ref: 50918a0b71ee777bbb3b7ba7bebb590aeb86f151
author: Ali Gholami Rudi <ali@rudi.ir>
date: Fri Sep 6 18:15:36 EDT 2013
mktrfn: otfdump and improved afm support
--- /dev/null
+++ b/Makefile
@@ -1,0 +1,11 @@
+CC = cc
+CFLAGS = -O2 -Wall
+LDFLAGS =
+
+all: mktrfn
+%.o: %.c
+ $(CC) -c $(CFLAGS) $<
+mktrfn: mktrfn.o trfn.o sbuf.o tab.o
+ $(CC) -o $@ $^ $(LDFLAGS)
+clean:
+ rm -f *.o mktrfn
--- /dev/null
+++ b/mktrfn.c
@@ -1,0 +1,120 @@
+/*
+ * mktrfn - produce troff font descriptions
+ *
+ * Copyright (C) 2012-2013 Ali Gholami Rudi <ali at rudi dot ir>
+ *
+ * This program is released under the Modified BSD license.
+ */
+#include <ctype.h>
+#include <fcntl.h>
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include "trfn.h"
+
+#define TOKLEN 256
+
+static void otfdump_read(void)
+{+ char cmd[TOKLEN];
+ char name[TOKLEN];
+ char ch[TOKLEN];
+ char c1[TOKLEN], c2[TOKLEN];
+ char wid[TOKLEN];
+ while (scanf("%s", cmd) == 1) {+ if (!strcmp("name", cmd)) {+ scanf("%s", name);+ trfn_psfont(name);
+ }
+ if (!strcmp("char", cmd)) {+ scanf("%s width %s", ch, wid);+ trfn_char(ch, NULL, atoi(wid), -1);
+ }
+ if (!strcmp("kernpair", cmd)) {+ scanf("%s %s width %s", c1, c2, wid);+ trfn_kern(c1, c2, atoi(wid));
+ }
+ if (!strcmp("feature", cmd)) {+ scanf("%s substitution %s %s", name, c1, c2);+ }
+ }
+}
+
+static void afm_read(void)
+{+ char ch[TOKLEN], pos[TOKLEN];
+ char c1[TOKLEN], c2[TOKLEN];
+ char wid[TOKLEN];
+ char ln[1024];
+ while (fgets(ln, sizeof(ln), stdin)) {+ if (ln[0] == '#')
+ continue;
+ if (!strncmp("FontName ", ln, 8)) {+ sscanf(ln, "FontName %s", ch);
+ trfn_psfont(ch);
+ continue;
+ }
+ if (!strncmp("StartCharMetrics", ln, 16))+ break;
+ }
+ while (fgets(ln, sizeof(ln), stdin)) {+ if (ln[0] == '#')
+ continue;
+ if (!strncmp("EndCharMetrics", ln, 14))+ break;
+ if (sscanf(ln, "C %s ; WX %s ; N %s", pos, wid, ch) == 3)
+ trfn_char(ch, pos, atoi(wid), -1);
+ }
+ while (fgets(ln, sizeof(ln), stdin)) {+ if (ln[0] == '#')
+ continue;
+ if (!strncmp("StartKernPairs", ln, 14))+ break;
+ }
+ while (fgets(ln, sizeof(ln), stdin)) {+ if (ln[0] == '#')
+ continue;
+ if (!strncmp("EndKernPairs", ln, 12))+ break;
+ if (sscanf(ln, "KPX %s %s %s", c1, c2, wid) == 3)
+ trfn_kern(c1, c2, atoi(wid));
+ }
+}
+
+int main(int argc, char *argv[])
+{+ int afm = 0;
+ int i = 1;
+ int res = 720;
+ for (i = 1; i < argc && argv[i][0] == '-'; i++) {+ switch (argv[i][1]) {+ case 'a':
+ afm = 1;
+ break;
+ case 'o':
+ afm = 0;
+ break;
+ case 'r':
+ res = atoi(argv[i][2] ? argv[i] + 2 : argv[++i]);
+ break;
+ case 't':
+ trfn_trfont(argv[i][2] ? argv[i] + 2 : argv[++i]);
+ break;
+ case 'p':
+ trfn_psfont(argv[i][2] ? argv[i] + 2 : argv[++i]);
+ break;
+ default:
+ printf("usage: mktrfn -a -o -t name -p psname\n");+ return 0;
+ }
+ }
+ trfn_init(res);
+ if (afm)
+ afm_read();
+ else
+ otfdump_read();
+ trfn_print();
+ trfn_done();
+ return 0;
+}
--- /dev/null
+++ b/sbuf.c
@@ -1,0 +1,84 @@
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include "sbuf.h"
+
+#define MAX(a, b) ((a) < (b) ? (b) : (a))
+#define SBUF_SZ 512
+
+static void sbuf_extend(struct sbuf *sbuf, int amount)
+{+ char *s = sbuf->s;
+ sbuf->sz = (MAX(1, amount) + SBUF_SZ - 1) & ~(SBUF_SZ - 1);
+ sbuf->s = malloc(sbuf->sz);
+ if (sbuf->n)
+ memcpy(sbuf->s, s, sbuf->n);
+ free(s);
+}
+
+void sbuf_init(struct sbuf *sbuf)
+{+ memset(sbuf, 0, sizeof(*sbuf));
+ sbuf_extend(sbuf, SBUF_SZ);
+}
+
+void sbuf_add(struct sbuf *sbuf, int c)
+{+ if (sbuf->n + 2 >= sbuf->sz)
+ sbuf_extend(sbuf, sbuf->sz * 2);
+ sbuf->s[sbuf->n++] = c;
+}
+
+void sbuf_append(struct sbuf *sbuf, char *s)
+{+ int len = strlen(s);
+ if (sbuf->n + len + 1 >= sbuf->sz)
+ sbuf_extend(sbuf, sbuf->n + len + 1);
+ memcpy(sbuf->s + sbuf->n, s, len);
+ sbuf->n += len;
+}
+
+void sbuf_printf(struct sbuf *sbuf, char *s, ...)
