tripal/tripal_chado/api/modules/tripal_chado.pub.api.inc

<?php
/**
 * @file
 * Provides an application programming interface (API) to manage chado publications
 */

/**
 * @defgroup tripal_pub_api Publication Module API
 * @ingroup tripal_api
 * @{
 * Provides an application programming interface (API) to manage chado publications
 *
 * @stephen add documentation here for how to add a new importer.
 *
 * @}
 */


/**
 * Retrieves a chado publication array
 *
 * @param $identifier
 *   An array used to uniquely identify a publication. This array has the same
 *   format as that used by the chado_generate_var(). The following keys can be
 *   useful for uniquely identifying a publication as they should be unique:
 *    - pub_id: the chado pub.pub_id primary key
 *    - nid: the drupal nid of the publication
 *    - uniquename: A value to matach with the pub.uniquename field
 *   There are also some specially handled keys. They are:
 *    - property: An array describing the property to select records for. It
 *      should at least have either a 'type_name' key (if unique across cvs) or
 *      'type_id' key. Other supported keys include: 'cv_id', 'cv_name' (of the type),
 *      'value' and 'rank'
 *    - dbxref: The database cross reference accession.  It should be in the form
 *        DB:ACCESSION, where DB is the database name and ACCESSION is the
 *        unique publication identifier (e.g. PMID:4382934)
 *    - dbxref_id:  The dbxref.dbxref_id of the publication.
 * @param $options
 *   An array of options. Supported keys include:
 *     - Any keys supported by chado_generate_var(). See that function definition for
 *       additional details.
 *
 * NOTE: the $identifier parameter can really be any array similar to $values passed into
 *   chado_select_record(). It should fully specify the pub record to be returned.
 *
 * @return
 *   If a singe publication is retreived using the identifiers, then a publication
 *   array will be returned.  The array is of the same format returned by the
 *   chado_generate_var() function. Otherwise, FALSE will be returned.
 *
 * @ingroup tripal_pub_api
 */
function tripal_get_publication($identifiers, $options = array()) {

  // Error Checking of parameters
  if (!is_array($identifiers)) {
    tripal_report_error('tripal_pub_api', TRIPAL_ERROR,
      "chado_get_publication: The identifier passed in is expected to be an array with the key
       matching a column name in the pub table (ie: pub_id or name). You passed in %identifier.",
      array('%identifier'=> print_r($identifiers, TRUE))
    );
  }
  elseif (empty($identifiers)) {
    tripal_report_error('tripal_pub_api', TRIPAL_ERROR,
      "chado_get_publication: You did not pass in anything to identify the publication you want. The identifier
       is expected to be an array with the key matching a column name in the pub table
       (ie: pub_id or name). You passed in %identifier.",
      array('%identifier'=> print_r($identifiers, TRUE))
    );
  }

  // If one of the identifiers is property then use chado_get_record_with_property()
  if (array_key_exists('property', $identifiers)) {
    $property = $identifiers['property'];
    unset($identifiers['property']);
    $pub = chado_get_record_with_property(
      array('table' => 'pub', 'base_records' => $identifiers),
      array('type_name' => $property),
      $options
    );
  }
  elseif (array_key_exists('dbxref', $identifiers)) {
    if(preg_match('/^(.*?):(.*?)$/', $identifiers['dbxref'], $matches)) {
      $dbname = $matches[1];
      $accession = $matches[2];

