You are here

Home » API reference » Bibliography Module 7

biblio_pm.module in Bibliography Module 7

Same filename and directory in other branches
  1. 6.2 modules/pubmed/biblio_pm.module
  2. 7.2 modules/pubmed/biblio_pm.module

File

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

/**
 * @file
 */

/**
 *
 */
function biblio_pm_cron() {
  if (variable_get('biblio_pm_auto_update', 0)) {
    module_load_include('inc', 'biblio', 'includes/biblio.import.export');

    // Defaults to once per hour.
    $interval = variable_get('biblio_pm_update_interval', 3600);

    // Only update 100 at a time.
    $count_limit = variable_get('biblio_pm_update_limit', 100);

    // Defaults to one month since last update.
    $age = variable_get('biblio_pm_age_limit', 2419200);
    $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(
        'nid',
        'biblio_pubmed_id',
      ))
        ->condition('biblio_pm_changed', $age_limit, '<')
        ->orderBy('nid', 'ASC')
        ->range(0, $count_limit)
        ->execute();
      foreach ($result as $pm) {
        $ids[$pm->nid] = $pm->biblio_pubmed_id;
      }
      if (count($ids)) {
        list($nids, $dups) = biblio_pm_import_ids($ids);
        if (count($nids)) {
          foreach ($nids as $nid) {
            $message = '';
            $message = t('!nid was updated due to changes originating at !url', array(
              '!nid' => l($nid, 'node/' . $nid),
              '!url' => l(t('PubMed'), 'https://www.ncbi.nlm.nih.gov/pubmed/' . $ids[$nid]),
            ));
            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('nid', $dups, 'IN')
            ->execute();
        }
      }
      else {
        $message = t('There were no PubMed entries older than @age to check.', array(
          '@age' => format_interval($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_node_form_alter(&$form, &$form_state, $form_id) {
  if ($form_id == 'biblio_node_form' && isset($form_state['biblio_fields'])) {
    if (isset($form_state['values']['biblio_pubmed_id'])) {
      $default_pubmed_id = $form_state['values']['biblio_pubmed_id'];
    }
    elseif (isset($form_state['node']->biblio_pubmed_id)) {
      $default_pubmed_id = $form_state['node']->biblio_pubmed_id;
    }
    else {
      $default_pubmed_id = '';
    }
    if (isset($form_state['values']['biblio_pmcid'])) {
      $default_pmcid = $form_state['values']['biblio_pmcid'];
    }
    elseif (isset($form_state['node']->biblio_pmcid)) {
      $default_pmcid = $form_state['node']->biblio_pmcid;
    }
    else {
      $default_pmcid = '';
    }
    $form['biblio_tabs'][4]['biblio_pubmed_id'] = array(
      '#type' => 'textfield',
      '#title' => t('PMID'),
      '#required' => FALSE,
      '#description' => t('PubMed ID'),
      '#default_value' => $default_pubmed_id,
      '#size' => 50,
      '#maxlength' => 50,
    );
    $form['biblio_tabs'][4]['biblio_pmcid'] = array(
      '#type' => 'textfield',
      '#title' => t('PMCID'),
      '#required' => FALSE,
      '#description' => t('PubMed Central ID'),
      '#default_value' => $default_pmcid,
      '#size' => 50,
      '#maxlength' => 50,
    );
  }
  if ((!isset($form_state['biblio_type']) || empty($form_state['biblio_type'])) && !isset($form_state['node']->nid)) {
    $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'),
      '#submit' => array(
        'biblio_pm_form_biblio_node_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' => $form_state['values']['biblio_pubmed_md5'],
    );
  }
  if (isset($form_state['values']['biblio_pmcid'])) {
    $form['biblio_pmcid'] = array(
      '#type' => 'value',
      '#value' => $form_state['values']['biblio_pmcid'],
    );
  }
  if (isset($form_state['values']['biblio_pubmed_grants'])) {
    $form['biblio_pubmed_grants'] = array(
      '#type' => 'value',
      '#value' => $form_state['values']['biblio_pubmed_grants'],
    );
  }
}

