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 / 2 : (w) + trfn_div / 2) / 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 */
  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 char trfn_ligs[8192];    /* font ligatures */
  27 static char trfn_trname[256];   /* font troff name */
  28 static char trfn_psname[256];   /* font ps name */
  29 /* character type */
  30 static int trfn_asc;            /* minimum height of glyphs with ascender */
  31 static int trfn_desc;           /* minimum depth of glyphs with descender */
  32
  33 /* lookup tables */
  34 static struct tab *tab_agl;     /* adobe glyph list table */
  35 static struct tab *tab_alts;    /* character aliases table */
  36
  37 static int utf8len(int c)
  38 {
  39         if (c > 0 && c <= 0x7f)
  40                 return 1;
  41         if (c >= 0xfc)
  42                 return 6;
  43         if (c >= 0xf8)
  44                 return 5;
  45         if (c >= 0xf0)
  46                 return 4;
  47         if (c >= 0xe0)
  48                 return 3;
  49         if (c >= 0xc0)
  50                 return 2;
  51         return c != 0;
  52 }
  53
  54 static int utf8get(char **src)
  55 {
  56         int result;
  57         int l = 1;
  58         char *s = *src;
  59         if (~((unsigned char) **src) & 0xc0)
  60                 return (unsigned char) *(*src)++;
  61         while (l < 6 && (unsigned char) *s & (0x40 >> l))
  62                 l++;
  63         result = (0x3f >> l) & (unsigned char) *s++;
  64         while (l--)
  65                 result = (result << 6) | ((unsigned char) *s++ & 0x3f);
  66         *src = s;
  67         return result;
  68 }
  69
  70 static void utf8put(char **d, int c)
  71 {
  72         int l;
  73         if (c > 0xffff) {
  74                 *(*d)++ = 0xf0 | (c >> 18);
  75                 l = 3;
  76         } else if (c > 0x7ff) {
  77                 *(*d)++ = 0xe0 | (c >> 12);
  78                 l = 2;
  79         } else if (c > 0x7f) {
  80                 *(*d)++ = 0xc0 | (c >> 6);
  81                 l = 1;
  82         } else {
  83                 *(*d)++ = c > 0 ? c : ' ';
  84                 l = 0;
  85         }
  86         while (l--)
  87                 *(*d)++ = 0x80 | ((c >> (l * 6)) & 0x3f);
  88         **d = '\0';
  89 }
  90
  91 static int hexval(char *s, int len)
  92 {
  93         char *digs = HEXDIGS;
  94         int n = 0;
  95         int i;
  96         for (i = 0; i < len; i++) {
  97                 if (s[i] && strchr(digs, s[i]))
  98                         n = n * 16 + (strchr(digs, s[i]) - digs);
  99                 else
 100                         break;
 101         }
 102         return i < 4 ? -1 : n;
 103 }
 104
 105 static int agl_map(char *d, char *s)
 106 {
 107         char *u = tab_get(tab_agl, s);  /* unicode code point like "FB8E" */
 108         if (!u)
 109                 return 1;
 110         while (u && *u) {
 111                 while (*u == ' ')
 112                         u++;
 113                 utf8put(&d, hexval(u, 6));
 114                 u = strchr(u, ' ');
 115         }
 116         *d = '\0';
 117         return 0;
 118 }
 119
 120 static int achar_map(char *name)
 121 {
 122         int i;
 123         for (i = 0; i < LEN(achars); i++) {
 124                 struct achar *a = &achars[i];
 125                 if (!strncmp(a->name, name, strlen(a->name))) {
 126                         char *postfix = name + strlen(a->name);
 127                         if (!*postfix)
 128                                 return a->c;
 129                         if (!strcmp("isolated", postfix))
 130                                 return a->s ? a->s : a->c;
 131                         if (!strcmp("initial", postfix))
 132                                 return a->i ? a->i : a->c;
 133                         if (!strcmp("medial", postfix))
 134                                 return a->m ? a->m : a->c;
 135                         if (!strcmp("final", postfix))
 136                                 return a->f ? a->f : a->c;
 137                 }
 138         }
 139         return 0;
 140 }
 141
 142 static int achar_shape(int c, int pjoin, int njoin)
 143 {
 144         int i;
 145         for (i = 0; i < LEN(achars); i++) {
 146                 struct achar *a = &achars[i];
 147                 if (a->c == c) {
 148                         if (!pjoin && !njoin)
 149                                 return a->c;
 150                         if (!pjoin && njoin)
 151                                 return a->i ? a->i : a->c;
 152                         if (pjoin && njoin)
 153                                 return a->m ? a->m : a->c;
 154                         if (pjoin && !njoin)
 155                                 return a->f ? a->f : a->c;
