You are here

biblio_pm.module in Bibliography Module 7.2

Same filename and directory in other branches
  1. 6.2 modules/pubmed/biblio_pm.module
  2. 7 modules/pubmed/biblio_pm.module

File

modules/pubmed/biblio_pm.module
View source
<?php

/*
 * @file pubmed.module
 *
 */
function biblio_pm_cron() {
  if (variable_get('biblio_pm_auto_update', 0)) {
    $interval = variable_get('biblio_pm_update_interval', 3600);

    // defaults to once per hour
    $count_limit = variable_get('biblio_pm_update_limit', 100);

    // only update 100 at a time
    $age = variable_get('biblio_pm_age_limit', 2419200);

    // defaults to one month since last update
    $age_limit = time() - $age;
    if (time() >= variable_get('biblio_pm_update_next_execution', 0)) {
      $ids = array();
      $result = db_select('biblio_pubmed', 'bpm')
        ->fields('bpm', array(
        'bid',
        'biblio_pubmed_id',
      ))
        ->condition('biblio_pm_changed', $age_limit, '<')
        ->orderBy('bid', 'ASC')
        ->range(0, $count_limit)
        ->execute();
      foreach ($result as $pm) {
        $ids[$pm->bid] = $pm->biblio_pubmed_id;
      }
      if (count($ids)) {
        list($bids, $dups) = biblio_pm_import_ids($ids);
        if (count($bids)) {
          foreach ($bids as $bid) {
            $message = '';
            $message = t('!bid was updated due to changes originating at !url', array(
              '!bid' => l($bid, 'biblio/' . $bid),
              '!url' => l('PubMed', 'http://www.ncbi.nlm.nih.gov/pubmed/' . $ids[$bid]),
            ));
            watchdog('biblio_pm', $message, array(), WATCHDOG_WARNING);
          }
        }
        if (count($dups)) {
          $count = count($dups);
          $message = format_plural($count, 'One duplicate PubMed entry was checked, but no changes were found.', '@count PubMed entries were checked, but no changes were found.');
          watchdog('biblio_pm', $message, array(
            '@count' => $count,
          ), WATCHDOG_INFO);
          $now = time();
          db_update('biblio_pubmed')
            ->fields(array(
            'biblio_pm_changed' => $now,
          ))
            ->condition('bid', $dups, 'IN')
            ->execute();
        }
      }
      $message = t('There were no PubMed entries older than @age to check.', array(
        '@age' => format_interval($age),
      ));
      watchdog('biblio_pm', $message, array(), WATCHDOG_INFO);
      variable_set('biblio_pm_update_next_execution', time() + $interval);
    }
  }
}
function biblio_pm_form_biblio_admin_settings_alter(&$form, &$form_state) {
  module_load_include('inc', 'biblio_pm', 'biblio_pm.admin');
  $form += biblio_pm_settings_form();
}
function biblio_pm_form_biblio_form_alter(&$form, &$form_state, $form_id) {
  if ((!isset($form_state['biblio_type']) || empty($form_state['biblio_type'])) && !isset($form_state['biblio']->bid)) {
    $form['biblio_pubmed_lookup'] = array(
      '#type' => 'fieldset',
      '#title' => t('PubMed Lookup'),
      '#weight' => -20,
      '#collapsible' => TRUE,
      '#collapsed' => TRUE,
    );
    $form['biblio_pubmed_lookup']['PMID'] = array(
      '#type' => 'textfield',
      '#title' => t('PubMed ID'),
      '#required' => FALSE,
      '#default_value' => '',
      '#description' => t('Enter a PubMed ID</b>'),
      '#size' => 60,
      '#maxlength' => 255,
      '#weight' => -4,
    );
    $form['biblio_pubmed_lookup']['pubmed_submit'] = array(
      '#type' => 'submit',
      '#value' => t('Populate using PubMed'),
      '#validate' => array(
        'biblio_pm_form_biblio_form_validate',
      ),
      '#submit' => array(
        'biblio_pm_form_biblio_form_submit',
      ),
    );