      // First make sure the dbxref is present.
      $values = array(
        'accession' => $accession,
        'db_id' => array(
          'name' => $dbname
        ),
      );
      $dbxref = chado_select_record('dbxref', array('dbxref_id'), $values);
      if (count($dbxref) == 0) {
        return FALSE;
      }
      $pub_dbxref = chado_select_record('pub_dbxref', array('pub_id'), array('dbxref_id' => $dbxref[0]->dbxref_id));
      if (count($pub_dbxref) == 0) {
        return FALSE;
      }
      $pub = chado_generate_var('pub', array('pub_id' => $pub_dbxref[0]->pub_id), $options);
    }
    else {
      tripal_report_error('tripal_pub_api', TRIPAL_ERROR,
        "chado_get_publication: The dbxref identifier is not correctly formatted.",
        array('%identifier'=> print_r($identifiers, TRUE))
      );
    }
  }
  elseif (array_key_exists('dbxref_id', $identifiers)) {
    // first get the pub_dbxref record
    $values = array('dbxref_id' => $identifiers['dbxref_id']);
    $pub_dbxref = chado_select_record('pub_dbxref', array('pub_id'), $values);

    // now get the pub
    if (count($pub_dbxref) > 0) {
      $pub = chado_generate_var('pub', array('pub_id' => $pub_dbxref[0]->pub_id), $options);
    }
    else {
      return FALSE;
    }

  }
  // Else we have a simple case and we can just use chado_generate_var to get the pub
  else {
    // Try to get the pub
    $pub = chado_generate_var('pub', $identifiers, $options);
  }

  // Ensure the pub is singular. If it's an array then it is not singular
  if (is_array($pub)) {
    tripal_report_error('tripal_pub_api', TRIPAL_ERROR,
      "chado_get_publication: The identifiers did not find a single unique record. Identifiers passed: %identifier.",
      array('%identifier'=> print_r($identifiers, TRUE))
    );
  }

  // Report an error if $pub is FALSE since then chado_generate_var has failed
  elseif ($pub === FALSE) {
    tripal_report_error('tripal_pub_api', TRIPAL_ERROR,
      "chado_get_publication: Could not find a publication using the identifiers
       provided. Check that the identifiers are correct. Identifiers passed: %identifier.",
      array('%identifier'=> print_r($identifiers, TRUE))
    );
  }

  // Else, as far we know, everything is fine so give them their pub :)
  else {
    return $pub;
  }
}
/**
 * The publication table of Chado only has a unique constraint for the
 * uniquename of the publiation, but in reality a publication can be considered
 * unique by a combination of the title, publication type, published year and
 * series name (e.g. journal name or conference name). The site administrator
 * can configure how publications are determined to be unique.  This function
 * uses the configuration specified by the administrator to look for publications
 * that match the details specified by the $pub_details argument
 * and indicates if one ore more publications match the criteria.
 *
 * @param $pub_details
 *   An associative array with details about the publications. The expected keys
 *   are:
 *     'Title':              The title of the publication
 *     'Year':               The published year of the publication
 *     'Publication Type':   An array of publication types. A publication can have more than one type.
 *     'Series Name':        The series name of the publication
 *     'Journal Name':       An alternative to 'Series Name'
 *     'Conference Name':    An alternative to 'Series Name'
 *     'Citation':           The publication citation (this is the value saved in the pub.uniquename field and must be unique)
 *   If this key is present it will also be checked
 *     'Publication Dbxref': A database cross reference of the form DB:ACCESSION where DB is the name
 *                           of the database and ACCESSION is the unique identifier (e.g PMID:3483139)
 *
 * @return
 *   An array containing the pub_id's of matching publications. Returns an
 *   empty array if no pubs match
 *
 * @ingroup tripal_pub_api
 */
function tripal_publication_exists($pub_details) {

  // first try to find the publication using the accession number if that key exists in the details array
  if (array_key_exists('Publication Dbxref', $pub_details)) {
    $pub = tripal_get_publication(array('dbxref' => $pub_details['Publication Dbxref']));
    if($pub) {
      return array($pub->pub_id);
    }
  }

  // make sure the citation is unique
  if (array_key_exists('Citation', $pub_details)) {
    $pub = tripal_get_publication(array('uniquename' => $pub_details['Citation']));
    if($pub) {
      return array($pub->pub_id);
    }
  }