+{+ char buf[1024];
+ va_list ap;
+ va_start(ap, s);
+ vsprintf(buf, s, ap);
+ va_end(ap);
+ sbuf_append(sbuf, buf);
+}
+
+void sbuf_putnl(struct sbuf *sbuf)
+{+ if (sbuf->n && sbuf->s[sbuf->n - 1] != '\n')
+ sbuf_add(sbuf, '\n');
+}
+
+int sbuf_empty(struct sbuf *sbuf)
+{+ return !sbuf->n;
+}
+
+char *sbuf_buf(struct sbuf *sbuf)
+{+ sbuf->s[sbuf->n] = '\0';
+ return sbuf->s;
+}
+
+int sbuf_len(struct sbuf *sbuf)
+{+ return sbuf->n;
+}
+
+/* shorten the sbuf */
+void sbuf_cut(struct sbuf *sbuf, int n)
+{+ if (sbuf->n > n)
+ sbuf->n = n;
+}
+
+void sbuf_done(struct sbuf *sbuf)
+{+ free(sbuf->s);
+}
--- /dev/null
+++ b/sbuf.h
@@ -1,0 +1,11 @@
+/* variable length string buffer */
+struct sbuf {+ char *s; /* allocated buffer */
+ int sz; /* buffer size */
+ int n; /* length of the string stored in s */
+};
+
+void sbuf_init(struct sbuf *sbuf);
+void sbuf_done(struct sbuf *sbuf);
+char *sbuf_buf(struct sbuf *sbuf);
+void sbuf_printf(struct sbuf *sbuf, char *s, ...);
--- /dev/null
+++ b/tab.c
@@ -1,0 +1,52 @@
+#include <stdlib.h>
+#include <string.h>
+#include "tab.h"
+
+struct tab {+ char **keys;
+ void **vals;
+ int n;
+ int *next;
+ int head[256];
+};
+
+struct tab *tab_alloc(int sz)
+{+ struct tab *tab = malloc(sizeof(*tab));
+ int i;
+ memset(tab, 0, sizeof(*tab));
+ tab->keys = malloc(sz * sizeof(tab->keys[0]));
+ tab->vals = malloc(sz * sizeof(tab->vals[0]));
+ tab->next = malloc(sz * sizeof(tab->next[0]));
+ for (i = 0; i < 256; i++)
+ tab->head[i] = -1;
+ return tab;
+}
+
+void tab_free(struct tab *tab)
+{+ free(tab->keys);
+ free(tab->vals);
+ free(tab->next);
+ free(tab);
+}
+
+void tab_put(struct tab *tab, char *k, void *v)
+{+ tab->keys[tab->n] = k;
+ tab->vals[tab->n] = v;
+ tab->next[tab->n] = tab->head[(unsigned char) k[0]];
+ tab->head[(unsigned char) k[0]] = tab->n;
+ tab->n++;
+}
+
+void *tab_get(struct tab *tab, char *k)
+{+ int i = tab->head[(unsigned char) k[0]];
+ while (i >= 0) {+ if (k[1] == tab->keys[i][1] && !strcmp(k, tab->keys[i]))
+ return tab->vals[i];
+ i = tab->next[i];
+ }
+ return NULL;
+}
--- /dev/null
+++ b/tab.h
@@ -1,0 +1,4 @@
+struct tab *tab_alloc(int sz);
+void tab_free(struct tab *tab);
+void tab_put(struct tab *tab, char *k, void *v);
+void *tab_get(struct tab *tab, char *k);
--- /dev/null
+++ b/trfn.c
@@ -1,0 +1,262 @@
+#include <ctype.h>
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+#include "sbuf.h"
+#include "tab.h"
+#include "trfn.h"
+#include "trfn_ch.h"
+
+#define WX(w) (((w) < 0 ? (w) - trfn_div / 2 : (w) + trfn_div / 2) / trfn_div)
+#define LEN(a) ((sizeof(a) / sizeof((a)[0])))
+#define HEXDIGS "0123456789abcdef"
+#define GNLEN 32
+#define AGLLEN (8 * 1024)
+
+static struct sbuf sbuf_char; /* charset section */
