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_ch.h"
   9
  10 #define WX(w)           (((w) < 0 ? (w) - trfn_div / 2 : (w) + trfn_div / 2) / trfn_div)
  11 #define LEN(a)          ((sizeof(a) / sizeof((a)[0])))
  12 #define HEXDIGS         "0123456789abcdef"
  13 #define NCHAR           8       /* number of characters per glyph */
  14 #define GNLEN           64      /* glyph name length */
  15 #define AGLLEN          8192    /* adobe glyphlist length */
  16
  17 static struct sbuf sbuf_char;   /* characters */
  18 static struct sbuf sbuf_kern;   /* kerning pairs */
  19 static int trfn_div;            /* divisor of widths */
  20 static int trfn_swid;           /* space width */
  21 static int trfn_special;        /* special flag */
  22 static int trfn_kmin;           /* minimum kerning value */
  23 static int trfn_bbox;           /* include bounding box */
  24 static char trfn_ligs[8192];    /* font ligatures */
  25 static char trfn_trname[256];   /* font troff name */
  26 static char trfn_psname[256];   /* font ps name */
  27 /* character type */
  28 static int trfn_asc;            /* minimum height of glyphs with ascender */
  29 static int trfn_desc;           /* minimum depth of glyphs with descender */
  30
  31 /* adobe glyphlist mapping */
  32 static char agl_key[AGLLEN][GNLEN];
  33 static char agl_val[AGLLEN][GNLEN];
  34 static int agl_n;
  35
  36 /* lookup tables */
  37 static struct tab *tab_agl;
  38 static struct tab *tab_alts;
  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;
  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                 l = 0;
  88         }
  89         while (l--)
  90                 *(*d)++ = 0x80 | ((c >> (l * 6)) & 0x3f);
  91         **d = '\0';
  92 }
  93
  94 static int hexval(char *s, int len)
  95 {
  96         char *digs = HEXDIGS;
  97         int n = 0;
  98         int i;
  99         for (i = 0; i < len; i++) {
 100                 if (s[i] && strchr(digs, tolower(s[i])))
 101                         n = n * 16 + (strchr(digs, tolower(s[i])) - digs);
 102                 else
 103                         break;
 104         }
 105         return len == 1 ? n << 4 : n;
 106 }
 107
 108 static int agl_read(char *path)
 109 {
 110         FILE *fin = fopen(path, "r");
 111         char ln[GNLEN];
 112         char val[GNLEN];
 113         char *s, *d;
 114         int i;
 115         if (!fin)
 116                 return 1;
 117         while (fgets(ln, sizeof(ln), fin)) {
 118                 s = strchr(ln, ';');
 119                 if (ln[0] == '#' || !s)
 120                         continue;
 121                 *s++ = '\0';
 122                 d = val;
 123                 while (s && *s) {
 124                         while (*s == ' ')
 125                                 s++;
 126                         utf8put(&d, hexval(s, 6));
 127                         s = strchr(s, ' ');
 128                 }
 129                 *d = '\0';
 130                 strcpy(agl_key[agl_n], ln);
 131                 strcpy(agl_val[agl_n], val);
 132                 agl_n++;
 133         }
 134         fclose(fin);
 135         tab_agl = tab_alloc(agl_n);
 136         for (i = 0; i < agl_n; i++)
 137                 tab_put(tab_agl, agl_key[i], agl_val[i]);
 138         return 0;
 139 }
 140
 141 static char *agl_map(char *s)
 142 {
 143         return tab_get(tab_agl, s);
 144 }
 145
 146 static int achar_map(char *name)
 147 {
 148         int i;
 149         for (i = 0; i < LEN(achars); i++) {
 150                 struct achar *a = &achars[i];
 151                 if (!strncmp(a->name, name, strlen(a->name))) {
 152                         char *postfix = name + strlen(a->name);
 153                         if (!*postfix)
 154                                 return a->c;
 155                         if (!strcmp("isolated", postfix))
 156                                 return a->s ? a->s : a->c;
 157                         if (!strcmp("initial", postfix))
 158                                 return a->i ? a->i : a->c;
 159                         if (!strcmp("medial", postfix))
 160                                 return a->m ? a->m : a->c;
 161                         if (!strcmp("final", postfix))
 162                                 return a->f ? a->f : a->c;
 163                 }
 164         }
 165         return 0;
 166 }
 167
 168 static int achar_shape(int c, int pjoin, int njoin)
 169 {
 170         int i;
 171         for (i = 0; i < LEN(achars); i++) {
 172                 struct achar *a = &achars[i];
 173                 if (a->c == c) {
 174                         if (!pjoin && !njoin)
 175                                 return a->c;
 176                         if (!pjoin && njoin)
 177                                 return a->i ? a->i : a->c;
 178                         if (pjoin && njoin)
 179                                 return a->m ? a->m : a->c;
 180                         if (pjoin && !njoin)
