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, *g, *e, *p, *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) != 0)
  69                         sysfatal("open: %r");
  70         }
  71         nbuf = 0;
  72         while(nbuf < sizeof(buf)-1){
  73                 if((n = read(0, buf + nbuf, sizeof(buf)-1-nbuf)) <= 0)
  74                         break;
  75                 nbuf += n;
  76                 buf[nbuf] = 0;
  77         }
  78
  79         p = buf;
  80         if(nbuf >= 3 && memcmp(p, "\xEF\xBB\xBF", 3)==0){
  81                 p += 3;
  82                 nbuf -= 3;
  83                 cset = "utf";
  84                 goto Found;
  85         }
  86         if(nbuf >= 2 && memcmp(p, "\xFE\xFF", 2) == 0){
  87                 p += 2;
  88                 nbuf -= 2;
  89                 cset = "unicode-be";
  90                 goto Found;
  91         }
  92         if(nbuf >= 2 && memcmp(p, "\xFF\xFE", 2) == 0){
  93                 p += 2;
  94                 nbuf -= 2;
  95                 cset = "unicode-le";
  96                 goto Found;
  97         }
  98
  99         s = p;
 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 == '=' && q == 0)
 108                                 q = '=';
 109                         else if(*s == '\'' || *s == '"'){
 110                                 if(q == '=')
 111                                         q = *s;
 112                                 else if(q == *s)
 113                                         q = 0;
 114                         }
 115                         else if(*s == '>' && q != '\'' && q != '"'){
 116                                 e = s;
 117                                 break;
 118                         }
 119                         else if(q == '=' && strchr(whitespace, *s) == nil)
 120                                 q = 0;
 121                         s++;
 122                 }
 123                 t = *e;
 124                 *e = 0;
 125                 if((a = attr(g, "encoding")) != nil || (a = attr(g, "charset")) != nil)
 126                 if(cistrcmp(a, "utf") != 0 && cistrcmp(a, "utf-8") != 0){
 127                         cset = a;
 128                         *e = t;
 129                         break;
 130                 }
 131                 *e = t;
 132                 s = ++e;
 133         } while(t);
 134
 135         s = p;
 136         while(s+UTFmax < p+nbuf){
 137                 s += chartorune(&r, s);
 138                 if(r == Runeerror){
 139                         if(cset == nil)
 140                                 cset = "latin1";
 141                         goto Found;
 142                 }
 143         }
 144         cset = "utf";
 145
 146 Found:
 147         if(pflag){
 148                 print("%s\n", cset);
 149                 exits(0);
 150         }
 151
 152         if(nbuf == 0){
 153                 write(1, p, 0);
 154                 exits(0);
 155         }
 156
 157         if(pipe(pfd) < 0)
 158                 sysfatal("pipe: %r");
 159
 160         switch(rfork(RFFDG|RFREND|RFPROC)){
 161         case -1:
 162                 sysfatal("fork: %r");
 163         case 0:
 164                 dup(pfd[0], 0);
 165                 close(pfd[0]);
 166                 close(pfd[1]);
 167
 168                 arg[0] = "rc";
 169                 arg[1] = "-c";
 170                 arg[2] = smprint("{tcs -f %s || cat} | tcs -f html", cset);
 171                 arg[3] = nil;
 172                 exec("/bin/rc", arg);
 173         }
 174
 175         dup(pfd[1], 1);
 176         close(pfd[0]);
 177         close(pfd[1]);
 178
 179         while(nbuf > 0){
 180                 if(write(1, p, nbuf) != nbuf)
 181                         sysfatal("write: %r");
 182                 p = buf;
 183                 if((nbuf = read(0, p, sizeof(buf))) < 0)
 184                         sysfatal("read: %r");
 185         }
 186         close(1);
 187         waitpid();
 188         exits(0);
 189 }