src/document/html/parse-meta-refresh.c

   1 /* Parse <meta http-equiv="refresh" content="..."> */
   2
   3 #ifdef HAVE_CONFIG_H
   4 #include "config.h"
   5 #endif
   6
   7 #include <errno.h>
   8 #include <stdlib.h>
   9
  10 #include "elinks.h"
  11
  12 #include "document/html/parse-meta-refresh.h"
  13 #include "osdep/ascii.h"
  14 #include "util/string.h"
  15
  16 #define LWS(c) ((c) == ' ' || (c) == ASCII_TAB)
  17
  18 int
  19 html_parse_meta_refresh(const unsigned char *content,
  20                         unsigned long *delay_out,
  21                         unsigned char **url_out)
  22 {
  23         const unsigned char *scan = content;
  24         char *delay_end;
  25         int saw_delay = 0;
  26         const unsigned char *url_begin;
  27         const unsigned char *url_end;
  28
  29         *url_out = NULL;
  30         *delay_out = 0;
  31
  32         while (LWS(*scan))
  33                 ++scan;
  34
  35         /* TODO: Do we need to switch to the "C" locale and back?  */
  36         *delay_out = strtoul(scan, &delay_end, 10);
  37         saw_delay = (scan != (const unsigned char *) delay_end);
  38         scan = (const unsigned char *) delay_end;
  39
  40         if (saw_delay) {
  41                 /* Omit any fractional part.  */
  42                 if (*scan == '.') {
  43                         ++scan;
  44                         while (!(*scan == '\0' || LWS(*scan)
  45                                  || *scan == ';' || *scan == ','))
  46                                 ++scan;
  47                 }
  48
  49                 if (!(*scan == '\0' || LWS(*scan)
  50                       || *scan == ';' || *scan == ',')) {
  51                         /* The delay is followed by garbage.  Give up.  */
  52                         return -1;
  53                 }
  54
  55                 /* Between the delay and the URL, there must be at
  56                  * least one LWS, semicolon, or comma; optionally with
  57                  * more LWS around it.  */
  58                 while (LWS(*scan))
  59                         ++scan;
  60                 if (*scan == ';' || *scan == ',')
  61                         ++scan;
  62         } else {
  63                 /* The delay was not specified.  The delimiter must be
  64                  * a semicolon or a comma, optionally with LWS.  LWS
  65                  * alone does not suffice.  */
  66                 while (*scan != '\0' && *scan != ';' && *scan != ',')
  67                         ++scan;
  68                 if (*scan == ';' || *scan == ',')
  69                         ++scan;
  70         }
  71
  72         while (LWS(*scan))
  73                 ++scan;
  74
  75         /* Presume the URL begins here...  */
  76         url_begin = scan;
  77
  78         /* ..unless there is "URL=" with at least one equals sign,
  79          * and optional spaces.  */
  80         if ((scan[0] == 'U' || scan[0] == 'u')
  81             && (scan[1] == 'R' || scan[1] == 'r')
  82             && (scan[2] == 'L' || scan[2] == 'l')) {
  83                 scan += 3;
  84                 while (LWS(*scan))
  85                         ++scan;
  86                 if (*scan == '=') {
  87                         ++scan;
  88                         while (LWS(*scan))
  89                                 ++scan;
  90                         url_begin = scan;
  91                 }
  92         }
  93
  94         if (*url_begin == '"' || *url_begin == '\'') {
  95                 unsigned char quote = *url_begin++;
  96
  97                 url_end = strchr(url_begin, quote);
  98                 if (url_end == NULL)
  99                         url_end = strchr(url_begin, '\0');
 100         } else {
 101                 url_end = strchr(url_begin, '\0');
 102         }
 103
 104         /* In any case, trim all spaces from the end of the URL.  */
 105         while (url_begin < url_end && LWS(url_end[-1]))
 106                 --url_end;
 107
 108         if (url_begin != url_end) {
 109                 *url_out = memacpy(url_begin, url_end - url_begin);
 110                 if (!*url_out)
 111                         return -1;
 112         } else if (!saw_delay) {
 113                 /* There is no delay and no URL.  */
 114                 return -1;
 115         }
 116
 117         return 0;
 118 }