/*
- * $XFree86$
+ * $RCSId: xc/lib/fontconfig/fc-lang/fc-lang.c,v 1.3 2002/08/22 07:36:43 keithp Exp $
*
- * Copyright © 2002 Keith Packard, member of The XFree86 Project, Inc.
+ * Copyright © 2002 Keith Packard
*
* Permission to use, copy, modify, distribute, and sell this software and its
* documentation for any purpose is hereby granted without fee, provided that
*/
#include "fcint.h"
+#include "fccharset.c"
+#include "fcstr.c"
+#include "fcserialize.c"
/*
* fc-lang
*
* Read a set of language orthographies and build C declarations for
* charsets which can then be used to identify which languages are
- * supported by a given font. Note that it would be nice if
- * this could be done while compiling the library, but this
- * code uses a number of routines from the library. It's
- * expediant to just ship the pre-built version along with the
- * source orthographies.
+ * supported by a given font. Note that this uses some utilities
+ * from the fontconfig library, so the necessary file is simply
+ * included in this compilation. A couple of extra utility
+ * functions are also needed in slightly modified form
*/
+void
+FcMemAlloc (int kind, int size)
+{
+}
+
+void
+FcMemFree (int kind, int size)
+{
+}
+
+FcChar8 *
+FcConfigHome (void)
+{
+ return (FcChar8 *) getenv ("HOME");
+}
+
static void
-fatal (char *file, int lineno, char *msg)
+fatal (const char *file, int lineno, const char *msg)
{
- fprintf (stderr, "%s:%d: %s\n", file, lineno, msg);
+ if (lineno)
+ fprintf (stderr, "%s:%d: %s\n", file, lineno, msg);
+ else
+ fprintf (stderr, "%s: %s\n", file, msg);
exit (1);
}
get_line (FILE *f, char *line, int *lineno)
{
char *hash;
+ int end;
if (!fgets (line, 1024, f))
return 0;
++(*lineno);
hash = strchr (line, '#');
if (hash)
*hash = '\0';
+
+ end = strlen (line);
+ while (end > 0 && isspace (line[end-1]))
+ line[--end] = '\0';
+
if (line[0] == '\0' || line[0] == '\n' || line[0] == '\032' || line[0] == '\r')
return get_line (f, line, lineno);
return line;
}
+static char *dir = 0;
+
+static FILE *
+scanopen (char *file)
+{
+ FILE *f;
+
+ f = fopen (file, "r");
+ if (!f && dir)
+ {
+ char path[1024];
+
+ strcpy (path, dir);
+ strcat (path, "/");
+ strcat (path, file);
+ f = fopen (path, "r");
+ }
+ return f;
+}
+
/*
* build a single charset from a source file
*
if (!strncmp (line, "include", 7))
{
file = strchr (line, ' ');
- while (*file == ' ')
+ if (!file)
+ fatal (line, lineno,
+ "invalid syntax, expected: include filename");
+ while (isspace(*file))
file++;
- end = strlen (file);
- if (file[end-1] == '\n')
- file[end-1] = '\0';
- f = fopen (file, "r");
+ f = scanopen (file);
if (!f)
fatal (file, 0, "can't open");
c = scan (f, file);
while ((c = *name++))
{
- if (isupper (c))
- c = tolower (c);
+ if (isupper ((int) (unsigned char) c))
+ c = tolower ((int) (unsigned char) c);
if (c == '_')
c = '-';
if (c == ' ')
return lang;
}
+static int compare (const void *a, const void *b)
+{
+ const FcChar8 *const *as = a, *const *bs = b;
+ return FcStrCmpIgnoreCase (*as, *bs);
+}
+
+#define MAX_LANG 1024
+#define MAX_LANG_SET_MAP ((MAX_LANG + 31) / 32)
+
+#define BitSet(map, id) ((map)[(id)>>5] |= ((FcChar32) 1 << ((id) & 0x1f)))
+#define BitGet(map, id) ((map)[(id)>>5] >> ((id) & 0x1f)) & 1)
+
int
main (int argc, char **argv)
{
- FcCharSet *sets[1024];
- char *names[1024];
+ static char *files[MAX_LANG];
+ static FcCharSet *sets[MAX_LANG];
+ static int duplicate[MAX_LANG];
+ static int country[MAX_LANG];
