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