/**
 *
 */
function biblio_pm_form_biblio_node_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');
      $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();
    }
    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['input']['biblio_pmcid'] = $form_state['biblio_pmcid'] = isset($node_data['biblio_pmcid']) ? $node_data['biblio_pmcid'] : '';
    $form_state['input']['biblio_pubmed_id'] = $form_state['biblio_pubmed_id'] = $node_data['biblio_pubmed_id'];
    $form_state['input']['biblio_pubmed_grants'] = $form_state['biblio_pubmed_grants'] = isset($node_data['biblio_pmcid']) ? $node_data['biblio_pmcid'] : array();
    $form_state['rebuild'] = TRUE;
  }
}

/**
 *
 */
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) {
  $nids = 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);
}

/**
 * Imports article(s) from an XML file (exported) from PubMed Central.
 *
 * @param file object $file
 * @param array $terms
 * @param bool $batch
 * @param string $session_id
 *
 * @return array
 */
function biblio_pm_xml_biblio_import($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(t("Could not parse file as PubMed XML"), 'error');
    return;
  }
  return _biblio_pm_create_node_from_xml($xml, $terms, $batch, $session_id);
}

/**
 * Imports multiple PMIDs.
 *
 * @param array $pmids
 * @param array $terms
 * @param bool $batch
 * @param string $session_id
 *
 * @return array
 */
function biblio_pm_import_ids($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);
  $nids = array();
  $dups = array();
  while ($start < $resmax && ($result = $Eclient
    ->fetchRecords($start))) {
    $start += count($result
      ->xpath('//PubmedArticle'));
    list($nid, $dup) = _biblio_pm_create_node_from_xml($result, $terms, $batch, $session_id);
    $nids = array_merge($nids, $nid);
    $dups = array_merge($dups, $dup);
  }
  return array(
    $nids,
    $dups,
  );
}

/**
 * Fetches the article for the given PMID and returns a biblio node object.
 *
 * Returns an empty node object if the PMID is not found.
 *
 * @param int $pmid
 *
 * @return stdClass $node
 */
function biblio_pm_fetch_pmid($pmid) {
  $node = new stdClass();
  $Eclient = new BiblioEntrezClient();
  $paser = new BiblioEntrezPubmedArticle();
  try {
    $xml = $Eclient
      ->fetch($pmid);
  } catch (Exception $e) {
    return $node;
  }
  $articles = $xml
    ->xpath('//PubmedArticle');
  if (count($articles)) {
    $node = $paser
      ->setArticle($articles[0])
      ->getBiblioAsObject();
    if (!empty($node)) {
      $node->type = 'biblio';
      node_object_prepare($node);
    }
  }
  return $node;
}

/**
 *
 */
function _biblio_pm_create_node_from_xml($xml, $terms, $batch, $session_id) {
  module_load_include('php', 'biblio_pm', 'EntrezPubmedArticle');
  $nids = array();
  $dups = array();
  $node = new stdClass();
  $data = new BiblioEntrezPubmedArticle();
  foreach ($xml
    ->xpath('//PubmedArticle') as $article) {
    $node = $data
      ->setArticle($article)
      ->getBiblioAsObject();
    if (isset($node)) {
      $dup = biblio_pm_check_md5($node->biblio_pubmed_id, $node->biblio_pubmed_md5);
      $action = variable_get('biblio_pm_dup_action', 'newrev');

      // Entry has be imported before, but may have changed.
      if ($dup < 0 && $action == 'newrev') {

        // Load the node in order to preserve all its data and merge the new
        // data from pubmed.
        $node = (object) array_merge((array) node_load(-$dup), (array) $node);
        $node->nid = -$dup;
        $node->revision = 1;
        $curr_date = format_date(time());
        $node->log = t('Automatically updated by the Biblio PubMed module on !date due to changes at !url', array(
          '!date' => $curr_date,
          '!url' => l('PubMed', 'https://www.ncbi.nlm.nih.gov/pubmed/' . $node->biblio_pubmed_id),
        ));
        $dup = NULL;
      }