 181                                 return a->f ? a->f : a->c;
 182                 }
 183         }
 184         return c;
 185 }
 186
 187 static void ashape(char *str, char *ext)
 188 {
 189         int s[NCHAR];
 190         char *src = str;
 191         int i, l;
 192         int bjoin = !strcmp(".medi", ext) || !strcmp(".fina", ext);
 193         int ejoin = !strcmp(".medi", ext) || !strcmp(".init", ext);
 194         for (l = 0; l < NCHAR && *src; l++)
 195                 s[l] = utf8get(&src);
 196         for (i = 0; i < l; i++)
 197                 s[i] = achar_shape(s[i], i > 0 || bjoin, i < l - 1 || ejoin);
 198         for (i = 0; i < l; i++)
 199                 utf8put(&str, s[i]);
 200 }
 201
 202 /* find the utf-8 name of src with the given unicode codepoint */
 203 static int trfn_name(char *dst, char *src, int codepoint)
 204 {
 205         char ch[GNLEN];
 206         char *d = dst;
 207         char *s;
 208         int i;
 209         if (codepoint) {
 210                 utf8put(&dst, codepoint);
 211                 return 0;
 212         }
 213         if (!src || src[0] == '.')
 214                 return 1;
 215         while (*src && *src != '.') {
 216                 s = ch;
 217                 if (src[0] == '_')
 218                         src++;
 219                 while (*src && *src != '_' && *src != '.')
 220                         *s++ = *src++;
 221                 *s = '\0';
 222                 if (agl_map(ch)) {
 223                         strcpy(d, agl_map(ch));
 224                         for (i = 0; i < LEN(agl_exceptions); i++) {
 225                                 if (!strcmp(agl_exceptions[i][0], d)) {
 226                                         strcpy(d, agl_exceptions[i][1]);
 227                                         break;
 228                                 }
 229                         }
 230                         d = strchr(d, '\0');
 231                 } else if (ch[0] == 'u' && ch[1] == 'n' && ch[2] == 'i') {
 232                         for (i = 0; strlen(ch + 3 + 4 * i) >= 4; i++)
 233                                 utf8put(&d, hexval(ch + 3 + 4 * i, 4));
 234                 } else if (ch[0] == 'u' && ch[1] && strchr(HEXDIGS, tolower(ch[1]))) {
 235                         utf8put(&d, hexval(ch + 1, 6));
 236                 } else if (achar_map(ch)) {
 237                         utf8put(&d, achar_map(ch));
 238                 } else {
 239                         return 1;
 240                 }
 241         }
 242         ashape(dst, src);
 243         return *src && strcmp(src, ".medi") && strcmp(src, ".fina") &&
 244                         strcmp(src, ".init") && strcmp(src, ".isol");
 245 }
 246
 247 static void trfn_lig(char *c)
 248 {
 249         int i;
 250         for (i = 0; i < LEN(ligs_exceptions); i++)
 251                 if (!strcmp(ligs_exceptions[i], c))
 252                         return;
 253         if (c[0] && c[1] && strlen(c) > utf8len((unsigned char) c[0])) {
 254                 sprintf(strchr(trfn_ligs, '\0'), "%s ", c);
 255         } else {
 256                 for (i = 0; i < LEN(ligs_utf8); i++)
 257                         if (!strcmp(ligs_utf8[i][0], c))
 258                                 sprintf(strchr(trfn_ligs, '\0'),
 259                                         "%s ", ligs_utf8[i][1]);
 260         }
 261 }
 262
 263 static int trfn_type(char *s, int lly, int ury)
 264 {
 265         int typ = 0;
 266         int c = !s[0] || s[1] ? 0 : (unsigned char) *s;
 267         if (c == 't' && !trfn_asc)
 268                 trfn_asc = ury;
 269         if ((c == 'g' || c == 'j' || c == 'p' || c == 'q' || c == 'y') &&
 270                         (!trfn_desc || trfn_desc < lly))
 271                 trfn_desc = lly;
 272         if (!trfn_desc || !trfn_asc) {
 273                 if (c > 0 && c < 128)
 274                         return ctype_ascii[c];
 275                 return 3;
 276         }
 277         if (!trfn_desc || lly <= trfn_desc)
 278                 typ |= 1;
 279         if (!trfn_asc || ury >= trfn_asc)
 280                 typ |= 2;
 281         return typ;
 282 }
 283
 284 /* n is the position and u is the unicode codepoint */
 285 void trfn_char(char *psname, int n, int u, int wid,
 286                 int llx, int lly, int urx, int ury)
 287 {
 288         char uc[GNLEN];                 /* mapping unicode character */
 289         char **a_tr;                    /* troff character names */
 290         char pos[GNLEN] = "";           /* postscript character position/name */
 291         int typ;                        /* character type */
 292         /* initializing character attributes */
 293         if (trfn_name(uc, psname, u))
 294                 strcpy(uc, "---");
 295         if (n >= 0 && n < 256)
 296                 sprintf(pos, "%d", n);
 297         if (n < 0 && !uc[1] && uc[0] >= 32 && uc[0] <= 125)
 298                 if (!strchr(psname, '.'))