    // $form['#submit'] = array_merge(array('biblio_pm_form_biblio_node_form_submit'), isset($form['#submit'])?$form['#submit']:array()); // put my validator first
  }
  if (isset($form_state['values']['biblio_pubmed_id'])) {
    $form['biblio_pubmed_id'] = array(
      '#type' => 'value',
      '#value' => $form_state['values']['biblio_pubmed_id'],
    );
  }
  if (isset($form_state['values']['biblio_pubmed_md5'])) {
    $form['biblio_pubmed_md5'] = array(
      '#type' => 'value',
      '#value' => $form_state['values']['biblio_pubmed_md5'],
    );
  }
}
function biblio_pm_form_biblio_form_validate($form, &$form_state) {
  $pmid = $form_state['values']['PMID'];

  // PubMed sometimes gives us spaces in the PMID. Take 'em out.
  if (isset($pmid) && !empty($pmid)) {
    $form_state['values']['PMID'] = str_replace(' ', '', $pmid);
  }
}
function biblio_pm_form_biblio_form_submit($form, &$form_state) {
  $node_data = array();
  if (strlen($pmid = $form_state['values']['PMID'])) {
    if (!($dup = biblio_pm_check_pmid($pmid))) {
      module_load_include('php', 'biblio_pm', 'EntrezClient');
      module_load_include('php', 'biblio_pm', 'EntrezPubmedArticle');
      module_load_include('inc', 'biblio', 'includes/biblio.import.export');
      $Eclient = new BiblioEntrezClient();
      try {
        $result = $Eclient
          ->fetch($pmid);
      } catch (Exception $e) {
        form_set_error($e
          ->getMessage());
      }
      if (!isset($result->PubmedArticle)) {
        unset($form_state['values']['biblio_type']);
        unset($form_state['post']['biblio_type']);
        form_set_error('PMID', 'No data available for PubMed ID: ' . check_plain($pmid));
        return;
      }
      $data = new BiblioEntrezPubmedArticle($result->PubmedArticle);
      $node_data = $data
        ->getBiblio();
      $biblio = biblio_create($node_data['biblio_type'], $node_data);
      $biblio->biblio_keywords = $node_data['biblio_keywords'];
    }
    else {
      $message = t('The PubMed ID that you are trying to import already exists in the database, see !url', array(
        '!url' => l('node/' . $dup, 'node/' . $dup),
      ));
      form_set_error('PMID', $message);
      $form_state['rebuild'] = TRUE;
      $form_state['submitted'] = FALSE;
      unset($form_state['values']['biblio_type']);
    }
  }
  if (!empty($node_data)) {
    $form_state['values'] = array_merge($form_state['values'], $node_data);
    $form_state['input']['biblio_type'] = $form_state['biblio_type'] = $node_data['biblio_type'];
    $form_state['biblio_imported'] = $biblio;
    $form_state['biblio_imported_contributors'] = $node_data['biblio_contributors'];
    $form_state['rebuild'] = TRUE;
  }
  return;
}
function biblio_pm_biblio_import_options() {
  return array(
    'biblio_pm' => t('PubMed ID List'),
    'biblio_pm_xml' => t('PubMed XML'),
  );
}
function biblio_pm_biblio_import($file, $terms = array(), $batch = FALSE, $session_id = NULL, $save = TRUE, $string = FALSE) {
  $bids = array();
  $dups = array();
  $pmids = file($file->uri, FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);
  if (empty($pmids)) {
    drupal_set_message(t("Could not open PubMed ID file"), 'error');
    return;
  }
  return biblio_pm_import_ids($pmids, $terms, $batch, $session_id);
}
function biblio_pm_xml_biblio_import($file, $terms = array(), $batch = FALSE, $session_id = NULL) {
  libxml_use_internal_errors(true);
  $xml = @simplexml_load_file($file->uri);
  if (empty($xml) || isset($xml->body->pre->ERROR)) {
    drupal_set_message("Could not parse file as PubMed XML", 'error');
    return;
  }
  return _biblio_pm_create_node_from_xml($xml, $terms, $batch, $session_id);
}
function biblio_pm_import_ids($pmids, $terms = array(), $batch = FALSE, $session_id = NULL) {
  module_load_include('php', 'biblio_pm', 'EntrezClient');
  $retmax = 100;
  $resmax = count($pmids);
  $start = 0;
  $Eclient = new BiblioEntrezClient();
  $Eclient
    ->post($pmids);
  $Eclient
    ->setReturnMax($retmax);
  $bids = array();
  $dups = array();
  while ($start < $resmax && ($result = $Eclient
    ->fetchRecords($start))) {
    $start += count($result
      ->xpath('//PubmedArticle'));
    list($bid, $dup) = _biblio_pm_create_node_from_xml($result, $terms, $batch, $session_id);
    $bids = array_merge($bids, $bid);
    $dups = array_merge($dups, $dup);
  }
  return array(
    $nids,
    $dups,
  );
}
function _biblio_pm_create_node_from_xml($xml, $terms, $batch, $session_id) {
  module_load_include('php', 'biblio_pm', 'EntrezPubmedArticle');
  $bids = array();
  $dups = array();
  $data = new BiblioEntrezPubmedArticle();
  foreach ($xml
    ->xpath('//PubmedArticle') as $article) {
    $raw_biblio = $data
      ->setArticle($article)
      ->getBiblioAsObject();
    $biblio = biblio_create($raw_biblio->biblio_type, (array) $raw_biblio);
    $biblio->biblio_keywords = $raw_biblio->biblio_keywords;
    $wrapper = biblio_wrapper($biblio);
    if (isset($biblio)) {
      $dup = biblio_pm_check_md5($biblio->biblio_pubmed_id, $biblio->biblio_pubmed_md5);
      $action = variable_get('biblio_pm_dup_action', 'replace');
      if ($dup < 0 && $action == 'newrev') {