      // Entry has be imported before, but may have changed.
      if ($dup < 0 && $action == 'replace') {
        $node->nid = -$dup;
        $existing_node = db_query("SELECT * FROM {node} WHERE nid=:nid", array(
          ':nid' => $node->nid,
        ))
          ->fetchObject();
        $node = (object) array_merge((array) $existing_node, (array) $node);
        $dup = NULL;
      }
      if (!$dup) {

        // Allows other modules to alter the node before it is being saved. (Note: $article is a SimpleXML object)
        drupal_alter('biblio_pm_node', $node, $article);
        biblio_save_node($node, $terms, $batch, $session_id);
        if (!empty($node->nid)) {
          $nids[] = $node->nid;
        }
      }
      else {
        $dups[] = $dup;
      }
      $node = NULL;
    }
  }
  return array(
    $nids,
    $dups,
  );
}

/**
 *
 */
function biblio_pm_check_pmid($pmid) {
  return db_query("SELECT nid 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 = 'https://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_node_view($node, $view_mode, $langcode) {
  if ($node->type == 'biblio' && isset($node->biblio_pubmed_id)) {
    switch ($view_mode) {
      case 'full':
      case 'teaser':
        $node->content['links']['biblio_pubmed'] = array(
          '#links' => biblio_pm_biblio_lookup_link($node),
          '#attributes' => array(
            'class' => array(
              'links',
              'inline',
            ),
          ),
        );
    }
  }
}

/**
 *
 */
function biblio_pm_node_delete($node) {
  if ($node->type != 'biblio') {
    return;
  }
  db_delete('biblio_pubmed')
    ->condition('nid', $node->nid)
    ->execute();
  db_delete('biblio_pubmed_grant_info')
    ->condition('nid', $node->nid)
    ->execute();
}

/**
 *
 */
function biblio_pm_node_insert($node) {
  if (isset($node->biblio_pubmed_id) && !empty($node->biblio_pubmed_id)) {
    $node->biblio_pm_changed = time();
    drupal_write_record('biblio_pubmed', $node);
  }
  if (isset($node->biblio_pubmed_grants) && is_array($node->biblio_pubmed_grants)) {
    foreach ($node->biblio_pubmed_grants as $grant) {
      $info = array(
        'nid' => $node->nid,
        'biblio_pubmed_id' => $node->biblio_pubmed_id,
      );
      $info += $grant;
      drupal_write_record('biblio_pubmed_grant_info', $info);
    }
  }
}

/**
 *
 */
function biblio_pm_node_update($node) {
  if (isset($node->biblio_pubmed_id) && !empty($node->biblio_pubmed_id)) {

    // Make sure stale entries are deleted.
    db_delete('biblio_pubmed')
      ->condition('nid', $node->nid)
      ->execute();

    // Write the new data.
    $node->biblio_pm_changed = time();
    drupal_write_record('biblio_pubmed', $node);
  }
  if (isset($node->biblio_pubmed_grants) && is_array($node->biblio_pubmed_grants) && !empty($node->biblio_pubmed_grants)) {
    db_delete('biblio_pubmed_grant_info')
      ->condition('nid', $node->nid)
      ->execute();
    foreach ($node->biblio_pubmed_grants as $grant) {
      $info = array(
        'nid' => $node->nid,
        'biblio_pubmed_id' => $node->biblio_pubmed_id,
      );
      $info += $grant;
      drupal_write_record('biblio_pubmed_grant_info', $info);
    }
  }
}

/**
 *
 */
function biblio_pm_node_load($nodes, $types) {
  $result = db_select('biblio_pubmed', 'bpm')
    ->fields('bpm', array(
    'nid',
    'biblio_pubmed_id',
    'biblio_pmcid',
    'biblio_pubmed_md5',
  ))
    ->condition('nid', array_keys($nodes))
    ->execute();

