els_update_search.php

<?php

  $article = $argv[1];

  $xml = new XMLReader();
  if(!$xml->open($article)){
    die('Failed to open file!');
  }
  $rc = 1;

  $startTimeAll = microtime(true);

  while ($xml->read()){
    if ($xml->name === "page") {
      $page = array();
      $node = new SimpleXMLElement($xml->readOuterXML());
      $page['id'] = (string)$node->id;
      $page['title'] = (string)$node->title;
      $page['text'] = (string)$node->revision->text;
      if ($page['id'] != ""){
        $update_startTime = microtime(true);
        $json_doc = array(
                          "_key" => $page['id'],
                          "title" => $page['id'] . "_" . $page['title'],
                          "text" => $page['text']
                         );
        $qry = json_encode($json_doc);

        $url="http://localhost:9200/wikipedia/text/" . $page['id'];
        $req="PUT";
        $ch = curl_init();
        curl_setopt($ch, CURLOPT_URL, $url);
        curl_setopt($ch, CURLOPT_PORT, 9200);
        curl_setopt($ch, CURLOPT_CUSTOMREQUEST, $req);
        curl_setopt($ch, CURLOPT_POSTFIELDS, $qry);
        curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
        curl_setopt($ch, CURLOPT_HEADER, 0);
        curl_exec ($ch);
        curl_close ($ch);

        $updateTime = microtime(true) - $update_startTime;

        $json_doc = array(
                          "from" => "0",
                          "size" => "0",
                          "fields" => array("id"),
                          "query" => array(
                             "query_string" => array(
                               "default_field" => "title",
                               "query" => "\"" . $page['id'] . "_" . $page['title'] . "\""
                             ),
                           )
                         );
        $qry = json_encode($json_doc);

        $search_startTime = microtime(true);

        $url="http://localhost:9200/wikipedia/text/_search";
        $req="GET";
        $ch = curl_init();
        curl_setopt($ch, CURLOPT_URL, $url);
        curl_setopt($ch, CURLOPT_PORT, 9200);
        curl_setopt($ch, CURLOPT_CUSTOMREQUEST, $req);
        curl_setopt($ch, CURLOPT_POSTFIELDS, $qry);
        curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
        curl_setopt($ch, CURLOPT_HEADER, 0);

        $count = 0;
        while($count == 0){
          $data = curl_exec ($ch);
          $json = json_decode($data);
          $count =  $json->hits->total;
          if($count == 1){
            break;
          } else {
            //sleep 10msec
            usleep(100);
          }
        }
        curl_close ($ch);

        $searchTime = microtime(true) - $search_startTime;

        $AllTime = microtime(true) - $update_startTime;

        $now = microtime(true);
        echo "${rc}, ${now}, ${updateTime}, ${searchTime}, ${AllTime}\n";
        $rc++;
        if($rc > 1000){
          break;
        }
      }
    }
  }

  $xml->close();
  $endTimeAll = microtime(true);
  $elapsedTimeAll = $endTimeAll - $startTimeAll;

  $rc--;
  echo $rc . " records done. Total = " . $elapsedTimeAll . "\n";


?>