  // get the publication type (use the first publication type)
  if (array_key_exists('Publication Type', $pub_details)) {
    $type_name = '';
    if(is_array($pub_details['Publication Type'])) {
      $type_name = $pub_details['Publication Type'][0];
    }
    else {
      $type_name = $pub_details['Publication Type'];
    }
    $identifiers = array(
      'name' => $type_name,
      'cv_id' => array(
        'name' => 'tripal_pub',
      ),
    );
    $pub_type = tripal_get_cvterm($identifiers);
  }
  else {
    tripal_report_error('tripal_pub', TRIPAL_ERROR,
      "tripal_publication_exists(): The Publication Type is a " .
      "required property but is missing", array());
    return array();
  }
  if (!$pub_type) {
    tripal_report_error('tripal_pub', TRIPAL_ERROR,
     "tripal_publication_exists(): Cannot find publication type: '%type'",
      array('%type' => $pub_details['Publication Type'][0]));
    return array();
  }

  // get the series name.  The pub.series_name field is only 255 chars so we must truncate to be safe
  $series_name = '';
  if (array_key_exists('Series Name', $pub_details)) {
    $series_name = substr($pub_details['Series Name'], 0, 255);
  }
  if (array_key_exists('Journal Name', $pub_details)) {
    $series_name = substr($pub_details['Journal Name'], 0, 255);
  }
  if (array_key_exists('Conference Name', $pub_details)) {
    $series_name = substr($pub_details['Conference Name'], 0, 255);
  }

  // make sure the publication is unique using the prefereed import duplication check
  $import_dups_check = variable_get('tripal_pub_import_duplicate_check', 'title_year_media');
  $pubs = array();
  switch ($import_dups_check) {
    case 'title_year':
      $identifiers = array(
        'title' => $pub_details['Title'],
        'pyear' => $pub_details['Year']
      );
      $pubs = chado_select_record('pub', array('pub_id'), $identifiers);
      break;
    case 'title_year_type':
      $identifiers = array(
        'title'   => $pub_details['Title'],
        'pyear'   => $pub_details['Year'],
        'type_id' => $pub_type->cvterm_id,
      );
      $pubs = chado_select_record('pub', array('pub_id'), $identifiers);
      break;
    case 'title_year_media':
      $identifiers = array(
        'title'       => $pub_details['Title'],
        'pyear'       => $pub_details['Year'],
        'series_name' => $series_name,
      );
      $pubs = chado_select_record('pub', array('pub_id'), $identifiers);
      break;
  }

  $return = array();
  foreach ($pubs as $pub) {
    $return[] = $pub->pub_id;
  }

  return $return;
}


/**
 * Used for autocomplete in forms for identifying for publications.
 *
 * @param $field
 *   The field in the publication to search on.
 * @param $string
 *   The string to search for
 *
 * @return
 *   A json array of terms that begin with the provided string
 *
 * @ingroup tripal_chado_api
 */
function tripal_autocomplete_pub($string = '') {
  $items = array();
  $sql = "
    SELECT pub_id, title, uniquename
    FROM {pub}
    WHERE lower(title) like lower(:str)
    ORDER by title
    LIMIT 25 OFFSET 0
  ";
  $pubs = chado_query($sql, array(':str' => $string . '%'));
  while ($pub = $pubs->fetchObject()) {
    $items[$pub->uniquename] = $pub->uniquename;
  }

  drupal_json_output($items);
}


/**
 * Imports a singe publication specified by a remote database cross reference.
 *
 * @param $pub_dbxref
 *   The unique database ID for the record to update.  This value must
 *   be of the format DB_NAME:ACCESSION where DB_NAME is the name of the
 *   database (e.g. PMID or AGL) and the ACCESSION is the unique identifier
 *   for the record in the database.
 * @param $do_contact
 *   Set to TRUE if authors should automatically have a contact record added
 *   to Chado.
 * @param $do_update
 *   If set to TRUE then the publication will be updated if it already exists
 *   in the database.
 *
 * @ingroup tripal_pub
 */
function tripal_import_pub_by_dbxref($pub_dbxref, $do_contact = FALSE, $do_update) {
  $num_to_retrieve = 1;
  $pager_id = 0;
  $page = 0;
  $num_pubs = 0;

  print "\nNOTE: Loading of publications is performed using a database transaction. \n" .
      "If the load fails or is terminated prematurely then the entire set of \n" .
      "insertions/updates is rolled back and will not be found in the database\n\n";