+static struct sbuf sbuf_kern; /* kernpairs section */
+static int trfn_div; /* divisor of widths */
+static int trfn_swid; /* space width */
+static char trfn_ligs[1024]; /* font ligatures */
+static char trfn_trname[256]; /* font troff name */
+static char trfn_psname[256]; /* font ps name */
+
+/* adobe glyphlist mapping */
+static char agl_key[AGLLEN][GNLEN];
+static char agl_val[AGLLEN][GNLEN];
+static int agl_n;
+
+/* lookup tables */
+static struct tab *tab_agl;
+static struct tab *tab_alts;
+static struct tab *tab_ctyp;
+
+static void pututf8(char **d, int c)
+{+ int l;
+ if (c > 0xffff) {+ *(*d)++ = 0xf0 | (c >> 18);
+ l = 3;
+ } else if (c > 0x7ff) {+ *(*d)++ = 0xe0 | (c >> 12);
+ l = 2;
+ } else if (c > 0x7f) {+ *(*d)++ = 0xc0 | (c >> 6);
+ l = 1;
+ } else {+ *(*d)++ = c > 0 ? c : ' ';
+ l = 0;
+ }
+ while (l--)
+ *(*d)++ = 0x80 | ((c >> (l * 6)) & 0x3f);
+ **d = '\0';
+}
+
+static int hexval(char *s, int len)
+{+ char *digs = HEXDIGS;
+ int n = 0;
+ int i;
+ for (i = 0; i < len; i++) {+ if (s[i] && strchr(digs, tolower(s[i])))
+ n = n * 16 + (strchr(digs, tolower(s[i])) - digs);
+ else
+ break;
+ }
+ return len == 1 ? n << 4 : n;
+}
+
+static int agl_read(char *path)
+{+ FILE *fin = fopen(path, "r");
+ char ln[GNLEN * 8];
+ char val[GNLEN];
+ char *s, *d;
+ int i;
+ if (!fin)
+ return 1;
+ while (fgets(ln, sizeof(ln), fin)) {+ s = strchr(ln, ';');
+ if (ln[0] == '#' || !s)
+ continue;
+ *s++ = '\0';
+ d = val;
+ while (s && *s) {+ while (*s == ' ')
+ s++;
+ pututf8(&d, hexval(s, 6));
+ s = strchr(s, ' ');
+ }
+ *d = '\0';
+ strcpy(agl_key[agl_n], ln);
+ strcpy(agl_val[agl_n], val);
+ agl_n++;
+ }
+ fclose(fin);
+ tab_agl = tab_alloc(agl_n);
+ for (i = 0; i < agl_n; i++)
+ tab_put(tab_agl, agl_key[i], agl_val[i]);
+ return 0;
+}
+
+static char *agl_map(char *s)
+{+ return tab_get(tab_agl, s);
+}
+
+static int achar_map(char *name)
+{+ int i;
+ for (i = 0; i < LEN(achars); i++) {+ struct achar *a = &achars[i];
+ if (!strncmp(a->name, name, strlen(a->name))) {+ char *postfix = name + strlen(a->name);
+ if (!*postfix)
+ return a->c;
+ if (!strcmp("isolated", postfix))+ return a->s ? a->s : a->c;
+ if (!strcmp("initial", postfix))+ return a->i ? a->i : a->c;
+ if (!strcmp("medial", postfix))+ return a->m ? a->m : a->c;
+ if (!strcmp("final", postfix))+ return a->f ? a->f : a->c;
+ }
+ }
+ return 0;
+}
+
+static int trfn_name(char *dst, char *src)
+{+ char ch[GNLEN];
+ char *s;
+ int i;
+ if (src[0] == '.')
+ return 1;
+ if (src[1] && strchr(src, '.'))
+ return 1; /* ignore opentype features for now */
+ while (*src && *src != '.') {+ s = ch;
+ if (src[0] == '_')
+ src++;
+ while (*src && *src != '_' && *src != '.')
