, int bbox, int ligs, int pos)
 344 {
 345         int i;
 346         trfn_div = 72000 / res;
 347         trfn_special = spc;
 348         trfn_kmin = kmin;
 349         trfn_bbox = bbox;
 350         trfn_noligs = !ligs;
 351         trfn_pos = pos;
 352         sbuf_char = sbuf_make();
 353         sbuf_kern = sbuf_make();
 354         tab_agl = tab_alloc(LEN(agl));
 355         for (i = 0; i < LEN(agl); i++)
 356                 tab_put(tab_agl, agl[i][0], agl[i][1]);
 357         tab_alts = tab_alloc(LEN(alts));
 358         for (i = 0; i < LEN(alts); i++)
 359                 tab_put(tab_alts, alts[i][0], alts[i] + 1);
 360 }
 361
 362 void trfn_done(void)
 363 {
 364         sbuf_free(sbuf_char);
 365         sbuf_free(sbuf_kern);
 366         tab_free(tab_alts);
 367         if (tab_agl)
 368                 tab_free(tab_agl);
 369 }