 156                 }
 157         }
 158         return c;
 159 }
 160
 161 static void ashape(char *str, char *ext)
 162 {
 163         int s[NCHAR];
 164         char *src = str;
 165         int i, l;
 166         int bjoin = !strcmp(".medi", ext) || !strcmp(".fina", ext);
 167         int ejoin = !strcmp(".medi", ext) || !strcmp(".init", ext);
 168         for (l = 0; l < NCHAR && *src; l++)
 169                 s[l] = utf8get(&src);
 170         for (i = 0; i < l; i++)
 171                 s[i] = achar_shape(s[i], i > 0 || bjoin, i < l - 1 || ejoin);
 172         for (i = 0; i < l; i++)
 173                 utf8put(&str, s[i]);
 174 }
 175
 176 /* find the utf-8 name of src with the given unicode codepoint */
 177 static int trfn_name(char *dst, char *src, int codepoint)
 178 {
 179         char ch[GNLEN];
 180         char *d = dst;
 181         char *s;
 182         int i;
 183         if (codepoint) {
 184                 utf8put(&dst, codepoint);
 185                 return 0;
 186         }
 187         if (!src || src[0] == '.')
 188                 return 1;
 189         while (*src && *src != '.') {
 190                 s = ch;
 191                 if (src[0] == '_')
 192                         src++;
 193                 while (*src && *src != '_' && *src != '.')
 194                         *s++ = *src++;
 195                 *s = '\0';
 196                 if (!agl_map(d, ch)) {
 197                         d = strchr(d, '\0');
 198                 } else if (ch[0] == 'u' && ch[1] == 'n' &&
 199                                 ch[2] == 'i' && hexval(ch + 3, 4) > 0) {
 200                         for (i = 0; strlen(ch + 3 + 4 * i) >= 4; i++)
 201                                 utf8put(&d, hexval(ch + 3 + 4 * i, 4));
 202                 } else if (ch[0] == 'u' && hexval(ch + 1, 4) > 0) {
 203                         utf8put(&d, hexval(ch + 1, 6));
 204                 } else if (achar_map(ch)) {
 205                         utf8put(&d, achar_map(ch));
 206                 } else {
 207                         return 1;
 208                 }
 209         }
 210         for (i = 0; i < LEN(agl_exceptions); i++) {
 211                 if (!strcmp(agl_exceptions[i][0], dst)) {
 212                         strcpy(dst, agl_exceptions[i][1]);
 213                         break;
 214                 }
 215         }
 216         ashape(dst, src);
 217         return *src && strcmp(src, ".medi") && strcmp(src, ".fina") &&
 218                         strcmp(src, ".init") && strcmp(src, ".isol");
 219 }
 220
 221 static void trfn_lig(char *c)
 222 {
 223         int i;
 224         for (i = 0; i < LEN(agl_exceptions); i++)
 225                 if (!strcmp(agl_exceptions[i][1], c))
 226                         return;
 227         if (c[0] && c[1] && strlen(c) > utf8len((unsigned char) c[0])) {
 228                 sprintf(strchr(trfn_ligs, '\0'), "%s ", c);
 229         } else {
 230                 for (i = 0; i < LEN(ligs_utf8); i++)
 231                         if (!strcmp(ligs_utf8[i][0], c))
 232                                 sprintf(strchr(trfn_ligs, '\0'),
 233                                         "%s ", ligs_utf8[i][1]);
 234         }
 235 }
 236
 237 static int trfn_type(char *s, int lly, int ury)
 238 {
 239         int typ = 0;
 240         int c = !s[0] || s[1] ? 0 : (unsigned char) *s;
 241         if (c == 't' && !trfn_asc)
 242                 trfn_asc = ury;
 243         if ((c == 'g' || c == 'j' || c == 'p' || c == 'q' || c == 'y') &&
 244                         (!trfn_desc || trfn_desc < lly))
 245                 trfn_desc = lly;
 246         if (!trfn_desc || !trfn_asc) {
 247                 if (c > 0 && c < 128)
 248                         return ctype_ascii[c];
 249                 return 3;
 250         }
 251         if (!trfn_desc || lly <= trfn_desc)
 252                 typ |= 1;
 253         if (!trfn_asc || ury >= trfn_asc)
 254                 typ |= 2;
 255         return typ;
 256 }
 257
 258 /* n is the position and u is the unicode codepoint */
 259 void trfn_char(char *psname, int n, int u, int wid,
 260                 int llx, int lly, int urx, int ury)
 261 {
 262         char uc[GNLEN];                 /* mapping unicode character */
 263         char **a_tr;                    /* troff character names */
 264         char pos[GNLEN] = "";           /* postscript character position/name */
 265         int typ;                        /* character type */
 266         /* initializing character attributes */
 267         if (trfn_name(uc, psname, u))
 268                 strcpy(uc, "---");
 269         if (n >= 0 && n < 256)
 270                 sprintf(pos, "%d", n);
 271         if (n < 0 && !uc[1] && uc[0] >= 32 && uc[0] <= 125)
 272                 if (!strchr(psname, '.'))