+ *s++ = *src++;
+ *s = '\0';
+ if (agl_map(ch)) {+ strcpy(dst, agl_map(ch));
+ for (i = 0; i < LEN(agl_exceptions); i++)
+ if (!strcmp(agl_exceptions[i][0], dst))
+ strcpy(dst, agl_exceptions[i][1]);
+ dst = strchr(dst, '\0');
+ } else if (ch[0] == 'u' && ch[1] == 'n' && ch[2] == 'i') {+ for (i = 0; strlen(ch + 3 + 4 * i) >= 4; i++)
+ pututf8(&dst, hexval(ch + 3 + 4 * i, 4));
+ } else if (ch[0] == 'u' && ch[1] && strchr(HEXDIGS, tolower(ch[1]))) {+ pututf8(&dst, hexval(ch + 1, 6));
+ } else if (achar_map(ch)) {+ pututf8(&dst, achar_map(ch));
+ } else {+ return 1;
+ }
+ }
+ return src[0];
+}
+
+static void trfn_lig(char *c)
+{+ int i;
+ for (i = 0; i < LEN(ligs); i++)
+ if (!strcmp(ligs[i], c))
+ sprintf(strchr(trfn_ligs, '\0'), "%s ", c);
+}
+
+static int trfn_type(char *c)
+{+ struct ctype *t = tab_get(tab_ctyp, c);
+ return t ? t->type : 3;
+}
+
+void trfn_char(char *c, char *n, int wid, int typ)
+{+ char uc[GNLEN];
+ char pos[GNLEN];
+ char **a;
+ if (trfn_name(uc, c))
+ strcpy(uc, "---");
+ if (strchr(uc, ' ')) { /* space not allowed in character names */+ if (!trfn_swid && !strcmp(" ", uc))+ trfn_swid = WX(wid);
+ return;
+ }
+ trfn_lig(uc);
+ if (typ < 0)
+ typ = trfn_type(uc);
+ strcpy(pos, c);
+ if (n && atoi(n) >= 0 && atoi(n) < 256)
+ strcpy(pos, n);
+ if (!n && !uc[1] && uc[0] >= 32 && uc[0] <= 125)
+ sprintf(pos, "%d", uc[0]);
+ sbuf_printf(&sbuf_char, "%s\t%d\t%d\t%s\n", uc, WX(wid), typ, pos);
+ a = tab_get(tab_alts, uc);
+ while (a && *a)
+ sbuf_printf(&sbuf_char, "%s\t\"\n", *a++);
+}
+
+void trfn_kern(char *c1, char *c2, int x)
+{+ char g1[GNLEN], g2[GNLEN];
+ if (!trfn_name(g1, c1) && !trfn_name(g2, c2) && abs(WX(x)) > WX(20))
+ if (!strchr(g1, ' ') && !strchr(g2, ' '))
+ sbuf_printf(&sbuf_kern, "%s\t%s\t%d\n", g1, g2, WX(x));
+}
+
+void trfn_trfont(char *name)
+{+ if (!trfn_trname[0])
+ strcpy(trfn_trname, name);
+}
+
+void trfn_psfont(char *name)
+{+ if (!trfn_psname[0])
+ strcpy(trfn_psname, name);
+}
+
+void trfn_print(void)
+{+ if (trfn_trname[0])
+ printf("name %s\n", trfn_trname);+ if (trfn_psname[0])
+ printf("fontname %s\n", trfn_psname);+ printf("spacewidth %d\n", trfn_swid);+ printf("ligatures %s 0\n", trfn_ligs);+ printf("charset\n");+ printf("%s", sbuf_buf(&sbuf_char));+ printf("kernpairs\n");+ printf("%s", sbuf_buf(&sbuf_kern));+}
+
+void trfn_init(int res)
+{+ int i;
+ trfn_div = 7200 / res;
+ agl_read("glyphlist.txt");+ sbuf_init(&sbuf_char);
+ sbuf_init(&sbuf_kern);
+ tab_alts = tab_alloc(LEN(alts));
+ tab_ctyp = tab_alloc(LEN(ctype));
+ for (i = 0; i < LEN(alts); i++)
+ tab_put(tab_alts, alts[i][0], alts[i] + 1);
+ for (i = 0; i < LEN(ctype); i++)
+ tab_put(tab_ctyp, ctype[i].ch, &ctype[i]);
+}
+
+void trfn_done(void)
+{+ sbuf_done(&sbuf_char);
+ sbuf_done(&sbuf_kern);
