include/Scrape.php

   1 <?php
   2
   3 require_once('library/HTML5/Parser.php');
   4
   5 if(! function_exists('attribute_contains')) {
   6 function attribute_contains($attr,$s) {
   7         $a = explode(' ', $attr);
   8         if(count($a) && in_array($s,$a))
   9                 return true;
  10         return false;
  11 }}
  12
  13
  14 if(! function_exists('scrape_dfrn')) {
  15 function scrape_dfrn($url) {
  16
  17         $ret = array();
  18         $s = fetch_url($url);
  19
  20         if(! $s)
  21                 return $ret;
  22
  23         $dom = HTML5_Parser::parse($s);
  24
  25         if(! $dom)
  26                 return $ret;
  27
  28         $items = $dom->getElementsByTagName('link');
  29
  30         // get DFRN link elements
  31
  32         foreach($items as $item) {
  33                 $x = $item->getAttribute('rel');
  34                 if(substr($x,0,5) == "dfrn-")
  35                         $ret[$x] = $item->getAttribute('href');
  36         }
  37
  38         // Pull out hCard profile elements
  39
  40         $items = $dom->getElementsByTagName('*');
  41         foreach($items as $item) {
  42                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
  43                         $level2 = $item->getElementsByTagName('*');
  44                         foreach($level2 as $x) {
  45                                 if(attribute_contains($x->getAttribute('class'),'fn'))
  46                                         $ret['fn'] = $x->textContent;
  47                                 if(attribute_contains($x->getAttribute('class'),'photo'))
  48                                         $ret['photo'] = $x->getAttribute('src');
  49                                 if(attribute_contains($x->getAttribute('class'),'key'))
  50                                         $ret['key'] = $x->textContent;
  51                         }
  52                 }
  53         }
  54
  55         return $ret;
  56 }}
  57
  58
  59
  60
  61
  62
  63 if(! function_exists('validate_dfrn')) {
  64 function validate_dfrn($a) {
  65         $errors = 0;
  66         if(! x($a,'key'))
  67                 $errors ++;
  68         if(! x($a,'dfrn-request'))
  69                 $errors ++;
  70         if(! x($a,'dfrn-confirm'))
  71                 $errors ++;
  72         if(! x($a,'dfrn-notify'))
  73                 $errors ++;
  74         if(! x($a,'dfrn-poll'))
  75                 $errors ++;
  76         return $errors;
  77 }}
  78
  79 if(! function_exists('scrape_meta')) {
  80 function scrape_meta($url) {
  81
  82         $ret = array();
  83         $s = fetch_url($url);
  84
  85         if(! $s)
  86                 return $ret;
  87
  88         $dom = HTML5_Parser::parse($s);
  89
  90         if(! $dom)
  91                 return $ret;
  92
  93         $items = $dom->getElementsByTagName('meta');
  94
  95         // get DFRN link elements
  96
  97         foreach($items as $item) {
  98                 $x = $item->getAttribute('name');
  99                 if(substr($x,0,5) == "dfrn-")
 100                         $ret[$x] = $item->getAttribute('content');
 101         }
 102
 103         return $ret;
 104 }}