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 int trfn_pos;            /* include glyph positions */
  27 static char trfn_ligs[8192];    /* font ligatures */
  28 static char trfn_trname[256];   /* font troff name */
  29 static char trfn_psname[256];   /* font ps name */
  30 /* character type */
  31 static int trfn_asc;            /* minimum height of glyphs with ascender */
  32 static int trfn_desc;           /* minimum depth of glyphs with descender */
  33
  34 /* lookup tables */
  35 static struct tab *tab_agl;     /* adobe glyph list table */
  36 static struct tab *tab_alts;    /* character aliases table */
  37
  38 static int utf8len(int c)
  39 {
  40         if (c > 0 && c <= 0x7f)
  41                 return 1;
  42         if (c >= 0xfc)
  43                 return 6;
  44         if (c >= 0xf8)
  45                 return 5;
  46         if (c >= 0xf0)
  47                 return 4;
  48         if (c >= 0xe0)
  49                 return 3;
  50         if (c >= 0xc0)
  51                 return 2;
  52         return c != 0;
  53 }
  54
  55 static int utf8get(char **src)
  56 {
  57         int result;
  58         int l = 1;
  59         char *s = *src;
  60         if (~((unsigned char) **src) & 0xc0)
  61                 return (unsigned char) *(*src)++;
  62         while (l < 6 && (unsigned char) *s & (0x40 >> l))
  63                 l++;
  64         result = (0x3f >> l) & (unsigned char) *s++;
  65         while (l--)
  66                 result = (result << 6) | ((unsigned char) *s++ & 0x3f);
  67         *src = s;
  68         return result;
  69 }
  70
  71 static void utf8put(char **d, int c)
  72 {
  73         int l;
  74         if (c > 0xffff) {
  75                 *(*d)++ = 0xf0 | (c >> 18);
  76                 l = 3;
  77         } else if (c > 0x7ff) {
  78                 *(*d)++ = 0xe0 | (c >> 12);
  79                 l = 2;
  80         } else if (c > 0x7f) {
  81                 *(*d)++ = 0xc0 | (c >> 6);
  82                 l = 1;
  83         } else {
  84                 *(*d)++ = c > 0 ? c : ' ';
  85                 l = 0;
  86         }
  87         while (l--)
  88                 *(*d)++ = 0x80 | ((c >> (l * 6)) & 0x3f);
  89         **d = '\0';
  90 }
  91
  92 static int hexval(char *s, int len)
  93 {
  94         char *digs = HEXDIGS;
  95         int n = 0;
  96         int i;
  97         for (i = 0; i < len; i++) {
  98                 if (s[i] && strchr(digs, s[i]))
  99                         n = n * 16 + (strchr(digs, s[i]) - digs);
 100                 else
 101                         break;
 102         }
 103         return i < 4 ? -1 : n;
 104 }
 105
 106 static int agl_map(char *d, char *s)
 107 {
 108         char *u = tab_get(tab_agl, s);  /* unicode code point like "FB8E" */
 109         if (!u)
 110                 return 1;
 111         while (u && *u) {
 112                 while (*u == ' ')
 113                         u++;
 114                 utf8put(&d, hexval(u, 6));
 115                 u = strchr(u, ' ');
 116         }
 117         *d = '\0';
 118         return 0;
 119 }
 120
 121 static int achar_map(char *name)
 122 {
 123         int i;
 124         for (i = 0; i < LEN(achars); i++) {
 125                 struct achar *a = &achars[i];
 126                 if (!strncmp(a->name, name, strlen(a->name))) {
 127                         char *postfix = name + strlen(a->name);
 128                         if (!*postfix)
 129                                 return a->c;
 130                         if (!strcmp("isolated", postfix))
 131                                 return a->s ? a->s : a->c;
 132                         if (!strcmp("initial", postfix))
 133                                 return a->i ? a->i : a->c;
 134                         if (!strcmp("medial", postfix))
 135                                 return a->m ? a->m : a->c;
 136                         if (!strcmp("final", postfix))
 137                                 return a->f ? a->f : a->c;
 138                 }
 139         }
 140         return 0;
 141 }
 142
 143 static int achar_shape(int c, int pjoin, int njoin)
 144 {
 145         int i;
 146         for (i = 0; i < LEN(achars); i++) {
 147                 struct achar *a = &achars[i];
 148                 if (a->c == c) {
 149                         if (!pjoin && !njoin)
 150                                 return a->c;
 151                         if (!pjoin && njoin)
 152                                 return a->i ? a->i : a->c;
 153                         if (pjoin && njoin)
 154                                 return a->m ? a->m : a->c;
 155                         if (pjoin && !njoin)
 156                                 return a->f ? a->f : a->c;
 157                 }
 158         }
 159         return c;
 160 }
 161
 162 static void ashape(char *str, char *ext)
 163 {
 164         int s[NCHAR];
 165         char *src = str;
 166         int i, l;
 167         int bjoin = !strcmp(".medi", ext) || !strcmp(".fina", ext);
 168         int ejoin = !strcmp(".medi", ext) || !strcmp(".init", ext);
 169         for (l = 0; l < NCHAR && *src; l++)
 170                 s[l] = utf8get(&src);
 171         for (i = 0; i < l; i++)
 172                 s[i] = achar_shape(s[i], i > 0 || bjoin, i < l - 1 || ejoin);
 173         for (i = 0; i < l; i++)
 174                 utf8put(&str, s[i]);
 175 }
 176
 177 /* find the utf-8 name of src with the given unicode codepoint */
 178 static int trfn_name(char *dst, char *src, int codepoint)
 179 {
 180         char ch[GNLEN];
 181         char *d = dst;
 182         char *s;
 183         int i;
 184         if (codepoint) {
 185                 utf8put(&dst, codepoint);
 186                 return 0;
 187         }
 188         if (!src || src[0] == '.')
