agls function scraper
[disclosr.git] / admin / genericAgencyFixer.php
blob:a/admin/genericAgencyFixer.php -> blob:b/admin/genericAgencyFixer.php
  <?php
   
  include_once("../include/common.inc.php");
  require($basePath . 'lib/phpquery/phpQuery/phpQuery.php');
   
  setlocale(LC_CTYPE, 'C');
   
   
  $db = $server->get_db('disclosr-agencies');
   
  try {
  $agencies = $db->get_view("app", "byCanonicalName", null, true)->rows;
  //print_r($rows);
  foreach ($agencies as $agency) {
  //echo $agency->value->name . " ".$agency->value->website."<br />\n";
  // print_r($agency);
  //hasRestricitiveLicence" hasRestrictiveLicense -> has Restrictive Licence
  // "hasYoutube" -> Tube
  // "comment" -> "comments"
  if (!isset($agency->value->metaTags) && isset($agency->value->website)) {
  echo $agency->value->name . " ".$agency->value->website."<br />\n";
  $agency->value->metaTags = Array();
  $request = Requests::get($agency->value->website);
  $html = phpQuery::newDocumentHTML($request->body);
  phpQuery::selectDocument($html);
  foreach (pq('meta')->elements as $meta) {
  $tagName = $meta->getAttribute('name');;
  $content = $meta->getAttribute('content');
  if ($tagName != "") {
  echo "$tagName == $content <br>\n";
  $agency->value->metaTags[$tagName] = $content;
  }
  }
  //print_r($agency->value->metaTags);
  $db->save($agency->value);
  echo "<hr>";
  flush();
  }
  }
  } catch (SetteeRestClientException $e) {
  setteErrorHandler($e);
  }
  ?>