+ static char *names[MAX_LANG];
+ static char *langs[MAX_LANG];
+ static int off[MAX_LANG];
FILE *f;
+ int ncountry = 0;
int i = 0;
- FcCharLeaf **leaves, **sleaves;
+ int nsets = 0;
+ int argi;
+ FcCharLeaf **leaves;
int total_leaves = 0;
- int l, sl, tl;
- char line[1024];
+ int l, sl, tl, tn;
+ static char line[1024];
+ static FcChar32 map[MAX_LANG_SET_MAP];
+ int num_lang_set_map;
+ int setRangeStart[26];
+ int setRangeEnd[26];
+ FcChar8 setRangeChar;
- while (*++argv)
+ argi = 1;
+ while (argv[argi])
{
- f = fopen (*argv, "r");
+ if (!strcmp (argv[argi], "-d"))
+ {
+ argi++;
+ dir = argv[argi++];
+ continue;
+ }
+ if (i == MAX_LANG)
+ fatal (argv[0], 0, "Too many languages");
+ files[i++] = argv[argi++];
+ }
+ files[i] = 0;
+ qsort (files, i, sizeof (char *), compare);
+ i = 0;
+ while (files[i])
+ {
+ f = scanopen (files[i]);
if (!f)
- fatal (*argv, 0, strerror (errno));
- sets[i] = scan (f, *argv);
- names[i] = get_name (*argv);
+ fatal (files[i], 0, strerror (errno));
+ sets[i] = scan (f, files[i]);
+ names[i] = get_name (files[i]);
+ langs[i] = get_lang(names[i]);
+ if (strchr (langs[i], '-'))
+ country[ncountry++] = i;
+
total_leaves += sets[i]->num;
i++;
fclose (f);
}
+ nsets = i;
sets[i] = 0;
leaves = malloc (total_leaves * sizeof (FcCharLeaf *));
tl = 0;
*/
for (i = 0; sets[i]; i++)
{
- sleaves = sets[i]->leaves;
for (sl = 0; sl < sets[i]->num; sl++)
{
for (l = 0; l < tl; l++)
- if (leaves[l] == sleaves[sl])
+ if (leaves[l] == FcCharSetLeaf(sets[i], sl))
break;
if (l == tl)
- leaves[tl++] = sleaves[sl];
+ leaves[tl++] = FcCharSetLeaf(sets[i], sl);
}
}
printf ("/* total size: %d unique leaves: %d */\n\n",
total_leaves, tl);
+
+ /*
+ * Find duplicate charsets
+ */
+ duplicate[0] = -1;
+ for (i = 1; sets[i]; i++)
+ {
+ int j;
+
+ duplicate[i] = -1;
+ for (j = 0; j < i; j++)
+ if (sets[j] == sets[i])
+ {
+ duplicate[i] = j;
+ break;
+ }
+ }
+
+ tn = 0;
+ for (i = 0; sets[i]; i++) {
+ if (duplicate[i] >= 0)
+ continue;
+ off[i] = tn;
+ tn += sets[i]->num;
+ }
+
+ printf ("#define LEAF0 (%d * sizeof (FcLangCharSet))\n", nsets);
+ printf ("#define OFF0 (LEAF0 + %d * sizeof (FcCharLeaf))\n", tl);
+ printf ("#define NUM0 (OFF0 + %d * sizeof (intptr_t))\n", tn);
+ printf ("#define SET(n) (n * sizeof (FcLangCharSet) + offsetof (FcLangCharSet, charset))\n");
+ printf ("#define OFF(s,o) (OFF0 + o * sizeof (intptr_t) - SET(s))\n");
+ printf ("#define NUM(s,n) (NUM0 + n * sizeof (FcChar16) - SET(s))\n");
+ printf ("#define LEAF(o,l) (LEAF0 + l * sizeof (FcCharLeaf) - (OFF0 + o * sizeof (intptr_t)))\n");
+ printf ("#define fcLangCharSets (fcLangData.langCharSets)\n");
+ printf ("\n");
+
+ printf ("static const struct {\n"
+ " FcLangCharSet langCharSets[%d];\n"
+ " FcCharLeaf leaves[%d];\n"
+ " intptr_t leaf_offsets[%d];\n"
+ " FcChar16 numbers[%d];\n"
+ "} fcLangData = {\n",
+ nsets, tl, tn, tn);
+
+ /*
+ * Dump sets
+ */
+
+ printf ("{\n");
+ for (i = 0; sets[i]; i++)
+ {
+ int j = duplicate[i];
+
+ if (j < 0)
+ j = i;
+
+ printf (" { (FcChar8 *) \"%s\", "
+ " { FC_REF_CONSTANT, %d, OFF(%d,%d), NUM(%d,%d) } }, /* %d */\n",
+ langs[i],
+ sets[j]->num, i, off[j], i, off[j], i);
+ }
+ printf ("},\n");
+
/*
* Dump leaves
*/
- printf ("static const FcCharLeaf leaves[%d] = {\n", tl);
+ printf ("{\n");
for (l = 0; l < tl; l++)
{
printf (" { { /* %d */", l);
}
printf ("\n } },\n");
}
- printf ("};\n\n");
- printf ("#define L(n) ((FcCharLeaf *) &leaves[n])\n\n");
+ printf ("},\n");
