tools/unigen.c

   1 #include <ctype.h>
   2 #include <stdio.h>
   3 #include <stdlib.h>
   4 #include <string.h>
   5 #include <unistd.h>
   6
   7
   8 static char *trim (char *str) {
   9   char *s;
  10   if (str == NULL) return NULL;
  11   for (s = str; *s && isspace(*s); ++s) ;
  12   if (s > str) memmove(str, s, strlen(s)+1);
  13   for (s = str+strlen(str)-1; s >= str && isspace(*s); --s) ;
  14   s[1] = 0;
  15   return str;
  16 }
  17
  18
  19 static int hex2num (char *s) {
  20   char *end;
  21   int res;
  22   trim(s);
  23   if (!s[0]) return -1;
  24   res = strtol(s, &end, 16);
  25   if (end[0] || res < 0) return -1;
  26   return res;
  27 }
  28
  29
  30 typedef struct {
  31   int code;
  32   const char *name;
  33   const char *class;
  34   int upper;
  35   int lower;
  36 } UCInfo;
  37
  38
  39 static char *stx = NULL;
  40
  41 static char *tok (void) {
  42   char *s = stx;
  43   for (; *stx && *stx != ';'; ++stx) ;
  44   if (*stx) *stx++ = 0;
  45   return s;
  46 }
  47
  48
  49 // <0: eof; 0: bad record; >0: good record
  50 static int read_record (FILE *fi, UCInfo *ui) {
  51   static char str[8192];
  52   char *tk, *u, *l;
  53   int f;
  54   if (fgets(str, sizeof(str)-1, fi) == NULL) return -1;
  55   stx = trim(str);
  56   if ((tk = tok()) == NULL) return 0;
  57   if ((ui->code = hex2num(tk)) < 0) return 0;
  58   if (ui->code > 65535) return -1;
  59   if ((ui->name = trim(tok())) == NULL) return 0;
  60   if ((ui->class = trim(tok())) == NULL) return 0;
  61   // skip unused fields
  62   for (f = 9; f > 0; --f) if (tok() == NULL) { printf("%d\n", f); return 0; }
  63   if ((u = trim(tok())) == NULL) return 0;
  64   if ((l = trim(tok())) == NULL) return 0;
  65   if (!u[0]) ui->upper = ui->code; else ui->upper = hex2num(u);
  66   if (!l[0]) ui->lower = ui->code; else ui->lower = hex2num(l);
  67   if (ui->upper < 0 || ui->lower < 0) return 0;
  68   if (ui->upper > 65535 || ui->lower > 65535) abort();
  69   return 1;
  70 }
  71
  72
  73 typedef struct {
  74   int code, l, u;
  75 } mm;
  76
  77 static mm map[65535];
  78
  79
  80 int main (int argc, char *argv[]) {
  81   UCInfo ui;
  82   int rc, f, totalmap = 0;
  83   FILE *fi, *fo;
  84   if (argc != 3) { fprintf(stderr, "usage: %s unitable.c unitable.txt\n", argv[0]); exit(1); }
  85   if ((fi = fopen(argv[2], "r")) == NULL) { fprintf(stderr, "FATAL: can't open input file: '%s'\n", argv[2]); exit(1); }
  86   for (;;) {
  87     rc = read_record(fi, &ui);
  88     if (rc < 0) break;
  89     if (rc == 0) continue;
  90     if (ui.code < 128) continue;
  91     if (strcmp(ui.class, "Lu") != 0 && strcmp(ui.class, "Ll") != 0) continue;
  92     if (ui.upper == ui.code && ui.lower == ui.code) continue;
  93     for (f = 0; f < totalmap; ++f) {
  94       if (map[f].code == ui.code) { fprintf(stderr, "FATAL: duplicate entries in the map!\n"); exit(1); }
  95       if (map[f].code > ui.code) { fprintf(stderr, "FATAL: invalid entry order in the map!\n"); exit(1); }
  96     }
  97     if (totalmap > 65535) { fprintf(stderr, "FATAL: too many entries in the map!\n"); exit(1); }
  98     map[totalmap].code = ui.code;
  99     map[totalmap].l = ui.lower;
 100     map[totalmap].u = ui.upper;
 101     ++totalmap;
 102   }
 103   fclose(fi);
 104   qsort(map, totalmap, sizeof(map[0]), ({
 105     int cmp (const void *p0, const void *p1) {
 106       const mm *i0 = (const mm *)p0;
 107       const mm *i1 = (const mm *)p1;
 108       return (i0->code < i1->code ? -1 : (i0->code > i1->code ? 1 : 0));
 109     }
 110     cmp;
 111   }));
 112   fo = fopen(argv[1], "w");
 113   if (fo == NULL) { fprintf(stderr, "FATAL: can't create output file: '%s'\n", argv[1]); exit(1); }
 114   fprintf(fo, "static const struct casemap unicode_case_mapping[%d] = {\n", totalmap);
 115   for (f = 0; f < totalmap; ++f) fprintf(fo, "{0x%04x,0x%04x,0x%04x},\n", map[f].code, map[f].l, map[f].u);
 116   fprintf(fo, "%s\n", "};");
 117   fclose(fo);
 118   return 0;
 119 }