        //entry has be imported before, but may have changed

        // Load the node in order to preserve all its data and merge the new
        // data from pubmed.
        // Entity API doesn't support revisions at time of current Biblio 7.x-2.x development
        // $node = (object) array_merge((array)node_load(-$dup), (array)$node);
        // $node->nid = -$dup;
        // $node->revision = 1;
        // $curr_date = format_date(time());
        // $node->log = t("Automatically updated on !date due to changes at source", array('!date' => $curr_date));
        // $dup = NULL;
      }
      if ($dup < 0 && $action == 'replace') {

        //entry has be imported before, but may have changed
        $biblio->bid = -$dup;
        $dup = NULL;
      }
      if (!$dup) {

        // Allows other modules to alter the biblio before it is being saved. (Note: $article is a SimpleXML object)
        drupal_alter('biblio_pm_biblio', $biblio, $article);
        biblio_save_node($biblio, $terms, $batch, $session_id);
        if (!empty($biblio->bid)) {
          $bids[] = $biblio->bid;
        }
      }
      else {
        $dups[] = $dup;
      }
      $biblio = NULL;
    }
  }
  return array(
    $bids,
    $dups,
  );
}
function biblio_pm_check_pmid($pmid) {
  return db_query("SELECT bid FROM {biblio_pubmed} WHERE biblio_pubmed_id = :pmid", array(
    ':pmid' => $pmid,
  ))
    ->fetchField();
}
function biblio_pm_biblio_lookup_link_settings() {
  return array(
    'pubmed' => t('PubMed'),
  );
}
function biblio_pm_biblio_lookup_link($node) {
  $show_link = variable_get('biblio_lookup_links', array(
    'pubmed' => TRUE,
  ));
  if (!isset($show_link['pubmed']) || !$show_link['pubmed'] || !isset($node) || $node->type != 'biblio' || !isset($node->biblio_pubmed_id)) {
    return array();
  }
  $link = 'http://www.ncbi.nlm.nih.gov/pubmed/' . $node->biblio_pubmed_id . '?dopt=Abstract';
  $attrs = array(
    'title' => t("Click to view the PubMed listing for this node"),
  );
  if (variable_get('biblio_links_target_new_window', null)) {
    $attrs = array_merge($attrs, array(
      'target' => '_blank',
    ));
  }
  return array(
    'biblio_pubmed' => array(
      'title' => t('PubMed'),
      'href' => $link,
      'attributes' => $attrs,
    ),
  );
}
function biblio_pm_entity_view($entity, $type, $view_mode, $langcode) {
  if ($type == 'biblio' && isset($biblio->biblio_pubmed_id)) {
    switch ($view_mode) {
      case 'full':
      case 'teaser':
        $biblio->content['links']['biblio_pubmed'] = array(
          '#links' => biblio_pm_biblio_lookup_link($biblio),
          '#attributes' => array(
            'class' => array(
              'links',
              'inline',
            ),
          ),
        );
    }
  }
}
function biblio_pm_entity_delete($entity, $type) {
  if ($type != 'biblio') {
    return;
  }
  db_delete('biblio_pubmed')
    ->condition('bid', $entity->bid)
    ->execute();
}
function biblio_pm_entity_insert($entity, $type) {
  if ($type != 'biblio') {
    return;
  }
  if (isset($entity->biblio_pubmed_id) && !empty($entity->biblio_pubmed_id)) {
    $entity->biblio_pm_changed = time();
    drupal_write_record('biblio_pubmed', $entity);
  }
}
function biblio_pm_entity_update($entity, $type) {
  if (isset($entity->biblio_pubmed_id) && !empty($entity->biblio_pubmed_id)) {
    $entity->biblio_pm_changed = time();
    drupal_write_record('biblio_pubmed', $entity, 'bid');
  }
}
function biblio_pm_entity_load($entities, $type) {
  if ($type != 'biblio') {
    return;
  }
  $result = db_select('biblio_pubmed', 'bpm')
    ->fields('bpm', array(
    'bid',
    'biblio_pubmed_id',
  ))
    ->condition('bid', array_keys($entities))
    ->execute();

