sys/src/libhttpd/httpunesc.c

   1 #include <u.h>
   2 #include <libc.h>
   3 #include <bin.h>
   4 #include <httpd.h>
   5
   6 /*
   7  *  go from http with latin1 escapes to utf,
   8  *  we assume that anything >= Runeself is already in utf
   9  */
  10 char *
  11 httpunesc(HConnect *cc, char *s)
  12 {
  13         char *t, *v;
  14         int c;
  15         Htmlesc *e;
  16
  17         v = halloc(cc, UTFmax*strlen(s) + 1);
  18         for(t = v; c = *s;){
  19                 if(c == '&'){
  20                         if(s[1] == '#' && s[2] && s[3] && s[4] && s[5] == ';'){
  21                                 c = strtol(s+2, 0, 10);
  22                                 if(c < Runeself){
  23                                         *t++ = c;
  24                                         s += 6;
  25                                         continue;
  26                                 }
  27                                 if(c < 256 && c >= 161){
  28                                         e = &htmlesc[c-161];
  29                                         t += runetochar(t, &e->value);
  30                                         s += 6;
  31                                         continue;
  32                                 }
  33                         } else {
  34                                 for(e = htmlesc; e->name != nil; e++)
  35                                         if(strncmp(e->name, s, strlen(e->name)) == 0)
  36                                                 break;
  37                                 if(e->name != nil){
  38                                         t += runetochar(t, &e->value);
  39                                         s += strlen(e->name);
  40                                         continue;
  41                                 }
  42                         }
  43                 }
  44                 *t++ = c;
  45                 s++;
  46         }
  47         *t = 0;
  48         return v;
  49 }