 189                 return 1;
 190         while (*src && *src != '.') {
 191                 s = ch;
 192                 if (src[0] == '_')
 193                         src++;
 194                 while (*src && *src != '_' && *src != '.')
 195                         *s++ = *src++;
 196                 *s = '\0';
 197                 if (!agl_map(d, ch)) {
 198                         d = strchr(d, '\0');
 199                 } else if (ch[0] == 'u' && ch[1] == 'n' &&
 200                                 ch[2] == 'i' && hexval(ch + 3, 4) > 0) {
 201                         for (i = 0; strlen(ch + 3 + 4 * i) >= 4; i++)
 202                                 utf8put(&d, hexval(ch + 3 + 4 * i, 4));
 203                 } else if (ch[0] == 'u' && hexval(ch + 1, 4) > 0) {
 204                         utf8put(&d, hexval(ch + 1, 6));
 205                 } else if (achar_map(ch)) {
 206                         utf8put(&d, achar_map(ch));
 207                 } else {
 208                         return 1;
 209                 }
 210         }
 211         for (i = 0; i < LEN(agl_exceptions); i++) {
 212                 if (!strcmp(agl_exceptions[i][0], dst)) {
 213                         strcpy(dst, agl_exceptions[i][1]);
 214                         break;
 215                 }
 216         }
 217         ashape(dst, src);
 218         return *src && strcmp(src, ".medi") && strcmp(src, ".fina") &&
 219                         strcmp(src, ".init") && strcmp(src, ".isol");
 220 }
 221
 222 static void trfn_lig(char *c)
 223 {
 224         int i;
 225         for (i = 0; i < LEN(agl_exceptions); i++)
 226                 if (!strcmp(agl_exceptions[i][1], c))
 227                         return;
 228         if (c[0] && c[1] && strlen(c) > utf8len((unsigned char) c[0])) {
 229                 sprintf(strchr(trfn_ligs, '\0'), "%s ", c);
 230         } else {
 231                 for (i = 0; i < LEN(ligs_utf8); i++)
 232                         if (!strcmp(ligs_utf8[i][0], c))
 233                                 sprintf(strchr(trfn_ligs, '\0'),
 234                                         "%s ", ligs_utf8[i][1]);
 235         }
 236 }
 237
 238 static int trfn_type(char *s, int lly, int ury)
 239 {
 240         int typ = 0;
 241         int c = !s[0] || s[1] ? 0 : (unsigned char) *s;
 242         if (c == 't' && !trfn_asc)
 243                 trfn_asc = ury;
 244         if ((c == 'g' || c == 'j' || c == 'p' || c == 'q' || c == 'y') &&
 245                         (!trfn_desc || trfn_desc < lly))
 246                 trfn_desc = lly;
 247         if (!trfn_desc || !trfn_asc) {
 248                 if (c > 0 && c < 128)
 249                         return ctype_ascii[c];
 250                 return 3;
 251         }
 252         if (!trfn_desc || lly <= trfn_desc)
 253                 typ |= 1;
 254         if (!trfn_asc || ury >= trfn_asc)
 255                 typ |= 2;
 256         return typ;
 257 }
 258
 259 /* n is the position and u is the unicode codepoint */
 260 void trfn_char(char *psname, int n, int u, int wid,
 261                 int llx, int lly, int urx, int ury)
 262 {
 263         char uc[GNLEN];                 /* mapping unicode character */
 264         char **a_tr;                    /* troff character names */
 265         char pos[GNLEN] = "";           /* postscript character position/name */
 266         int typ;                        /* character type */
 267         /* initializing character attributes */
 268         if (trfn_name(uc, psname, u))
 269                 strcpy(uc, "---");
 270         if (trfn_pos && n >= 0 && n < 256)
 271                 sprintf(pos, "%d", n);
 272         if (trfn_pos && n < 0 && !uc[1] && uc[0] >= 32 && uc[0] <= 125)
 273                 if (!strchr(psname, '.'))
 274                         sprintf(pos, "%d", uc[0]);
 275         typ = trfn_type(!strchr(psname, '.') ? uc : "", lly, ury);
 276         if (!trfn_swid && (!strcmp(" ", uc) || !strcmp(" ", uc)))
 277                 trfn_swid = WX(wid);
 278         /* printing troff charset */
 279         if (strchr(uc, ' '))    /* space not allowed in char names */
 280                 return;
 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, int pos)
 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         trfn_pos = pos;
 335         sbuf_init(&sbuf_char);
 336         sbuf_init(&sbuf_kern);
 337         tab_agl = tab_alloc(LEN(agl));
 338         for (i = 0; i < LEN(agl); i++)
 339                 tab_put(tab_agl, agl[i][0], agl[i][1]);
 340         tab_alts = tab_alloc(LEN(alts));
 341         for (i = 0; i < LEN(alts); i++)
 342                 tab_put(tab_alts, alts[i][0], alts[i] + 1);
 343 }
 344
 345 void trfn_done(void)
 346 {
 347         sbuf_done(&sbuf_char);
 348         sbuf_done(&sbuf_kern);
 349         tab_free(tab_alts);
 350         if (tab_agl)
 351                 tab_free(tab_agl);
 352 }