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_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 /* character type */
  32 static int trfn_asc;            /* minimum height of glyphs with ascender */
  33 static int trfn_desc;           /* minimum depth of glyphs with descender */
  34
  35 /* lookup tables */
  36 static struct tab *tab_agl;     /* adobe glyph list table */
  37 static struct tab *tab_alts;    /* character aliases table */
  38
  39 static int utf8len(int c)
  40 {
  41         if (c > 0 && c <= 0x7f)
  42                 return 1;
  43         if (c >= 0xfc)
  44                 return 6;
  45         if (c >= 0xf8)
  46                 return 5;
  47         if (c >= 0xf0)
  48                 return 4;
  49         if (c >= 0xe0)
  50                 return 3;
  51         if (c >= 0xc0)
  52                 return 2;
  53         return c != 0;
  54 }
  55
  56 static int utf8get(char **src)
  57 {
  58         int result;
  59         int l = 1;
  60         char *s = *src;
  61         if (~((unsigned char) **src) & 0xc0)
  62                 return (unsigned char) *(*src)++;
  63         while (l < 6 && (unsigned char) *s & (0x40 >> l))
  64                 l++;
  65         result = (0x3f >> l) & (unsigned char) *s++;
  66         while (l--)
  67                 result = (result << 6) | ((unsigned char) *s++ & 0x3f);
  68         *src = s;
  69         return result;
  70 }
  71
  72 static void utf8put(char **d, int c)
  73 {
  74         int l = 0;
  75         if (c > 0xffff) {
  76                 *(*d)++ = 0xf0 | (c >> 18);
  77                 l = 3;
  78         } else if (c > 0x7ff) {
  79                 *(*d)++ = 0xe0 | (c >> 12);
  80                 l = 2;
  81         } else if (c > 0x7f) {
  82                 *(*d)++ = 0xc0 | (c >> 6);
  83                 l = 1;
  84         } else {
  85                 *(*d)++ = c > 0 ? c : ' ';
  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         ashape(dst, src);
 212         return *src && strcmp(src, ".medi") && strcmp(src, ".fina") &&
 213                         strcmp(src, ".init") && strcmp(src, ".isol");
 214 }
 215
 216 static void trfn_aglexceptions(char *dst)
 217 {
 218         int i;
 219         for (i = 0; i < LEN(agl_exceptions); i++)
 220                 if (!strcmp(agl_exceptions[i][0], dst))
 221                         strcpy(dst, agl_exceptions[i][1]);
 222 }
 223
 224 static void trfn_ligput(char *c)
 225 {
 226         char *dst = strlen(c) == 2 ? trfn_ligs2 : trfn_ligs;
 227         sprintf(strchr(dst, '\0'), "%s ", c);
 228 }
 229
 230 static void trfn_lig(char *c)
 231 {
 232         int i;
 233         for (i = 0; i < LEN(agl_exceptions); i++)
 234                 if (!strcmp(agl_exceptions[i][1], c))
 235                         return;
 236         if (c[0] && c[1] && strlen(c) > utf8len((unsigned char) c[0])) {
 237                 trfn_ligput(c);
 238         } else {
 239                 for (i = 0; i < LEN(ligs_utf8); i++)
 240                         if (!strcmp(ligs_utf8[i][0], c))
 241                                 trfn_ligput(ligs_utf8[i][1]);
 242         }
 243 }
 244
 245 static int trfn_type(char *s, int lly, int ury)
 246 {
 247         int typ = 0;
 248         int c = !s[0] || s[1] ? 0 : (unsigned char) *s;
 249         if (c == 't' && !trfn_asc)
 250                 trfn_asc = ury;
 251         if ((c == 'g' || c == 'j' || c == 'p' || c == 'q' || c == 'y') &&
 252                         (!trfn_desc || trfn_desc < lly))
 253                 trfn_desc = lly;
 254         if (!trfn_desc || !trfn_asc) {
 255                 if (c > 0 && c < 128)
 256                         return ctype_ascii[c];
 257                 return 3;
 258         }
 259         if (!trfn_desc || lly <= trfn_desc)
 260                 typ |= 1;
 261         if (!trfn_asc || ury >= trfn_asc)
 262                 typ |= 2;
 263         return typ;
 264 }
 265
 266 /* n is the position and u is the unicode codepoint */
 267 void trfn_char(char *psname, int n, int u, int wid,
 268                 int llx, int lly, int urx, int ury)
 269 {
 270         char uc[GNLEN];                 /* mapping unicode character */
 271         char **a_tr;                    /* troff character names */
 272         char pos[GNLEN] = "";           /* postscript character position/name */
 273         int typ;                        /* character type */
 274         /* initializing character attributes */
 275         if (trfn_name(uc, psname, u))
 276                 strcpy(uc, "---");
 277         trfn_aglexceptions(uc);
 278         if (trfn_pos && n >= 0 && n < 256)
 279                 sprintf(pos, "%d", n);
 280         if (trfn_pos && n < 0 && !uc[1] && uc[0] >= 32 && uc[0] <= 125)
 281                 if (!strchr(psname, '.'))
