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 char *whitespace = " \t\r\n";
   9
  10 void
  11 usage(void)
  12 {
  13         fprint(2, "%s [ -p ] [ -c charset ] [ file ]\n", argv0);
  14         exits("usage");
  15 }
  16
  17 char*
  18 attr(char *s, char *a)
  19 {
  20         char *e, q;
  21
  22         if((s = cistrstr(s, a)) == nil)
  23                 return nil;
  24         s += strlen(a);
  25         while(*s && strchr(whitespace, *s))
  26                 s++;
  27         if(*s++ != '=')
  28                 return nil;
  29         while(*s && strchr(whitespace, *s))
  30                 s++;
  31         q = 0;
  32         if(*s == '"' || *s == '\'')
  33                 q = *s++;
  34         for(e = s; *e; e++){
  35                 if(*e == q)
  36                         break;
  37                 if(isalnum(*e))
  38                         continue;
  39                 if(*e == '-' || *e == '_')
  40                         continue;
  41                 break;
  42         }
  43         if((e - s) > 1)
  44                 return smprint("%.*s", (int)(e - s), s);
  45         return nil;
  46 }
  47
  48 void
  49 main(int argc, char *argv[])
  50 {
  51         int n, q, pfd[2], pflag = 0;
  52         char *arg[4], *s, *e, *p, *g, *a, t;
  53         Rune r;
  54
  55         ARGBEGIN {
  56         case 'c':
  57                 cset = EARGF(usage());
  58                 break;
  59         case 'p':
  60                 pflag = 1;
  61                 break;
  62         default:
  63                 usage();
  64         } ARGEND;
  65
  66         if(*argv){
  67                 close(0);
  68                 if(open(*argv, OREAD) != 1)
  69                         sysfatal("open: %r");
  70         }
  71         nbuf = 0;
  72         p = buf;
  73         g = buf;
  74         while(nbuf < sizeof(buf)-1){
  75                 if((n = read(0, buf + nbuf, sizeof(buf)-1-nbuf)) <= 0)
  76                         break;
  77                 nbuf += n;
  78                 buf[nbuf] = 0;
  79                 if(nbuf == n){
  80                         if(memcmp(p, "\xEF\xBB\xBF", 3)==0){
  81                                 p += 3;
  82                                 nbuf -= 3;
  83                                 cset = "utf";
  84                                 goto Found;
  85                         }
  86                         if(memcmp(p, "\xFE\xFF", 2) == 0){
  87                                 p += 2;
  88                                 nbuf -= 2;
  89                                 cset = "unicode-be";
  90                                 goto Found;
  91                         }
  92                         if(memcmp(p, "\xFF\xFE", 2) == 0){
  93                                 p += 2;
  94                                 nbuf -= 2;
  95                                 cset = "unicode-le";
  96                                 goto Found;
  97                         }
  98                 }
  99                 s = g;
 100                 do {
 101                         if((s = strchr(s, '<')) == nil)
 102                                 break;
 103                         q = 0;
 104                         g = ++s;
 105                         e = buf+nbuf;
 106                         while(s < e){
 107                                 if(*s == '\'' || *s == '"'){
 108                                         if(q == 0)
 109                                                 q = *s;
 110                                         else if(q == *s)
 111                                                 q = 0;
 112                                 } else if(*s == '>' && q == 0){
 113                                         e = s;
 114                                         break;
 115                                 }
 116                                 s++;
 117                         }
 118                         t = *e;
 119                         *e = 0;
 120                         if((a = attr(g, "encoding")) || (a = attr(g, "charset"))){
 121                                 *e = t;
 122                                 cset = a;
 123                                 goto Found;
 124                         }
 125                         *e = t;
 126                         s = ++e;
 127                 } while(t);
 128         }
 129         s = p;
 130         while(s+UTFmax < p+nbuf){
 131                 s += chartorune(&r, s);
 132                 if(r == Runeerror){
 133                         cset = "latin1";
 134                         goto Found;
 135                 }
 136         }
 137         cset = "utf";
 138 Found:
 139         if(pflag){
 140                 print("%s\n", cset);
 141                 exits(0);
 142         }
 143
 144         if(nbuf == 0){
 145                 write(1, p, 0);
 146                 exits(0);
 147         }
 148
 149         if(pipe(pfd) < 0)
 150                 sysfatal("pipe: %r");
 151
 152         switch(rfork(RFFDG|RFREND|RFPROC)){
 153         case -1:
 154                 sysfatal("fork: %r");
 155         case 0:
 156                 dup(pfd[0], 0);
 157                 close(pfd[0]);
 158                 close(pfd[1]);
 159
 160                 arg[0] = "rc";
 161                 arg[1] = "-c";
 162                 arg[2] = smprint("{tcs -f %s || cat} | tcs -f html", cset);
 163                 arg[3] = nil;
 164                 exec("/bin/rc", arg);
 165         }
 166
 167         dup(pfd[1], 1);
 168         close(pfd[0]);
 169         close(pfd[1]);
 170
 171         while(nbuf > 0){
 172                 if(write(1, p, nbuf) != nbuf)
 173                         sysfatal("write: %r");
 174                 p = buf;
 175                 if((nbuf = read(0, p, sizeof(buf))) < 0)
 176                         sysfatal("read: %r");
 177         }
 178         close(1);
 179         waitpid();
 180         exits(0);
 181 }