patents_search_php/site/solr.php

<?php

function solr_search($search, $core, $sort){

  if ($sort == 'relevance'){
    // Assemble a query string to send to Solr. This uses the Solr hostname from config.env. Solr's query syntax can be found at many sites including https://lucene.apache.org/solr/guide/6_6/the-standard-query-parser.html
    $solrurl = 'http://' . $_ENV["SOLR_HOSTNAME"] . ':' . $_ENV["SOLR_PORT"] . '/solr/' . $core . '/select?q.op=OR&q=content%3A' . $search . '&wt=json';
  }
  else{
    // Assemble a query string to send to Solr. This uses the Solr hostname from config.env. Solr's query syntax can be found at many sites including https://lucene.apache.org/solr/guide/6_6/the-standard-query-parser.html
    $solrurl = 'http://' . $_ENV["SOLR_HOSTNAME"] . ':' . $_ENV["SOLR_PORT"] . '/solr/' . $core . '/select?q.op=OR&q=content%3A' . $search . '&wt=json&sort=' . $sort . '%20asc';
  }

  // Perform Curl request on the Solr API
  $ch = curl_init();
  curl_setopt($ch, CURLOPT_URL, $solrurl);
  curl_setopt($ch, CURLOPT_RETURNTRANSFER, TRUE);
  curl_setopt($ch, CURLOPT_HEADER, FALSE);
  curl_setopt($ch, CURLOPT_CUSTOMREQUEST, 'GET');
  $response = curl_exec($ch);
  curl_close($ch);

  // Turn the API response into useful Json
  $json = json_decode($response);

  // If no results are found, display a message
  if ($json->response->numFound == '0'){
    $output = 'no results found';
  }
  else{
    foreach ($json->response->docs as $result){
      $id = $result->id;
      $content = $result->content;
      $result_output = parse_result($id, $content);
      $output[] = $result_output;
     }
  }
  return $output;
}

function solr_search_id($id, $core){

  // URL encode the ID string
  $id = urlencode($id);

  // Assemble a query string to send to Solr. This uses the Solr hostname from config.env. Solr's query syntax can be found at many sites including https://lucene.apache.org/solr/guide/6_6/the-standard-query-parser.html
  $solrurl = 'http://' . $_ENV["SOLR_HOSTNAME"] . ':' . $_ENV["SOLR_PORT"] . '/solr/' . $core . '/select?q.op=OR&q=id%3A"' . $id . '"&wt=json';

  // Perform Curl request on the Solr API
  $ch = curl_init();
  curl_setopt($ch, CURLOPT_URL, $solrurl);
  curl_setopt($ch, CURLOPT_RETURNTRANSFER, TRUE);
  curl_setopt($ch, CURLOPT_HEADER, FALSE);
  curl_setopt($ch, CURLOPT_CUSTOMREQUEST, 'GET');
  $response = curl_exec($ch);
  curl_close($ch);

  // Turn the API response into useful Json
  $json = json_decode($response);

  // If no results are found, display a message
  if ($json->response->numFound == '0'){
    $output = 'no results found';
  }
  else{
    foreach ($json->response->docs as $result){
      $id = $result->id;
      $content = $result->content;
      $result_output = parse_result($id, $content);
      $output[] = $result_output;
    }
  }
  return $output;
}

function parse_result($id, $input){

  $output['id'] = $id;

  //Set document reference number (used for OPS API)
  if (preg_match('/=D\s(([^\s]*)\s([^\s]*)\s([^\s]*))/', $input, $doc_ref)){
    $output['doc_ref'] = str_replace(' ','',$doc_ref[1]);
  }

  // Search for the application ID in the content element and display it
  preg_match('/Application.*\n(.*)\n/', $input, $application_id);
  $output['application_id'] = $application_id[1];

  // Search for the EPO publication URL in the content element and display it
  preg_match('/Publication.*\n(.*)\n/', $input, $epo_publication);
  $output['epo_publication_url'] = $epo_publication[1];

