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