  $transaction = db_transaction();
  try {
    if(preg_match('/^(.*?):(.*?)$/', $pub_dbxref, $matches)) {
      $dbname = $matches[1];
      $accession = $matches[2];

      $criteria = array(
        'num_criteria' => 1,
        'remote_db' => $dbname,
        'criteria' => array(
          '1' => array(
            'search_terms' => "$dbname:$accession",
            'scope' => 'id',
            'operation' => '',
            'is_phrase' => 0,
          ),
        ),
      );
      $remote_db = $criteria['remote_db'];
      $results = tripal_get_remote_pubs($remote_db, $criteria, $num_to_retrieve, $page);
      $pubs          = $results['pubs'];
      $search_str    = $results['search_str'];
      $total_records = $results['total_records'];
      $pub_id = tripal_pub_add_publications($pubs, $do_contact, $do_update);
    }

    // For backwards compatibility check to see if the legacy pub module
    // is enabled. If so, then sync the nodes.
    if (module_exists('tripal_pub')) {

      // sync the newly added publications with Drupal
      print "Syncing publications with Drupal...\n";
      chado_node_sync_records('pub');

      // if any of the importers wanted to create contacts from the authors then sync them
      if($do_contact) {
        print "Syncing contacts with Drupal...\n";
        chado_node_sync_records('contact');
      }
    }
  }
  catch (Exception $e) {
    $transaction->rollback();
    print "\n"; // make sure we start errors on new line
    watchdog_exception('T_pub_import', $e);
    print "FAILED: Rolling back database changes...\n";
    return;
  }

  print "Done.\n";
}

/**
 * Imports all publications for all active import setups.
 *
 * @param $report_email
 *   A list of email address, separated by commas, that should be notified
 *   once importing has completed
 * @param $do_update
 *   If set to TRUE then publications that already exist in the Chado database
 *   will be updated, whereas if FALSE only new publications will be added
 *
 * @ingroup tripal_pub
 */
function tripal_execute_active_pub_importers($report_email = FALSE, $do_update = FALSE) {
  $num_to_retrieve = 100;
  $page = 0;

  print "\nNOTE: Loading of publications is performed using a database transaction. \n" .
      "If the load fails or is terminated prematurely then the entire set of \n" .
      "insertions/updates is rolled back and will not be found in the database\n\n";

  // start the transaction
  $transaction = db_transaction();

  try {
    // get all of the loaders
    $args = array();
    $sql = "SELECT * FROM {tripal_pub_import} WHERE disabled = 0 ";
    $results = db_query($sql, $args);
    $do_contact = FALSE;
    $reports = array();
    foreach ($results as $import) {
      $page = 0;
      print "Executing importer: '" . $import->name . "'\n";
      // keep track if any of the importers want to create contacts from authors
      if ($import->do_contact == 1) {
        $do_contact = TRUE;
      }
      $criteria = unserialize($import->criteria);
      $remote_db = $criteria['remote_db'];
      do {
        // retrieve the pubs for this page. We'll retreive 100 at a time
        $results = tripal_get_remote_pubs($remote_db, $criteria, $num_to_retrieve, $page);
        $pubs = $results['pubs'];
        $reports[$import->name] = tripal_pub_add_publications($pubs, $import->do_contact, $do_update);
        $page++;
      }
      // continue looping until we have a $pubs array that does not have
      // our requested numer of records.  This means we've hit the end
      while (count($pubs) == $num_to_retrieve);
    }