  // Search for the IPC publication URL in the content element and display it
  preg_match('/IPC.*\n(.*)\n/', $input, $ipc_publication);
  $output['ipc_publication_url'] = $ipc_publication[1];

  // Search for the title in the content element and display it
  if (preg_match('/Title.*\n(.*)\n/', $input, $title)){
    $output['title'] = $title[1];
  }

  // Search for the abstract in the content element and display it
  if (preg_match('/Abstract.*\n(.*)\n/', $input, $abstract)){
    $output['abstract'] = $abstract[1];
  }
  elseif (preg_match('/\(.\) \\n\\n(.*)\\n/', $input, $abstract)) {
    $output['abstract'] = $abstract[1];
  }

  // Search for the year in the content element and display it
  if (preg_match('/=D[^\s]*\s[^\s]*\s[^\s]*\s[^\s]*\s(\d{4})/', $input, $year)){
    $output['year'] = $year[1];
  }
  return $output;
}

function get_random_record($core){

    // Generate a random number for sorting by random
    $random = rand();

    // Assemble a query string to send to Solr. This uses the Solr hostname from config.env. Solr's query syntax can be found at many sites including https://lucene.apache.org/solr/guide/6_6/the-standard-query-parser.html
    // This query retrieves only the bib identifier field for records which satisfy the search query
    $solrurl = 'http://' . $_ENV["SOLR_HOSTNAME"] . ':' . $_ENV["SOLR_PORT"] . '/solr/' . $core . '/select?q.op=OR&q=*%3A*&wt=json&sort=random_' . $random . '%20asc';

    // Perform Curl request on the Solr API
    $ch = curl_init();
    curl_setopt($ch, CURLOPT_URL, $solrurl);
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, TRUE);
    curl_setopt($ch, CURLOPT_HEADER, FALSE);
    curl_setopt($ch, CURLOPT_CUSTOMREQUEST, 'GET');
    $response = curl_exec($ch);
    curl_close($ch);

    // Turn the API response into useful Json
    $json = json_decode($response);

    // Pick a random key out of the docs array
    $random = array_rand($json->response->docs);

    //Set ID variable
    $id = $json->response->docs[$random]->id;

    //Set content variable
    $content = $json->response->docs[$random]->content;

    //Construct associative array with ID and content
    $result_array = array($id=>$content);

    return $result_array;
}

function one_random_record ($core){

  $random = get_random_record($core);

  foreach ($random as $id => $content){
    $output = parse_result($id, $content);
  }
  return $output;
}

function ten_random_titles ($core){

  $x = 0;
  while ($x <= 9) {
    $random = get_random_record($core);

    foreach($random as $id => $content){
      // Search for the title in the content element and display it
      if (preg_match('/Title.*\n(.*)\n/', $content, $title)){
        $output[$x] = array($id=>$title[1]);
        ++$x;
      }
    }
  }
  return $output;
}

function ten_random_abstracts ($core){

  $x = 0;
  while ($x <= 9) {
    $random = get_random_record($core);

    foreach($random as $id => $content){
      // Search for the abstract in the content element and display it
      if (preg_match('/Abstract.*\n(.*)\n/', $content, $abstract)){
        $output[$x] = array($id=>$abstract[1]);
        ++$x;
      }
      elseif (preg_match('/\(.\) \\n\\n(.*)\\n/', $content, $abstract)) {
        $output[$x] = array($id=>$abstract[1]);
        ++$x;
      }
    }
  }
  return $output;
}

function ten_random_doc_refs ($core){

  $x = 0;
  while ($x <= 9) {
    $random = get_random_record($core);

    foreach($random as $id => $content){
      //Set document reference number (used for OPS API)
      preg_match('/=D\s(([^\s]*)\s([^\s]*)\s([^\s]*))/', $content, $doc_ref);
      $doc_ref = str_replace(' ','',$doc_ref[1]);
      if (check_for_images($doc_ref)){
        $output[$x] = $doc_ref;
        ++$x;
      }
    }
  }
  return $output;
}

?>