 282                         sprintf(pos, "%d", uc[0]);
 283         typ = trfn_type(!strchr(psname, '.') ? uc : "", lly, ury);
 284         if (!trfn_swid && (!strcmp(" ", uc) || !strcmp(" ", uc)))
 285                 trfn_swid = WX(wid);
 286         /* printing troff charset */
 287         if (strchr(uc, ' '))    /* space not allowed in char names */
 288                 return;
 289         if (strcmp("---", uc))
 290                 trfn_lig(uc);
 291         sbuf_printf(sbuf_char, "char %s\t%d", uc, WX(wid));
 292         if (trfn_bbox && (llx || lly || urx || ury))
 293                 sbuf_printf(sbuf_char, ",%d,%d,%d,%d",
 294                         WX(llx), WX(lly), WX(urx), WX(ury));
 295         sbuf_printf(sbuf_char, "\t%d\t%s\t%s\n", typ, psname, pos);
 296         a_tr = tab_get(tab_alts, uc);
 297         while (a_tr && *a_tr)
 298                 sbuf_printf(sbuf_char, "char %s\t\"\n", *a_tr++);
 299 }
 300
 301 void trfn_kern(char *c1, char *c2, int x)
 302 {
 303         if (WX(x) && abs(WX(x)) >= trfn_kmin)
 304                 sbuf_printf(sbuf_kern, "kern %s\t%s\t%d\n", c1, c2, WX(x));
 305 }
 306
 307 void trfn_trfont(char *name)
 308 {
 309         if (!trfn_trname[0])
 310                 strcpy(trfn_trname, name);
 311 }
 312
 313 void trfn_psfont(char *name)
 314 {
 315         if (!trfn_psname[0])
 316                 strcpy(trfn_psname, name);
 317 }
 318
 319 void trfn_print(void)
 320 {
 321         if (trfn_trname[0])
 322                 printf("name %s\n", trfn_trname);
 323         if (trfn_psname[0])
 324                 printf("fontname %s\n", trfn_psname);
 325         printf("spacewidth %d\n", trfn_swid);
 326         if (!trfn_noligs)
 327                 printf("ligatures %s%s0\n", trfn_ligs, trfn_ligs2);
 328         if (trfn_special)
 329                 printf("special\n");
 330         fputs(sbuf_buf(sbuf_char), stdout);
 331         fputs(sbuf_buf(sbuf_kern), stdout);
 332 }
 333
 334 void trfn_init(int res, int spc, int kmin, int bbox, int ligs, int pos)
 335 {
 336         int i;
 337         trfn_div = 7200 / res;
 338         trfn_special = spc;
 339         trfn_kmin = kmin;
 340         trfn_bbox = bbox;
 341         trfn_noligs = !ligs;
 342         trfn_pos = pos;
 343         sbuf_char = sbuf_make();
 344         sbuf_kern = sbuf_make();
 345         tab_agl = tab_alloc(LEN(agl));
 346         for (i = 0; i < LEN(agl); i++)
 347                 tab_put(tab_agl, agl[i][0], agl[i][1]);
 348         tab_alts = tab_alloc(LEN(alts));
 349         for (i = 0; i < LEN(alts); i++)
 350                 tab_put(tab_alts, alts[i][0], alts[i] + 1);
 351 }
 352
 353 void trfn_done(void)
 354 {
 355         sbuf_free(sbuf_char);
 356         sbuf_free(sbuf_kern);
 357         tab_free(tab_alts);
 358         if (tab_agl)
 359                 tab_free(tab_agl);
 360 }