+ tab_free(tab_alts);
+ tab_free(tab_ctyp);
+ if (tab_agl)
+ tab_free(tab_agl);
+}
--- /dev/null
+++ b/trfn.h
@@ -1,0 +1,7 @@
+void trfn_init(int res);
+void trfn_done(void);
+void trfn_trfont(char *name);
+void trfn_psfont(char *fontname);
+void trfn_print(void);
+void trfn_char(char *c, char *n, int wid, int typ);
+void trfn_kern(char *c1, char *c2, int x);
--- /dev/null
+++ b/trfn_ch.h
@@ -1,0 +1,688 @@
+/* ligatures */
+static char *ligs[] = {+ "fh", "fi", "fj", "fk", "fl", "ff", "ffi", "ffj", "ffl", "fft", "ft", "Th"
+};
+
+/* AGL exceptions; expanding ligatures */
+static char *agl_exceptions[][2] = {+ {"ff", "ff"},+ {"fi", "fi"},+ {"fl", "fl"},+ {"ffi", "ffi"},+ {"ffl", "ffl"},+ {"st", "st"},+};
+
+/* troff aliases */
+static char *alts[][8] = {+ {"\\", "bs"},+ {"`", "ga"},+ {"¡", "!!"},+ {"¢", "c|", "ct"},+ {"£", "L-", "ps"},+ {"¤", "xo", "cr"},+ {"¥", "Y-", "yn"},+ {"¦", "||"},+ {"§", "so", "sc"},+ {"¨", "\"\"", ":a"},+ {"©", "co"},+ {"ª", "a_"},+ {"«", "<<"},+ {"»", ">>"},+ {"¬", "-,"},+ {"¬", "-,"},+ {"-", "hy"},+ {"−", "--"},+ {"®", "ro", "rg"},+ {"¯", "a^", "-a"},+ {"°", "0^"},+ {"±", "+-"},+ {"²", "2^"},+ {"³", "3^"},+ {"´", "\\'", "aa"},+ {"µ", "mu"},+ {"¶", "P!", "pg"},+ {"·", ".^"},+ {"¸", ",,", ",a"},+ {"¹", "1^"},+ {"º", "o_"},+ {"»", ">>"},+ {"¼", "14"},+ {"½", "12"},+ {"¾", "34"},+ {"¿", "??"},+ {"À", "A`"},+ {"Á", "A'"},+ {"Â", "A^"},+ {"Ã", "A~"},+ {"Ä", "A:", "A\""},+ {"Å", "A*"},+ {"Æ", "AE"},+ {"Ç", "C,"},+ {"Č", "C<"},+ {"È", "E`"},+ {"É", "E'"},+ {"Ê", "E^"},+ {"Ë", "E:", "E\""},+ {"Ì", "I`"},+ {"Í", "I'"},+ {"Î", "I^"},+ {"Ï", "I:", "I\""},+ {"Ð", "D-"},+ {"Ď", "D<"},+ {"Ñ", "N~"},+ {"Ò", "O`"},+ {"Ó", "O'"},+ {"Ô", "O^"},+ {"Õ", "O~"},+ {"Ö", "O:"},+ {"Ő", "O\""},+ {"×", "xx"},+ {"Ø", "O/"},+ {"Ř", "R<"},+ {"Š", "S<"},+ {"Ť", "T<"},+ {"Ù", "U`"},+ {"Ú", "U'"},+ {"Û", "U^"},+ {"Ü", "U:"},+ {"Ű", "U\""},+ {"Ů", "U0"},+ {"Ý", "Y'"},+ {"Ÿ", "Y:", "Y\""},+ {"Ž", "Z<"},+ {"Þ", "TH"},+ {"ß", "ss"},+ {"à", "a`"},+ {"á", "a'"},+ {"â", "a^"},+ {"ã", "a~"},+ {"ä", "a:", "a\""},+ {"å", "a*"},+ {"æ", "ae"},+ {"ç", "c,"},+ {"č", "c<"},+ {"ď", "d<"},+ {"è", "e`"},+ {"é", "e'"},+ {"ê", "e^"},+ {"ë", "e:", "e\""},+ {"ě", "e<"},+ {"ì", "i`"},+ {"í", "i'"},+ {"î", "i^"},+ {"ï", "i\""},+ {"ð", "d-"},+ {"ñ", "n~"},+ {"ň", "n<"},+ {"ò", "o`"},+ {"ó", "o'"},+ {"ô", "o^"},+ {"õ", "o~"},+ {"ö", "o:"},+ {"ő", "o\""},+ {"÷", "-:"},+ {"ø", "o/"},+ {"ř", "r<"},+ {"š", "s<"},+ {"ť", "t<"},+ {"ù", "u`"},+ {"ú", "u'"},+ {"û", "u^"},+ {"ü", "u:"},+ {"ű", "u\""},+ {"ů", "u0"},+ {"ý", "y'"},+ {"ž", "z<"},+ {"þ", "th"},+ {"ÿ", "y:", "y\""},+ {"˘", "Ua"},+ {"˙", ".a"},+ {"˚", "oa"},+ {"˝", "\"a"},+ {"˛", "Ca"},+ {"ˇ", "va"},+ {"∀", "fa"},+ {"∃", "te"},+ {"∋", "st"},+ {"∗", "**"},+ {"+", "pl"},+ {"−", "mi"},+ {"/", "sl"},+ {"=", "eq"},+ {"≅", "cg"},+ {"Α", "*A"},+ {"Β", "*B"},+ {"Χ", "*X"},+ {"∆", "*D"},+ {"Ε", "*E"},+ {"Φ", "*F"},+ {"Γ", "*G"},+ {"Η", "*Y"},+ {"Ι", "*I"},+ {"Κ", "*K"},+ {"Λ", "*L"},+ {"Μ", "*M"},+ {"Ν", "*N"},+ {"Ο", "*O"},+ {"Π", "*P"},+ {"Θ", "*H"},+ {"Ρ", "*R"},+ {"Σ", "*S"},+ {"Τ", "*T"},+ {"Υ", "*U"},+ {"ς", "ts"},+ {"Ω", "*W"},+ {"Ξ", "*C"},+ {"Ψ", "*Q"},+ {"Ζ", "*Z"},+ {"∴", "tf"},+ {"⊥", "pp"},+ {"", "rn"},+ {"α", "*a"},+ {"β", "*b"},+ {"χ", "*x"},+ {"δ", "*d"},+ {"ε", "*e"},+ {"φ", "*f"},+ {"γ", "*g"},+ {"η", "*y"},+ {"ι", "*i"},+ {"κ", "*k"},+ {"λ", "*l"},+ {"μ", "*m"},+ {"ν", "*n"},+ {"ο", "*o"},+ {"π", "*p"},+ {"θ", "*h"},+ {"ρ", "*r"},+ {"σ", "*s"},+ {"τ", "*t"},+ {"υ", "*u"},+ {"ω", "*w"},+ {"ξ", "*c"},+ {"ψ", "*q"},+ {"ζ", "*z"},+ {"|", "or"},+ {"∼", "ap"},+ {"′", "fm"},+ {"≤", "<="},+ {"⁄", "fr"},+ {"∞", "if"},+ {"ƒ", "fn",},+ {"↔", "ab"},+ {"←", "<-"},+ {"↑", "ua"},+ {"→", "->"},+ {"↓", "da"},+ {"°", "de"},+ {"±", "+-"},+ {"≥", ">="},+ {"×", "mu"},+ {"∝", "pt"},+ {"∂", "pd"},+ {"•", "bu"},+ {"÷", "di"},+ {"≠", "!="},+ {"≡", "=="},+ {"≈", "~~"},+ {"…", "el"},+ {"", "av"},+ {"", "ah"},+ {"↵", "CR"},+ {"ℵ", "af"},+ {"ℑ", "If"},+ {"ℜ", "Rf"},+ {"ℛ", "ws"},+ {"⊗", "Ox"},+ {"⊕", "O+"},+ {"∅", "es"},+ {"∩", "ca"},+ {"∪", "cu"},+ {"⊃", "sp"},+ {"⊇", "ip"},+ {"⊄", "!b"},+ {"⊂", "sb"},+ {"⊆", "ib"},+ {"∈", "mo"},+ {"∉", "!m"},+ {"∠", "an"},+ {"∇", "gr"},+ {"", "rg"},+ {"", "co"},+ {"", "tm"},+ {"∏", "pr"},+ {"√", "sr"},+ {"⋅", "c."},+ {"¬", "no"},+ {"∧", "l&"},+ {"∨", "l|"},+ {"◊", "lz"},+ {"〈", "b<"},+ {"", "RG"},+ {"", "CO"},+ {"", "TM"},+ {"∑", "su"},+ {"", "LT"},+ {"", "br", "LX"},+ {"", "LB"},+ {"", "lc"},+ {"", "lx"},+ {"", "lf"},+ {"", "lt"},+ {"", "lk"},+ {"", "lb"},+ {"", "bv", "|",},+ {"〉", "b>"},+ {"∫", "is"},+ {"", "RT"},+ {"", "RX"},+ {"", "RB"},+ {"", "rc"},+ {"", "rx"},+ {"", "rf"},+ {"", "rt"},+ {"", "rk"},+ {"", "rb"},+ {"ff", "ff"},+ {"fi", "fi"},+ {"fl", "fl"},+ {"ffi", "ffi"},+ {"ffl", "ffl"},+ {"st", "st"},+};
+
+/* different shapes of arabic and farsi characters */
+static struct achar {+ char *name;
+ unsigned c;
+ unsigned s;
+ unsigned i;
+ unsigned m;
+ unsigned f;
+} achars[] = {+ {"hamza", 0x0621, 0xfe80},+ {"alefwithmaddaabove", 0x0622, 0xfe81, 0, 0, 0xfe82},+ {"alefwithhamzaabove", 0x0623, 0xfe83, 0, 0, 0xfe84},+ {"wawwithhamzaabove", 0x0624, 0xfe85, 0, 0, 0xfe86},+ {"alefwithhamzabelow", 0x0625, 0xfe87, 0, 0, 0xfe88},+ {"yehwithhamzaabove", 0x0626, 0xfe89, 0xfe8b, 0xfe8c, 0xfe8a},+ {"alef", 0x0627, 0xfe8d, 0, 0, 0xfe8e},+ {"arabicalef", 0x0627},+ {"beh", 0x0628, 0xfe8f, 0xfe91, 0xfe92, 0xfe90},+ {"tehmarbuta", 0x0629, 0xfe93, 0, 0, 0xfe94},+ {"teh", 0x062a, 0xfe95, 0xfe97, 0xfe98, 0xfe96},+ {"theh", 0x062b, 0xfe99, 0xfe9b, 0xfe9c, 0xfe9a},+ {"jeem", 0x062c, 0xfe9d, 0xfe9f, 0xfea0, 0xfe9e},+ {"hah", 0x062d, 0xfea1, 0xfea3, 0xfea4, 0xfea2},+ {"khah", 0x062e, 0xfea5, 0xfea7, 0xfea8, 0xfea6},+ {"dal", 0x062f, 0xfea9, 0, 0, 0xfeaa},+ {"thal", 0x0630, 0xfeab, 0, 0, 0xfeac},+ {"reh", 0x0631, 0xfead, 0, 0, 0xfeae},+ {"zain", 0x0632, 0xfeaf, 0, 0, 0xfeb0},+ {"seen", 0x0633, 0xfeb1, 0xfeb3, 0xfeb4, 0xfeb2},+ {"sheen", 0x0634, 0xfeb5, 0xfeb7, 0xfeb8, 0xfeb6},+ {"sad", 0x0635, 0xfeb9, 0xfebb, 0xfebc, 0xfeba},+ {"dad", 0x0636, 0xfebd, 0xfebf, 0xfec0, 0xfebe},+ {"tah", 0x0637, 0xfec1, 0xfec3, 0xfec4, 0xfec2},+ {"zah", 0x0638, 0xfec5, 0xfec7, 0xfec8, 0xfec6},+ {"ain", 0x0639, 0xfec9, 0xfecb, 0xfecc, 0xfeca},+ {"ghain", 0x063a, 0xfecd, 0xfecf, 0xfed0, 0xfece},+ {"tatweel", 0x0640},+ {"feh", 0x0641, 0xfed1, 0xfed3, 0xfed4, 0xfed2},+ {"qaf", 0x0642, 0xfed5, 0xfed7, 0xfed8, 0xfed6},+ {"kaf", 0x0643, 0xfed9, 0xfedb, 0xfedc, 0xfeda},+ {"lam", 0x0644, 0xfedd, 0xfedf, 0xfee0, 0xfede},+ {"meem", 0x0645, 0xfee1, 0xfee3, 0xfee4, 0xfee2},+ {"noon", 0x0646, 0xfee5, 0xfee7, 0xfee8, 0xfee6},+ {"heh", 0x0647, 0xfee9, 0xfeeb, 0xfeec, 0xfeea},+ {"waw", 0x0648, 0xfeed, 0, 0, 0xfeee},+ {"alefmaksura", 0x0649, 0xfeef, 0, 0, 0xfef0},+ {"yeh", 0x064a, 0xfef1, 0xfef3, 0xfef4, 0xfef2},+ {"fathatan", 0x064b, 0xfe70},+ {"dammatan", 0x064c, 0xfe72},+ {"kasratan", 0x064d, 0xfe74},+ {"fatha", 0x064e, 0xfe76, 0, 0xfe77, 0},+ {"damma", 0x064f, 0xfe78, 0, 0xfe79, 0},+ {"kasra", 0x0650, 0xfe7a, 0, 0xfe7b, 0},+ {"shadda", 0x0651, 0xfe7c, 0, 0xfe7c, 0},+ {"sukun", 0x0652, 0xfe7e, 0, 0xfe7f, 0},+ {"peh", 0x067e, 0xfb56, 0xfb58, 0xfb59, 0xfb57},+ {"tcheh", 0x0686, 0xfb7a, 0xfb7c, 0xfb7d, 0xfb7b},+ {"jeh", 0x0698, 0xfb8a, 0, 0, 0xfb8b},+ {"keheh", 0x06a9, 0xfb8e, 0xfb90, 0xfb91, 0xfb8f},+ {"gaf", 0x06af, 0xfb92, 0xfb94, 0xfb95, 0xfb93},+ {"farsiyeh", 0x06cc, 0xfbfc, 0xfbfe, 0xfbff, 0xfbfd},+ {"lamwithalef", 0xfefb, 0xfefb, 0, 0, 0xfefc},+};
+
+static struct ctype {+ char *ch; /* character name */
+ int type; /* ascender/descender type */
+} ctype[] = {+ {"Ï", 2},+ {"²", 2},+ {"ì", 2},+ {"u", 0},+ {"Ì", 2},+ {"U", 2},+ {"∞", 0},+ {"ú", 2},+ {"∫", 3},+ {"Ú", 2},+ {"a", 0},+ {"ι", 0},+ {"û", 2},+ {"A", 2},+ {"Ι", 2},+ {"Û", 2},+ {"á", 2},+ {"j", 3},+ {"ü", 2},+ {"Á", 2},+ {"J", 3},+ {"Ü", 2},+ {"â", 2},+ {"k", 2},+ {"ù", 2},+ {"Â", 2},+ {"K", 2},+ {"Ù", 2},+ {"´", 2},+ {"κ", 0},+ {"ű", 2},+ {"ä", 2},+ {"Κ", 2},+ {"Ű", 2},+ {"Ä", 2},+ {"l", 2},+ {"_", 1},+ {"æ", 0},+ {"L", 2},+ {"υ", 0},+ {"Æ", 2},+ {"λ", 2},+ {"Υ", 2},+ {"à", 2},+ {"Λ", 2},+ {"ů", 2},+ {"À", 2},+ {"<", 0},+ {"Ů", 2},+ {"α", 0},+ {"≤", 2},+ {"v", 0},+ {"Α", 2},+ {"¬", 0},+ {"V", 2},+ {"&", 2},+ {"◊", 2},+ {"w", 0},+ {"≈", 0},+ {"m", 0},+ {"W", 2},+ {"å", 2},+ {"M", 2},+ {"x", 0},+ {"Å", 2},+ {"¯", 2},+ {"X", 2},+ {"^", 2},+ {"−", 0},+ {"ξ", 3},+ {"~", 0},+ {"µ", 1},+ {"Ξ", 2},+ {"*", 2},+ {"Μ", 2},+ {"y", 1},+ {"@", 3},+ {"×", 0},+ {"Y", 2},+ {"ã", 2},+ {"n", 0},+ {"ý", 3},+ {"Ã", 2},+ {"N", 2},+ {"Ý", 2},+ {"b", 2},+ {"ň", 2},+ {"ÿ", 3},+ {"B", 2},+ {"9", 2},+ {"¥", 2},+ {"\\", 2},+ {"≠", 0},+ {"z", 0},+ {"|", 3},+ {"ñ", 2},+ {"Z", 2},+ {"β", 3},+ {"Ñ", 2},+ {"ž", 2},+ {"Β", 2},+ {"ν", 0},+ {"Ž", 2},+ {"{", 3},+ {"Ν", 2},+ {"0", 2},+ {"}", 3},+ {"#", 2},+ {"ζ", 3},+ {"[", 3},+ {"o", 0},+ {"Ζ", 2},+ {"]", 3},+ {"O", 2},+ {"", 3},+ {"˘", 2},+ {"ó", 2},+ {"", 2},+ {"¦", 3},+ {"Ó", 2},+ {"", 3},+ {"•", 0},+ {"ô", 2},+ {"", 2},+ {"c", 0},+ {"Ô", 2},+ {"", 2},+ {"C", 2},+ {"ö", 2},+ {"", 2},+ {"ˇ", 2},+ {"Ö", 2},+ {"", 2},+ {"č", 2},+ {"˛", 1},+ {"", 2},+ {"Č", 2},+ {"ò", 2},+ {"", 2},+ {"ç", 1},+ {"Ò", 2},+ {"،", 0},+ {"Ç", 3},+ {"ő", 2},+ {"؛", 2},+ {"¸", 1},+ {"Ő", 2},+ {"؟", 2},+ {"¢", 2},+ {"ω", 0},+ {"٠", 0},+ {"χ", 1},+ {"Ω", 2},+ {"١", 2},+ {"Χ", 2},+ {"ο", 0},+ {"٢", 2},+ {"ˆ", 2},+ {"Ο", 2},+ {"٣", 2},+ {":", 0},+ {"1", 2},+ {"٤", 2},+ {",", 1},+ {"½", 2},+ {"٥", 2},+ {"©", 2},+ {"¼", 2},+ {"٦", 2},+ {"¤", 2},+ {"¹", 2},+ {"٧", 2},+ {"d", 2},+ {"ª", 2},+ {"٨", 2},+ {"D", 2},+ {"º", 2},+ {"٩", 2},+ {"ď", 2},+ {"ø", 0},+ {"۰", 0},+ {"Ď", 2},+ {"Ø", 2},+ {"۱", 2},+ {"°", 2},+ {"õ", 2},+ {"۲", 2},+ {"δ", 2},+ {"Õ", 2},+ {"۳", 2},+ {"∆", 2},+ {"p", 1},+ {"۴", 2},+ {"¨", 2},+ {"P", 2},+ {"۵", 2},+ {"÷", 0},+ {"¶", 3},+ {"۶", 2},+ {"$", 2},+ {"(", 3},+ {"۷", 2},+ {"˙", 2},+ {")", 3},+ {"۸", 2},+ {"e", 0},+ {"∂", 2},+ {"۹", 2},+ {"E", 2},+ {"%", 2},+ {"٪", 2},+ {"é", 2},+ {".", 0},+ {"", 0},+ {"É", 2},+ {"·", 0},+ {"", 0},+ {"ě", 2},+ {"φ", 3},+ {"ê", 2},+ {"Φ", 2},+ {"Ê", 2},+ {"π", 0},+ {"ë", 2},+ {"Π", 2},+ {"Ë", 2},+ {"+", 0},+ {"è", 2},+ {"±", 2},+ {"È", 2},+ {"∏", 2},+ {"8", 2},+ {"ψ", 3},+ {"…", 0},+ {"Ψ", 2},+ {"ε", 0},+ {"q", 1},+ {"Ε", 2},+ {"Q", 3},+ {"=", 0},+ {"?", 2},+ {"η", 1},+ {"¿", 1},+ {"Η", 2},+ {"\"", 2},+ {"ð", 2},+ {"‘", 2},+ {"Ð", 2},+ {"’", 2},+ {"!", 2},+ {"r", 0},+ {"¡", 1},+ {"R", 2},+ {"f", 2},+ {"√", 2},+ {"F", 2},+ {"ř", 2},+ {"ff", 2},+ {"Ř", 2},+ {"ffi", 2},+ {"®", 2},+ {"ffj", 3},+ {"ρ", 1},+ {"ffl", 2},+ {"Ρ", 2},+ {"fi", 2},+ {"˚", 2},+ {"fi", 2},+ {"s", 0},+ {"5", 2},+ {"S", 2},+ {"fj", 3},+ {"š", 2},+ {"fl", 2},+ {"Š", 2},+ {"fl", 2},+ {"§", 2},+ {"ƒ", 3},+ {";", 1},+ {"4", 2},+ {"7", 2},+ {"⁄", 2},+ {"σ", 0},+ {"g", 1},+ {"Σ", 2},+ {"G", 2},+ {"6", 2},+ {"γ", 1},+ {"/", 2},+ {"Γ", 2},+ {"£", 2},+ {"ß", 2},+ {"∑", 2},+ {"`", 2},+ {"t", 2},+ {">", 0},+ {"T", 2},+ {"≥", 2},+ {"τ", 0},+ {"«", 0},+ {"Τ", 2},+ {"»", 0},+ {"ť", 2},+ {"h", 2},+ {"Ť", 2},+ {"H", 2},+ {"Th", 2},+ {"˝", 2},+ {"θ", 2},+ {"-", 0},+ {"Θ", 2},+ {"i", 2},+ {"þ", 3},+ {"I", 2},+ {"Þ", 2},+ {"í", 2},+ {"3", 2},+ {"Í", 2},+ {"¾", 2},+ {"î", 2},+ {"³", 2},+ {"Î", 2},+ {"˜", 2},+ {"ï", 2},+ {"2", 2},+};