 273                         sprintf(pos, "%d", uc[0]);
 274         typ = trfn_type(!strchr(psname, '.') ? uc : "", lly, ury);
 275         /* printing troff charset */
 276         if (strchr(uc, ' ')) {  /* space not allowed in char names */
 277                 if (!trfn_swid && !strcmp(" ", uc))
 278                         trfn_swid = WX(wid);
 279                 return;
 280         }
 281         if (strcmp("---", uc))
 282                 trfn_lig(uc);
 283         sbuf_printf(&sbuf_char, "char %s\t%d", uc, WX(wid));
 284         if (trfn_bbox && (llx || lly || urx || ury))
 285                 sbuf_printf(&sbuf_char, ",%d,%d,%d,%d",
 286                         WX(llx), WX(lly), WX(urx), WX(ury));
 287         sbuf_printf(&sbuf_char, "\t%d\t%s\t%s\n", typ, psname, pos);
 288         a_tr = tab_get(tab_alts, uc);
 289         while (a_tr && *a_tr)
 290                 sbuf_printf(&sbuf_char, "char %s\t\"\n", *a_tr++);
 291 }
 292
 293 void trfn_kern(char *c1, char *c2, int x)
 294 {
 295         if (WX(x) && abs(WX(x)) >= trfn_kmin)
 296                 sbuf_printf(&sbuf_kern, "kern %s\t%s\t%d\n", c1, c2, WX(x));
 297 }
 298
 299 void trfn_trfont(char *name)
 300 {
 301         if (!trfn_trname[0])
 302                 strcpy(trfn_trname, name);
 303 }
 304
 305 void trfn_psfont(char *name)
 306 {
 307         if (!trfn_psname[0])
 308                 strcpy(trfn_psname, name);
 309 }
 310
 311 void trfn_print(void)
 312 {
 313         if (trfn_trname[0])
 314                 printf("name %s\n", trfn_trname);
 315         if (trfn_psname[0])
 316                 printf("fontname %s\n", trfn_psname);
 317         printf("spacewidth %d\n", trfn_swid);
 318         if (!trfn_noligs)
 319                 printf("ligatures %s0\n", trfn_ligs);
 320         if (trfn_special)
 321                 printf("special\n");
 322         printf("%s", sbuf_buf(&sbuf_char));
 323         printf("%s", sbuf_buf(&sbuf_kern));
 324 }
 325
 326 void trfn_init(int res, int spc, int kmin, int bbox, int ligs)
 327 {
 328         int i;
 329         trfn_div = 7200 / res;
 330         trfn_special = spc;
 331         trfn_kmin = kmin;
 332         trfn_bbox = bbox;
 333         trfn_noligs = !ligs;
 334         sbuf_init(&sbuf_char);
 335         sbuf_init(&sbuf_kern);
 336         tab_agl = tab_alloc(LEN(agl));
 337         for (i = 0; i < LEN(agl); i++)
 338                 tab_put(tab_agl, agl[i][0], agl[i][1]);
 339         tab_alts = tab_alloc(LEN(alts));
 340         for (i = 0; i < LEN(alts); i++)
 341                 tab_put(tab_alts, alts[i][0], alts[i] + 1);
 342 }
 343
 344 void trfn_done(void)
 345 {
 346         sbuf_done(&sbuf_char);
 347         sbuf_done(&sbuf_kern);
 348         tab_free(tab_alts);
 349         if (tab_agl)
 350                 tab_free(tab_agl);
 351 }