sys/src/cmd/uhtml.c

   1 #include <u.h>
   2 #include <libc.h>
   3 #include <ctype.h>
   4
   5 int nbuf;
   6 char buf[64*1024+1];
   7 char *cset = nil;
   8
   9 void
  10 usage(void)
  11 {
  12         fprint(2, "%s [ -h ] [ -c charset ] [ file ]\n", argv0);
  13         exits("usage");
  14 }
  15
  16 char*
  17 strval(char *s)
  18 {
  19         char *e, q;
  20
  21         while(strchr("\t ", *s))
  22                 s++;
  23         q = 0;
  24         if(*s == '"' || *s == '\'')
  25                 q = *s++;
  26         for(e = s; *e; e++){
  27                 if(*e == q)
  28                         break;
  29                 if(isalnum(*e))
  30                         continue;
  31                 if(*e == '-' || *e == '_')
  32                         continue;
  33                 break;
  34         }
  35         if(e - s > 1)
  36                 return smprint("%.*s", (int)(e-s), s);
  37         return nil;
  38 }
  39
  40 void
  41 main(int argc, char *argv[])
  42 {
  43         int n, pfd[2], pflag = 0;
  44         char *arg[4], *s, *p;
  45         Rune r;
  46
  47         ARGBEGIN {
  48         case 'h':
  49                 usage();
  50         case 'c':
  51                 cset = EARGF(usage());
  52                 break;
  53         case 'p':
  54                 pflag = 1;
  55                 break;
  56         } ARGEND;
  57
  58         if(*argv){
  59                 close(0);
  60                 if(open(*argv, OREAD) != 1)
  61                         sysfatal("open: %r");
  62         }
  63         nbuf = 0;
  64         p = buf;
  65         while(nbuf < sizeof(buf)-1){
  66                 if((n = read(0, buf + nbuf, sizeof(buf)-1-nbuf)) <= 0)
  67                         break;
  68                 nbuf += n;
  69                 buf[nbuf] = 0;
  70                 if(nbuf == n){
  71                         if(memcmp(p, "\xEF\xBB\xBF", 3)==0){
  72                                 p += 3;
  73                                 cset = "utf";
  74                                 break;
  75                         }
  76                         if(memcmp(p, "\xFE\xFF", 2) == 0){
  77                                 p += 2;
  78                                 cset = "unicode-be";
  79                                 break;
  80                         }
  81                         if(memcmp(p, "\xFF\xFE", 2) == 0){
  82                                 p += 2;
  83                                 cset = "unicode-le";
  84                                 break;
  85                         }
  86                 }
  87                 if(s = cistrstr(p, "encoding="))
  88                         if(s = strval(s+9)){
  89                                 cset = s;
  90                                 break;
  91                         }
  92                 if(s = cistrstr(p, "charset="))
  93                         if(s = strval(s+8)){
  94                                 cset = s;
  95                                 break;
  96                         }
  97         }
  98         nbuf -= p - buf;
  99
 100         if(cset == nil){
 101                 cset = "utf";
 102                 s = p;
 103                 while(s+UTFmax < p+nbuf){
 104                         s += chartorune(&r, s);
 105                         if(r == Runeerror){
 106                                 cset = "latin1";
 107                                 break;
 108                         }
 109                 }
 110         }
 111
 112         if(pflag){
 113                 print("%s\n", cset);
 114                 exits(0);
 115         }
 116
 117         if(nbuf == 0){
 118                 write(1, p, 0);
 119                 exits(0);
 120         }
 121
 122         if(pipe(pfd) < 0)
 123                 sysfatal("pipe: %r");
 124
 125         switch(rfork(RFFDG|RFREND|RFPROC)){
 126         case -1:
 127                 sysfatal("fork: %r");
 128         case 0:
 129                 dup(pfd[0], 0);
 130                 close(pfd[0]);
 131                 close(pfd[1]);
 132
 133                 arg[0] = "rc";
 134                 arg[1] = "-c";
 135                 arg[2] = smprint("{tcs -f %s | tcs -f html} || cat", cset);
 136                 arg[3] = nil;
 137                 exec("/bin/rc", arg);
 138         }
 139
 140         dup(pfd[1], 1);
 141         close(pfd[0]);
 142         close(pfd[1]);
 143
 144         while(nbuf > 0){
 145                 if(write(1, p, nbuf) != nbuf)
 146                         sysfatal("write: %r");
 147                 p = buf;
 148                 if((nbuf = read(0, p, sizeof(buf))) < 0)
 149                         sysfatal("read: %r");
 150         }
 151         close(1);
 152         waitpid();
 153         exits(0);
 154 }