src/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 FILE *fi;
   9
  10 static void openFile (const char *fname) {
  11   fi = fopen(fname, "r");
  12   if (fi == NULL) { fprintf(stderr, "FATAL: can't open input file: '%s'\n", fname); exit(1); }
  13 }
  14
  15
  16 static char *trim (char *str) {
  17   char *s;
  18   //
  19   if (str == NULL) return NULL;
  20   for (s = str; *s && isspace(*s); ++s) ;
  21   if (s > str) memmove(str, s, strlen(s)+1);
  22   for (s = str+strlen(str)-1; s >= str && isspace(*s); --s) ;
  23   s[1] = 0;
  24   return str;
  25 }
  26
  27
  28 static int hex2num (char *s) {
  29   char *end;
  30   int res;
  31   //
  32   trim(s);
  33   if (!s[0]) return -1;
  34   res = strtol(s, &end, 16);
  35   if (end[0] || res < 0) return -1;
  36   return res;
  37 }
  38
  39
  40 typedef struct {
  41   int code;
  42   const char *name;
  43   const char *class;
  44   int upper;
  45   int lower;
  46 } UCInfo;
  47
  48
  49 static char *stx = NULL;
  50
  51 static char *tok (void) {
  52   char *s = stx;
  53   for (; *stx && *stx != ';'; ++stx) ;
  54   if (*stx) *stx++ = 0;
  55   return s;
  56 }
  57
  58
  59 // <0: eof; 0: bad record; >0: good record
  60 static int readRecord (UCInfo *ui) {
  61   static char str[8192];
  62   char *tk, *u, *l;
  63   int f;
  64   //
  65   if (fgets(str, sizeof(str)-1, fi) == NULL) return -1;
  66   stx = trim(str);
  67   if ((tk = tok()) == NULL) return 0;
  68   if ((ui->code = hex2num(tk)) < 0) return 0;
  69   if (ui->code > 65535) return -1;
  70   if ((ui->name = trim(tok())) == NULL) return 0;
  71   if ((ui->class = trim(tok())) == NULL) return 0;
  72   // skip unused fields
  73   for (f = 9; f > 0; --f) if (tok() == NULL) { printf("%d\n", f); return 0; }
  74   if ((u = trim(tok())) == NULL) return 0;
  75   if ((l = trim(tok())) == NULL) return 0;
  76   if (!u[0]) ui->upper = ui->code; else ui->upper = hex2num(u);
  77   if (!l[0]) ui->lower = ui->code; else ui->lower = hex2num(l);
  78   if (ui->upper < 0 || ui->lower < 0) return 0;
  79   if (ui->upper > 65535 || ui->lower > 65535) abort();
  80   return 1;
  81 }
  82
  83
  84 static struct {
  85   int code, l, u;
  86 } map[65535];
  87
  88
  89 int main (int argc, char *argv[]) {
  90   UCInfo ui;
  91   int rc, f, totalmap = 0;
  92   FILE *fo;
  93   //
  94   if (argc != 3) { fprintf(stderr, "usage: %s unitable.c unitable.txt\n", argv[0]); exit(1); }
  95   //
  96   openFile(argv[2]);
  97   for (;;) {
  98     rc = readRecord(&ui);
  99     if (rc < 0) break;
 100     if (rc == 0) continue;
 101     if (ui.code < 128) continue;
 102     if (strcmp(ui.class, "Lu") != 0 && strcmp(ui.class, "Ll") != 0) continue;
 103     if (ui.upper == ui.code && ui.lower == ui.code) continue;
 104     //
 105     for (f = 0; f < totalmap; ++f) {
 106       if (map[f].code == ui.code) { fprintf(stderr, "FATAL: duplicate entries in the map!\n"); exit(1); }
 107       if (map[f].code > ui.code) { fprintf(stderr, "FATAL: invalid entry order in the map!\n"); exit(1); }
 108     }
 109     //
 110     map[totalmap].code = ui.code;
 111     map[totalmap].l = ui.lower;
 112     map[totalmap].u = ui.upper;
 113     ++totalmap;
 114     if (totalmap > 65535) { fprintf(stderr, "FATAL: too many entries in the map!\n"); exit(1); }
 115   }
 116   fclose(fi);
 117   //
 118   fo = fopen(argv[1], "w");
 119   if (fo == NULL) { fprintf(stderr, "FATAL: can't create output file: '%s'\n", argv[1]); exit(1); }
 120   fprintf(fo, "static const struct casemap unicode_case_mapping[%d] = {\n", totalmap);
 121   for (f = 0; f < totalmap; ++f) fprintf(fo, "{0x%04x,0x%04x,0x%04x},\n", map[f].code, map[f].l, map[f].u);
 122   fprintf(fo, "%s\n", "};");
 123   fclose(fo);
 124   return 0;
 125 }