sys/src/cmd/unicode.c

   1 #include <u.h>
   2 #include <libc.h>
   3 #include <bio.h>
   4
   5 char    usage[] = "unicode { [-t] hex hex ... | hexmin-hexmax ... | [-n] char ... }";
   6 char    hex[] = "0123456789abcdefABCDEF";
   7 int     numout = 0;
   8 int     text = 0;
   9 char    *err;
  10 Biobuf  bout;
  11
  12 char    *range(char*[]);
  13 char    *nums(char*[]);
  14 char    *chars(char*[]);
  15
  16 void
  17 main(int argc, char *argv[])
  18 {
  19         ARGBEGIN{
  20         case 'n':
  21                 numout = 1;
  22                 break;
  23         case 't':
  24                 text = 1;
  25                 break;
  26         }ARGEND
  27         Binit(&bout, 1, OWRITE);
  28         if(argc == 0){
  29                 fprint(2, "usage: %s\n", usage);
  30                 exits("usage");
  31         }
  32         if(!numout && utfrune(argv[0], '-'))
  33                 exits(range(argv));
  34         if(numout || strchr(hex, argv[0][0])==0)
  35                 exits(nums(argv));
  36         exits(chars(argv));
  37 }
  38
  39 char*
  40 range(char *argv[])
  41 {
  42         char *q;
  43         int min, max;
  44         int i;
  45
  46         while(*argv){
  47                 q = *argv;
  48                 if(strchr(hex, q[0]) == 0){
  49     err:
  50                         fprint(2, "unicode: bad range %s\n", *argv);
  51                         return "bad range";
  52                 }
  53                 min = strtoul(q, &q, 16);
  54                 if(min<0 || min>Runemax || *q!='-')
  55                         goto err;
  56                 q++;
  57                 if(strchr(hex, *q) == 0)
  58                         goto err;
  59                 max = strtoul(q, &q, 16);
  60                 if(max<0 || max>Runemax || max<min || *q!=0)
  61                         goto err;
  62                 i = 0;
  63                 do{
  64                         Bprint(&bout, "%.4x %C", min, min);
  65                         i++;
  66                         if(min==max || (i&7)==0)
  67                                 Bprint(&bout, "\n");
  68                         else
  69                                 Bprint(&bout, "\t");
  70                         min++;
  71                 }while(min<=max);
  72                 argv++;
  73         }
  74         return 0;
  75 }
  76
  77 char*
  78 nums(char *argv[])
  79 {
  80         char *q;
  81         Rune r;
  82         int w;
  83
  84         while(*argv){
  85                 q = *argv;
  86                 while(*q){
  87                         w = chartorune(&r, q);
  88                         if(r==Runeerror){
  89                                 char b[UTFmax];
  90
  91                                 if(runetochar(b, &r) != w || memcmp(b, q, w) != 0){
  92                                         fprint(2, "unicode: invalid utf string %s\n", *argv);
  93                                         return "bad utf";
  94                                 }
  95                         }
  96                         Bprint(&bout, "%.4x\n", r);
  97                         q += w;
  98                 }
  99                 argv++;
 100         }
 101         return 0;
 102 }
 103
 104 char*
 105 chars(char *argv[])
 106 {
 107         char *q;
 108         int m;
 109
 110         while(*argv){
 111                 q = *argv;
 112                 if(strchr(hex, q[0]) == 0){
 113     err:
 114                         fprint(2, "unicode: bad unicode value %s\n", *argv);
 115                         return "bad char";
 116                 }
 117                 m = strtoul(q, &q, 16);
 118                 if(m<0 || m>Runemax || *q!=0)
 119                         goto err;
 120                 Bprint(&bout, "%C", m);
 121                 if(!text)
 122                         Bprint(&bout, "\n");
 123                 argv++;
 124         }
 125         return 0;
 126 }