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         $a = get_app();
   9
  10         $ret = array();
  11         $s = fetch_url($url);
  12
  13         if(! $s)
  14                 return $ret;
  15
  16         $headers = $a->get_curl_headers();
  17         $lines = explode("\n",$headers);
  18         if(count($lines)) {
  19                 foreach($lines as $line) {
  20                         // don't try and run feeds through the html5 parser
  21                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
  22                                 return ret;
  23                 }
  24         }
  25
  26
  27         $dom = HTML5_Parser::parse($s);
  28
  29         if(! $dom)
  30                 return $ret;
  31
  32         $items = $dom->getElementsByTagName('link');
  33
  34         // get DFRN link elements
  35
  36         foreach($items as $item) {
  37                 $x = $item->getAttribute('rel');
  38                 if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml'))
  39                         $ret['feed_atom'] = $item->getAttribute('href');
  40                 if(substr($x,0,5) == "dfrn-")
  41                         $ret[$x] = $item->getAttribute('href');
  42                 if($x === 'lrdd') {
  43                         $decoded = urldecode($item->getAttribute('href'));
  44                         if(preg_match('/acct:([^@]*)@/',$decoded,$matches))
  45                                 $ret['nick'] = $matches[1];
  46                 }
  47         }
  48
  49         // Pull out hCard profile elements
  50
  51         $items = $dom->getElementsByTagName('*');
  52         foreach($items as $item) {
  53                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
  54                         $level2 = $item->getElementsByTagName('*');
  55                         foreach($level2 as $x) {
  56                                 if(attribute_contains($x->getAttribute('class'),'fn'))
  57                                         $ret['fn'] = $x->textContent;
  58                                 if(attribute_contains($x->getAttribute('class'),'photo'))
  59                                         $ret['photo'] = $x->getAttribute('src');
  60                                 if(attribute_contains($x->getAttribute('class'),'key'))
  61                                         $ret['key'] = $x->textContent;
  62                         }
  63                 }
  64         }
  65
  66         return $ret;
  67 }}
  68
  69
  70
  71
  72
  73
  74 if(! function_exists('validate_dfrn')) {
  75 function validate_dfrn($a) {
  76         $errors = 0;
  77         if(! x($a,'key'))
  78                 $errors ++;
  79         if(! x($a,'dfrn-request'))
  80                 $errors ++;
  81         if(! x($a,'dfrn-confirm'))
  82                 $errors ++;
  83         if(! x($a,'dfrn-notify'))
  84                 $errors ++;
  85         if(! x($a,'dfrn-poll'))
  86                 $errors ++;
  87         return $errors;
  88 }}
  89
  90 if(! function_exists('scrape_meta')) {
  91 function scrape_meta($url) {
  92
  93         $a = get_app();
  94
  95         $ret = array();
  96         $s = fetch_url($url);
  97
  98         if(! $s)
  99                 return $ret;
 100
 101         $headers = $a->get_curl_headers();
 102         $lines = explode("\n",$headers);
 103         if(count($lines)) {
 104                 foreach($lines as $line) {
 105                         // don't try and run feeds through the html5 parser
 106                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 107                                 return ret;
 108                 }
 109         }
 110
 111
 112
 113         $dom = HTML5_Parser::parse($s);
 114
 115         if(! $dom)
 116                 return $ret;
 117
 118         $items = $dom->getElementsByTagName('meta');
 119
 120         // get DFRN link elements
 121
 122         foreach($items as $item) {
 123                 $x = $item->getAttribute('name');
 124                 if(substr($x,0,5) == "dfrn-")
 125                         $ret[$x] = $item->getAttribute('content');
 126         }
 127
 128         return $ret;
 129 }}
 130
 131
 132 if(! function_exists('scrape_vcard')) {
 133 function scrape_vcard($url) {
 134
 135         $a = get_app();
 136
 137         $ret = array();
 138         $s = fetch_url($url);
 139
 140         if(! $s)
 141                 return $ret;
 142
 143         $headers = $a->get_curl_headers();
 144         $lines = explode("\n",$headers);
 145         if(count($lines)) {
 146                 foreach($lines as $line) {
 147                         // don't try and run feeds through the html5 parser
 148                         if(stristr($line,'content-type:') && ((stristr($line,'application/atom+xml')) || (stristr($line,'application/rss+xml'))))
 149                                 return ret;
 150                 }
 151         }
 152
 153         $dom = HTML5_Parser::parse($s);
 154
 155         if(! $dom)
 156                 return $ret;
 157
 158         // Pull out hCard profile elements
 159
 160         $items = $dom->getElementsByTagName('*');
 161         foreach($items as $item) {
 162                 if(attribute_contains($item->getAttribute('class'), 'vcard')) {
 163                         $level2 = $item->getElementsByTagName('*');
 164                         foreach($level2 as $x) {
 165                                 if(attribute_contains($x->getAttribute('class'),'fn'))
 166                                         $ret['fn'] = $x->textContent;
 167                                 if((attribute_contains($x->getAttribute('class'),'photo'))
 168                                         || (attribute_contains($x->getAttribute('class'),'avatar')))
 169                                         $ret['photo'] = $x->getAttribute('src');
 170                                 if((attribute_contains($x->getAttribute('class'),'nickname'))
 171                                         || (attribute_contains($x->getAttribute('class'),'uid')))
 172                                         $ret['nick'] = $x->textContent;
 173                         }
 174                 }
 175         }
 176
 177         return $ret;
 178 }}
 179
 180
 181 if(! function_exists('scrape_feed')) {
 182 function scrape_feed($url) {
 183
 184         $a = get_app();
 185
 186         $ret = array();
 187         $s = fetch_url($url);
 188
 189         if(! $s)
 190                 return $ret;
 191
 192         $headers = $a->get_curl_headers();
 193         $lines = explode("\n",$headers);
 194         if(count($lines)) {
 195                 foreach($lines as $line) {
 196                         if(stristr($line,'content-type:')) {
 197                                 if(stristr($line,'application/atom+xml')) {
 198                                         $ret['feed_atom'] = $url;
 199                                         return $ret;
 200                                 }
 201                                 if(stristr($line,'application/rss+xml')) {
 202                                         $ret['feed_rss'] = $url;
 203                                         return ret;
 204                                 }
 205                         }
 206                 }
 207         }
 208
 209         $dom = HTML5_Parser::parse($s);
 210
 211         if(! $dom)
 212                 return $ret;
 213
 214         $items = $dom->getElementsByTagName('link');
 215
 216         // get Atom link elements
 217
 218         foreach($items as $item) {
 219                 $x = $item->getAttribute('rel');
 220                 if(($x === 'alternate') && ($item->getAttribute('type') === 'application/atom+xml'))
 221                         $ret['feed_atom'] = $item->getAttribute('href');
 222                 if(($x === 'alternate') && ($item->getAttribute('type') === 'application/rss+xml'))
 223                         $ret['feed_rss'] = $item->getAttribute('href');
 224         }
 225
 226         return $ret;
 227 }}