 299                         sprintf(pos, "%d", uc[0]);
 300         typ = trfn_type(!strchr(psname, '.') ? uc : "", lly, ury);
 301         /* printing troff charset */
 302         if (strchr(uc, ' ')) {  /* space not allowed in char names */
 303                 if (!trfn_swid && !strcmp(" ", uc))
 304                         trfn_swid = WX(wid);
 305                 return;
 306         }
 307         if (strcmp("---", uc))
 308                 trfn_lig(uc);
 309         sbuf_printf(&sbuf_char, "char %s\t%d", uc, WX(wid));
 310         if (trfn_bbox && (llx || lly || urx || ury))
 311                 sbuf_printf(&sbuf_char, ",%d,%d,%d,%d",
 312                         WX(llx), WX(lly), WX(urx), WX(ury));
 313         sbuf_printf(&sbuf_char, "\t%d\t%s\t%s\n", typ, psname, pos);
 314         a_tr = tab_get(tab_alts, uc);
 315         while (a_tr && *a_tr)
 316                 sbuf_printf(&sbuf_char, "char %s\t\"\n", *a_tr++);
 317 }
 318
 319 void trfn_kern(char *c1, char *c2, int x)
 320 {
 321         if (WX(x) && abs(WX(x)) >= trfn_kmin)
 322                 sbuf_printf(&sbuf_kern, "kern %s\t%s\t%d\n", c1, c2, WX(x));
 323 }
 324
 325 void trfn_trfont(char *name)
 326 {
 327         if (!trfn_trname[0])
 328                 strcpy(trfn_trname, name);
 329 }
 330
 331 void trfn_psfont(char *name)
 332 {
 333         if (!trfn_psname[0])
 334                 strcpy(trfn_psname, name);
 335 }
 336
 337 void trfn_print(void)
 338 {
 339         if (trfn_trname[0])
 340                 printf("name %s\n", trfn_trname);
 341         if (trfn_psname[0])
 342                 printf("fontname %s\n", trfn_psname);
 343         printf("spacewidth %d\n", trfn_swid);
 344         printf("ligatures %s0\n", trfn_ligs);
 345         if (trfn_special)
 346                 printf("special\n");
 347         printf("%s", sbuf_buf(&sbuf_char));
 348         printf("%s", sbuf_buf(&sbuf_kern));
 349 }
 350
 351 void trfn_init(int res, int spc, int kmin, int bbox)
 352 {
 353         int i;
 354         trfn_div = 7200 / res;
 355         trfn_special = spc;
 356         trfn_kmin = kmin;
 357         trfn_bbox = bbox;
 358         if (agl_read("glyphlist.txt"))
 359                 fprintf(stderr, "mktrfn: could not open glyphlist.txt\n");
 360         sbuf_init(&sbuf_char);
 361         sbuf_init(&sbuf_kern);
 362         tab_alts = tab_alloc(LEN(alts));
 363         for (i = 0; i < LEN(alts); i++)
 364                 tab_put(tab_alts, alts[i][0], alts[i] + 1);
 365 }
 366
 367 void trfn_done(void)
 368 {
 369         sbuf_done(&sbuf_char);
 370         sbuf_done(&sbuf_kern);
 371         tab_free(tab_alts);
 372         if (tab_agl)
 373                 tab_free(tab_agl);
 374 }