include/Scrape.php

   1 <?php
   2
   3 require_once('library/HTML5/Parser.php');
   4
   5 if(! function_exists('scrape_dfrn')) {
   6 function scrape_dfrn($url) {
   7
   8         $ret = array();
   9         $s = fetch_url($url);
  10
  11         if(! $s)
  12                 return $ret;
  13
  14         $dom = HTML5_Parser::parse($s);
  15
  16         if(! $dom)
  17                 return $ret;
  18
  19         $items = $dom->getElementsByTagName('link');
  20
  21         // get DFRN link elements
  22
  23         foreach($items as $item) {
  24                 $x = $item->getAttribute('rel');
  25                 if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml'))
  26                         $ret['feed_atom'] = $item->getAttribute('href');
  27                 if(substr($x,0,5) == "dfrn-")
  28                         $ret[$x] = $item->getAttribute('href');
  29                 if($x === 'lrdd') {
  30                         $decoded = urldecode($item->getAttribute('href'));
  31                         if(preg_match('/acct:([^@]*)@/',$decoded,$matches))
  32                                 $ret['nick'] = $matches[1];
  33                 }
  34         }
  35
  36         // Pull out hCard profile elements
  37
  38         $items = $dom->getElementsByTagName('*');
  39         foreach($items as $item) {
  40                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
  41                         $level2 = $item->getElementsByTagName('*');
  42                         foreach($level2 as $x) {
  43                                 if(attribute_contains($x->getAttribute('class'),'fn'))
  44                                         $ret['fn'] = $x->textContent;
  45                                 if(attribute_contains($x->getAttribute('class'),'photo'))
  46                                         $ret['photo'] = $x->getAttribute('src');
  47                                 if(attribute_contains($x->getAttribute('class'),'key'))
  48                                         $ret['key'] = $x->textContent;
  49                         }
  50                 }
  51         }
  52
  53         return $ret;
  54 }}
  55
  56
  57
  58
  59
  60
  61 if(! function_exists('validate_dfrn')) {
  62 function validate_dfrn($a) {
  63         $errors = 0;
  64         if(! x($a,'key'))
  65                 $errors ++;
  66         if(! x($a,'dfrn-request'))
  67                 $errors ++;
  68         if(! x($a,'dfrn-confirm'))
  69                 $errors ++;
  70         if(! x($a,'dfrn-notify'))
  71                 $errors ++;
  72         if(! x($a,'dfrn-poll'))
  73                 $errors ++;
  74         return $errors;
  75 }}
  76
  77 if(! function_exists('scrape_meta')) {
  78 function scrape_meta($url) {
  79
  80         $ret = array();
  81         $s = fetch_url($url);
  82
  83         if(! $s)
  84                 return $ret;
  85
  86         $dom = HTML5_Parser::parse($s);
  87
  88         if(! $dom)
  89                 return $ret;
  90
  91         $items = $dom->getElementsByTagName('meta');
  92
  93         // get DFRN link elements
  94
  95         foreach($items as $item) {
  96                 $x = $item->getAttribute('name');
  97                 if(substr($x,0,5) == "dfrn-")
  98                         $ret[$x] = $item->getAttribute('content');
  99         }
 100
 101         return $ret;
 102 }}
 103
 104
 105 if(! function_exists('scrape_vcard')) {
 106 function scrape_vcard($url) {
 107
 108         $ret = array();
 109         $s = fetch_url($url);
 110
 111         if(! $s)
 112                 return $ret;
 113
 114         $dom = HTML5_Parser::parse($s);
 115
 116         if(! $dom)
 117                 return $ret;
 118
 119         // Pull out hCard profile elements
 120
 121         $items = $dom->getElementsByTagName('*');
 122         foreach($items as $item) {
 123                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
 124                         $level2 = $item->getElementsByTagName('*');
 125                         foreach($level2 as $x) {
 126                                 if(attribute_contains($x->getAttribute('class'),'fn'))
 127                                         $ret['fn'] = $x->textContent;
 128                                 if((attribute_contains($x->getAttribute('class'),'photo'))
 129                                         || (attribute_contains($x->getAttribute('class'),'avatar')))
 130                                         $ret['photo'] = $x->getAttribute('src');
 131                                 if((attribute_contains($x->getAttribute('class'),'nickname'))
 132                                         || (attribute_contains($x->getAttribute('class'),'uid')))
 133                                         $ret['nick'] = $x->textContent;
 134                         }
 135                 }
 136         }
 137
 138         return $ret;
 139 }}
 140
 141
 142 if(! function_exists('scrape_feed')) {
 143 function scrape_feed($url) {
 144
 145         $ret = array();
 146         $s = fetch_url($url);
 147
 148         if(! $s)
 149                 return $ret;
 150
 151         $dom = HTML5_Parser::parse($s);
 152
 153         if(! $dom)
 154                 return $ret;
 155
 156         $items = $dom->getElementsByTagName('link');
 157
 158         // get Atom link elements
 159
 160         foreach($items as $item) {
 161                 $x = $item->getAttribute('rel');
 162                 if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml'))
 163                         $ret['feed_atom'] = $item->getAttribute('href');
 164         }
 165
 166         return $ret;
 167 }}