    // sync the newly added publications with Drupal. If the user
    // requested a report then we don't want to print any syncing information
    // so pass 'FALSE' to the sync call
    // For backwards compatibility check to see if the legacy pub module
    // is enabled. If so, then sync the nodes.
    if (module_exists('tripal_pub')) {
      print "Syncing publications with Drupal...\n";
      chado_node_sync_records('pub');
    }

    // iterate through each of the reports and generate a final report with HTML links
    $HTML_report = '';
    if ($report_email) {
      $HTML_report .= "<html>";
      global $base_url;
      foreach ($reports as $importer => $report) {
        $total = count($report['inserted']);
        $HTML_report .= "<b>$total new publications from importer: $importer</b><br><ol>\n";
        foreach ($report['inserted'] as $pub) {
          $item = $pub['Title'];
          if (array_key_exists('pub_id', $pub)) {
            $item = l($pub['Title'], "$base_url/pub/" . $pub['pub_id']);
          }
          $HTML_report .= "<li>$item</li>\n";
        }
        $HTML_report .= "</ol>\n";
      }
      $HTML_report .= "</html>";
      $site_email = variable_get('site_mail', '');
      $params = array(
        'message' => $HTML_report
      );
      drupal_mail('tripal_pub', 'import_report', $report_email, language_default(), $params, $site_email, TRUE);
    }

    // For backwards compatibility check to see if the legacy pub module
    // is enabled. If so, then sync the nodes.
    if (module_exists('tripal_pub')) {
      // if any of the importers wanted to create contacts from the authors then sync them
      if($do_contact) {
        print "Syncing contacts with Drupal...\n";
        chado_node_sync_records('contact');
      }
    }
  }
  catch (Exception $e) {
    $transaction->rollback();
    print "\n"; // make sure we start errors on new line
    watchdog_exception('T_pub_import', $e);
    print "FAILED: Rolling back database changes...\n";
    return;
  }
  print "Done.\n";
}

/**
 * Updates publication records.
 *
 * Updates publication records that currently exist in the Chado pub table
 * with the most recent data in the remote database.
 *
 * @param $do_contact
 *   Set to TRUE if authors should automatically have a contact record added
 *   to Chado. Contacts are added using the name provided by the remote
 *   database.
 * @param $dbxref
 *   The unique database ID for the record to update.  This value must
 *   be of the format DB_NAME:ACCESSION where DB_NAME is the name of the
 *   database (e.g. PMID or AGL) and the ACCESSION is the unique identifier
 *   for the record in the database.
 * @param $db
 *   The name of the remote database to update.  If this value is provided and
 *   no dbxref then all of the publications currently in the Chado database
 *   for this remote database will be updated.
 *
 * @ingroup tripal_pub
 */
function tripal_reimport_publications($do_contact = FALSE, $dbxref = NULL, $db = NULL) {

  print "\nNOTE: Loading of publications is performed using a database transaction. \n" .
      "If the load fails or is terminated prematurely then the entire set of \n" .
      "insertions/updates is rolled back and will not be found in the database\n\n";
  $transaction = db_transaction();
  try {

    // get a list of all publications by their Dbxrefs that have supported databases
    $sql = "
      SELECT DB.name as db_name, DBX.accession
      FROM pub P
        INNER JOIN pub_dbxref PDBX ON P.pub_id = PDBX.pub_id
        INNER JOIN dbxref DBX      ON DBX.dbxref_id = PDBX.dbxref_id
        INNER JOIN db DB           ON DB.db_id = DBX.db_id
    ";
    $args = array();
    if ($dbxref and preg_match('/^(.*?):(.*?)$/', $dbxref, $matches)) {
      $dbname = $matches[1];
      $accession = $matches[2];
      $sql .= "WHERE DBX.accession = :accession and DB.name = :dbname ";
      $args[':accession'] = $accession;
      $args[':dbname'] = $dbname;
    }
    elseif ($db) {
      $sql .= " WHERE DB.name = :dbname ";
      $args[':dbname'] = $db;
    }
    $sql .= "ORDER BY DB.name, P.pub_id";
    $results = chado_query($sql, $args);

