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] = $item->getAttribute('href');
  36                 if($x === 'lrdd') {
  37                         $decoded = urldecode($item->getAttribute('href'));
  38                         if(preg_match('/acct:([^@]*)@/',$decoded,$matches))
  39                                 $ret['nick'] = $matches[1];
  40                 }
  41         }
  42
  43         // Pull out hCard profile elements
  44
  45         $items = $dom->getElementsByTagName('*');
  46         foreach($items as $item) {
  47                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
  48                         $level2 = $item->getElementsByTagName('*');
  49                         foreach($level2 as $x) {
  50                                 if(attribute_contains($x->getAttribute('class'),'fn'))
  51                                         $ret['fn'] = $x->textContent;
  52                                 if(attribute_contains($x->getAttribute('class'),'photo'))
  53                                         $ret['photo'] = $x->getAttribute('src');
  54                                 if(attribute_contains($x->getAttribute('class'),'key'))
  55                                         $ret['key'] = $x->textContent;
  56                         }
  57                 }
  58         }
  59
  60         return $ret;
  61 }}
  62
  63
  64
  65
  66
  67
  68 if(! function_exists('validate_dfrn')) {
  69 function validate_dfrn($a) {
  70         $errors = 0;
  71         if(! x($a,'key'))
  72                 $errors ++;
  73         if(! x($a,'dfrn-request'))
  74                 $errors ++;
  75         if(! x($a,'dfrn-confirm'))
  76                 $errors ++;
  77         if(! x($a,'dfrn-notify'))
  78                 $errors ++;
  79         if(! x($a,'dfrn-poll'))
  80                 $errors ++;
  81         return $errors;
  82 }}
  83
  84 if(! function_exists('scrape_meta')) {
  85 function scrape_meta($url) {
  86
  87         $ret = array();
  88         $s = fetch_url($url);
  89
  90         if(! $s)
  91                 return $ret;
  92
  93         $dom = HTML5_Parser::parse($s);
  94
  95         if(! $dom)
  96                 return $ret;
  97
  98         $items = $dom->getElementsByTagName('meta');
  99
 100         // get DFRN link elements
 101
 102         foreach($items as $item) {
 103                 $x = $item->getAttribute('name');
 104                 if(substr($x,0,5) == "dfrn-")
 105                         $ret[$x] = $item->getAttribute('content');
 106         }
 107
 108         return $ret;
 109 }}