+
/*
- * Dump arrays
+ * Dump leaves
*/
+ printf ("{\n");
for (i = 0; sets[i]; i++)
{
int n;
- printf ("static const FcCharLeaf *leaves_%s[%d] = {\n",
- names[i], sets[i]->num);
+ if (duplicate[i] >= 0)
+ continue;
+ printf (" /* %s */\n", names[i]);
for (n = 0; n < sets[i]->num; n++)
{
- if (n % 8 == 0)
+ if (n % 4 == 0)
printf (" ");
for (l = 0; l < tl; l++)
- if (leaves[l] == sets[i]->leaves[n])
+ if (leaves[l] == FcCharSetLeaf(sets[i], n))
break;
if (l == tl)
fatal (names[i], 0, "can't find leaf");
- printf (" L(%3d),", l);
- if (n % 8 == 7)
+ printf (" LEAF(%3d,%3d),", off[i], l);
+ if (n % 4 == 3)
printf ("\n");
}
- if (n % 8 != 0)
+ if (n % 4 != 0)
printf ("\n");
- printf ("};\n\n");
+ }
+ printf ("},\n");
- printf ("static const FcChar16 numbers_%s[%d] = {\n",
- names[i], sets[i]->num);
+ printf ("{\n");
+ for (i = 0; sets[i]; i++)
+ {
+ int n;
+
+ if (duplicate[i] >= 0)
+ continue;
+ printf (" /* %s */\n", names[i]);
for (n = 0; n < sets[i]->num; n++)
{
if (n % 8 == 0)
printf (" ");
- printf (" 0x%04x,", sets[i]->numbers[n]);
+ printf (" 0x%04x,", FcCharSetNumbers (sets[i])[n]);
if (n % 8 == 7)
printf ("\n");
}
if (n % 8 != 0)
printf ("\n");
+ }
+ printf ("}\n");
+
+ printf ("};\n\n");
+
+ printf ("#define NUM_LANG_CHAR_SET %d\n", i);
+ num_lang_set_map = (i + 31) / 32;
+ printf ("#define NUM_LANG_SET_MAP %d\n", num_lang_set_map);
+ /*
+ * Dump indices with country codes
+ */
+ if (ncountry)
+ {
+ int c;
+ int ncountry_ent = 0;
+ printf ("\n");
+ printf ("static const FcChar32 fcLangCountrySets[][NUM_LANG_SET_MAP] = {\n");
+ for (c = 0; c < ncountry; c++)
+ {
+ i = country[c];
+ if (i >= 0)
+ {
+ int lang = strchr (langs[i], '-') - langs[i];
+ int d, k;
+
+ for (k = 0; k < num_lang_set_map; k++)
+ map[k] = 0;
+
+ BitSet (map, i);
+ for (d = c + 1; d < ncountry; d++)
+ {
+ int j = country[d];
+ if (j >= 0 && !strncmp (langs[j], langs[i], l))
+ {
+ BitSet(map, j);
+ country[d] = -1;
+ }
+ }
+ printf (" {");
+ for (k = 0; k < num_lang_set_map; k++)
+ printf (" 0x%08x,", map[k]);
+ printf (" }, /* %*.*s */\n",
+ lang, lang, langs[i]);
+ ++ncountry_ent;
+ }
+ }
printf ("};\n\n");
+ printf ("#define NUM_COUNTRY_SET %d\n", ncountry_ent);
}
- printf ("#undef L\n\n");
+
+
/*
- * Dump sets
+ * Find ranges for each letter for faster searching
*/
- printf ("static const FcLangCharSet fcLangCharSets[] = {\n");
+ setRangeChar = 'a';
+ memset(setRangeStart, '\0', sizeof (setRangeStart));
+ memset(setRangeEnd, '\0', sizeof (setRangeEnd));
for (i = 0; sets[i]; i++)
{
- printf (" { (FcChar8 *) \"%s\",\n"
- " { 1, FcTrue, %d, "
- "(FcCharLeaf **) leaves_%s, "
- "(FcChar16 *) numbers_%s } },\n",
- get_lang(names[i]),
- sets[i]->num, names[i], names[i]);
+ char c = names[i][0];
+
+ while (setRangeChar <= c && c <= 'z')
+ setRangeStart[setRangeChar++ - 'a'] = i;
+ }
+ for (setRangeChar = 'a'; setRangeChar < 'z'; setRangeChar++)
+ setRangeEnd[setRangeChar - 'a'] = setRangeStart[setRangeChar+1-'a'] - 1;
+ setRangeEnd[setRangeChar - 'a'] = i - 1;
+
+ /*
+ * Dump sets start/finish for the fastpath
+ */
+ printf ("static const FcLangCharSetRange fcLangCharSetRanges[] = {\n");
+ for (setRangeChar = 'a'; setRangeChar <= 'z' ; setRangeChar++)
+ {
+ printf (" { %d, %d }, /* %c */\n",
+ setRangeStart[setRangeChar - 'a'],
+ setRangeEnd[setRangeChar - 'a'], setRangeChar);
}
printf ("};\n\n");
+
while (fgets (line, sizeof (line), stdin))
fputs (line, stdout);
git.wh0rd.org / fontconfig.git / blobdiff