  // $result = db_query('SELECT  biblio_pubmed_id  FROM {biblio_pubmed} WHERE nid IN(:nids)', array(':nids' => array_keys($nodes)));
  foreach ($result as $record) {
    $nodes[$record->nid]->biblio_pubmed_id = $record->biblio_pubmed_id;
    $nodes[$record->nid]->biblio_pmcid = $record->biblio_pmcid;
    $nodes[$record->nid]->biblio_pubmed_md5 = $record->biblio_pubmed_md5;
  }
  $result = db_select('biblio_pubmed_grant_info', 'bpmgi')
    ->fields('bpmgi')
    ->condition('nid', array_keys($nodes))
    ->execute();
  foreach ($result as $record) {
    $nodes[$record->nid]->biblio_pubmed_grants[] = array(
      'grantid' => $record->grantid,
      'acronym' => $record->acronym,
      'agency' => $record->agency,
      'country' => $record->country,
    );
  }
}

/**
 *
 */
function biblio_pm_check_md5($pmid, $md5) {
  static $pm_md5s = array();
  static $pm_nids = array();
  if (empty($pm_md5s)) {
    $result = db_query("SELECT * FROM {biblio_pubmed} ");
    foreach ($result as $row) {
      $pm_md5s[$row->biblio_pubmed_md5] = $row->nid;
      $pm_nids[$row->biblio_pubmed_id] = $row->nid;
    }
  }

  // Must be an exact duplicate of an existing node (return the nid)
  if (isset($pm_nids[$pmid]) && isset($pm_md5s[$md5])) {
    return $pm_md5s[$md5];
  }
  elseif (isset($pm_nids[$pmid]) && !isset($pm_md5s[$md5])) {
    return -$pm_nids[$pmid];
  }
  else {

    // Gaurd against duplicates in the same import.
    $pm_md5s[$md5] = TRUE;
    $pm_nids[$pmid] = TRUE;
    return;
  }
}

/**
 *
 */
function biblio_pm_views_api() {
  return array(
    'api' => 2,
  );
}

/**
 *
 */
function biblio_pm_biblio_node_table_rows_alter(&$rows, $node) {
  if (isset($node->biblio_pubmed_id) && !empty($node->biblio_pubmed_id)) {
    $rows[] = array(
      array(
        'data' => t('PubMed ID'),
        'class' => array(
          'biblio-row-title',
        ),
      ),
      array(
        'data' => l($node->biblio_pubmed_id, 'https://www.ncbi.nlm.nih.gov/pubmed/' . $node->biblio_pubmed_id . '?dopt=Abstract'),
      ),
    );
  }
  if (isset($node->biblio_pmcid) && !empty($node->biblio_pmcid)) {
    $rows[] = array(
      array(
        'data' => t('PubMed Central ID'),
        'class' => array(
          'biblio-row-title',
        ),
      ),
      array(
        'data' => check_plain($node->biblio_pmcid),
      ),
    );
  }
  if (isset($node->biblio_pubmed_grants) && is_array($node->biblio_pubmed_grants)) {
    foreach ($node->biblio_pubmed_grants as $grant) {
      $list[] = check_plain(implode(' / ', $grant));
    }
    $rows[] = array(
      array(
        'data' => t('Grant List'),
        'class' => array(
          'biblio-row-title',
        ),
      ),
      array(
        'data' => implode('<br>', $list),
      ),
    );
  }
}

Functions

Namesort descending Description
biblio_pm_biblio_import
biblio_pm_biblio_import_options
biblio_pm_biblio_lookup_link
biblio_pm_biblio_lookup_link_settings
biblio_pm_biblio_node_table_rows_alter
biblio_pm_check_md5
biblio_pm_check_pmid
biblio_pm_cron
biblio_pm_fetch_pmid Fetches the article for the given PMID and returns a biblio node object.
biblio_pm_form_biblio_admin_settings_alter
biblio_pm_form_biblio_node_form_alter
biblio_pm_form_biblio_node_form_submit
biblio_pm_import_ids Imports multiple PMIDs.
biblio_pm_node_delete
biblio_pm_node_insert
biblio_pm_node_load
biblio_pm_node_update
biblio_pm_node_view
biblio_pm_views_api
biblio_pm_xml_biblio_import Imports article(s) from an XML file (exported) from PubMed Central.
_biblio_pm_create_node_from_xml