  // $result = db_query('SELECT  biblio_pubmed_id  FROM {biblio_pubmed} WHERE nid IN(:nids)', array(':nids' => array_keys($nodes)));
  foreach ($result as $record) {
    $entities[$record->bid]->biblio_pubmed_id = $record->biblio_pubmed_id;
  }
}
function biblio_pm_check_md5($pmid, $md5) {
  static $pm_md5s = array();
  static $pm_bids = array();
  if (empty($pm_md5s)) {
    $result = db_query("SELECT * FROM {biblio_pubmed} ");
    foreach ($result as $row) {
      $pm_md5s[$row->biblio_pubmed_md5] = $row->bid;
      $pm_bids[$row->biblio_pubmed_id] = $row->bid;
    }
  }
  if (isset($pm_bids[$pmid]) && isset($pm_md5s[$md5])) {

    // must be an exact duplicate of an existing node (return the bid)
    return $pm_md5s[$md5];
  }
  elseif (isset($pm_bids[$pmid]) && !isset($pm_md5s[$md5])) {

    //pmid has been save previously but content must have changed (return negative bid)
    return -$pm_bids[$pmid];
  }
  else {
    $pm_md5s[$md5] = TRUE;

    // gaurd against duplicates in the same import
    $pm_bids[$pmid] = TRUE;
    return;
  }
}
function biblio_pm_add_fields() {
  $field = array(
    'field_name' => 'biblio_pmid',
    'type' => 'number_integer',
  );
  if (!field_info_field($field['field_name'])) {
    field_create_field($field);
  }
}
function biblio_pm_add_field_instances() {

  // pubtypes with existing fields
  foreach (biblio_get_fielded_pubtypes() as $pubtype) {
    $instance = biblio_pm_default_instance_info($pubtype);
    field_create_instance($instance);
  }
}
function biblio_pm_default_instance_info($pubtype) {
  $instance = array(
    'field_name' => 'biblio_pmid',
    'entity_type' => 'biblio',
    'bundle' => $pubtype,
    'label' => 'PubMed ID',
    'settings' => array(
      'vtab' => 'identifiers',
    ),
    'widget' => array(
      'settings' => array(
        'size' => 10,
      ),
    ),
    'display' => array(
      'default' => array(
        'settings' => array(
          'thousand_separator' => '',
        ),
      ),
    ),
  );
  return $instance;
}

/**
 * Implements hook_field_create_instance().
 */
function biblio_pm_field_create_instance($instance) {
  if ($instance['entity_type'] != 'biblio') {
    return;
  }
  $pubtype = $instance['bundle'];
  $field_name = 'biblio_pmid';
  $instance_info = field_info_instance('biblio', $field_name, $pubtype);

  // If biblio_pmid field exists, but no instances exist for current pubtype
  if (field_info_field($field_name) && !$instance_info) {
    $instance = biblio_pm_default_instance_info($pubtype);
    field_create_instance($instance);
  }
}