    $num_to_retrieve = 100;
    $i = 0;                 // count the number of IDs. When we hit $num_to_retrieve we'll do the query
    $curr_db = '';          // keeps track of the current current database
    $ids = array();         // the list of IDs for the database
    $search = array();      // the search array passed to the search function

    // iterate through the pub IDs
    while ($pub = $results->fetchObject()) {
      $accession = $pub->accession;
      $remote_db = $pub->db_name;

      // here we need to only update publications for databases we support
      $supported_dbs = variable_get('tripal_pub_supported_dbs', array());
      if(!in_array($remote_db, $supported_dbs)) {
        continue;
      }
      $search = array(
        'num_criteria' => 1,
        'remote_db' => $remote_db,
        'criteria' => array(
          '1' => array(
            'search_terms' => "$remote_db:$accession",
            'scope' => 'id',
            'operation' => '',
            'is_phrase' => 0,
          ),
        ),
      );
      $pubs = tripal_get_remote_pubs($remote_db, $search, 1, 0);
      tripal_pub_add_publications($pubs, $do_contact, TRUE);

      $i++;
    }

    // For backwards compatibility check to see if the legacy pub module
    // is enabled. If so, then sync the nodes.
    if (module_exists('tripal_pub')) {
      // sync the newly added publications with Drupal
      print "Syncing publications with Drupal...\n";
      chado_node_sync_records('pub');

      // if the caller wants to create contacts then we should sync them
      if ($do_contact) {
        print "Syncing contacts with Drupal...\n";
        chado_node_sync_records('contact');
      }
    }
  }
  catch (Exception $e) {
    $transaction->rollback();
    print "\n"; // make sure we start errors on new line
    watchdog_exception('T_pub_import', $e);
    print "FAILED: Rolling back database changes...\n";
    return;
  }
  print "Done.\n";
}

/**
 * Launch the Tripal job to generate citations.
 *
 * This function will recreate citations for all publications currently
 * loaded into Tripal.  This is useful to create a consistent format for
 * all citations.
 *
 * @param $options
 *  Options pertaining to what publications to generate citations for.
 *  One of the following must be present:
 *   - all: Create and replace citation for all pubs
 *   - new: Create citation for pubs that don't already have one
 *
 * @ingroup tripal_pub
 */
function tripal_pub_create_citations($options) {
  $skip_existing = TRUE;
  $sql = "
    SELECT cvterm_id
    FROM {cvterm}
    WHERE
      name = 'Citation' AND
      cv_id = (SELECT cv_id FROM {cv} WHERE name = 'tripal_pub')
  ";
  $citation_type_id = chado_query($sql)->fetchField();

  // Create and replace citation for all pubs
  if ($options == 'all') {
    $sql = "SELECT pub_id FROM {pub} P WHERE pub_id <> 1";
    $skip_existing = FALSE;
  }
  // Create citation for pubs that don't already have one
  else if ($options == 'new') {
    $sql = "
      SELECT pub_id
      FROM {pub} P
      WHERE
        (SELECT value
         FROM {pubprop} PB
         WHERE type_id = :type_id AND P.pub_id = PB.pub_id AND rank = 0) IS NULL
        AND  pub_id <> 1
    ";
    $skip_existing = TRUE;
  }

  $result = chado_query($sql, array(':type_id' => $citation_type_id));
  $counter_updated = 0;
  $counter_generated = 0;
  while ($pub = $result->fetchObject()) {
    $pub_arr = tripal_pub_get_publication_array($pub->pub_id, $skip_existing);
    if ($pub_arr) {
      $citation = tripal_pub_create_citation($pub_arr);
      print $citation . "\n\n";
      // Replace if citation exists. This condition is never TRUE if $skip_existing is TRUE
      if ($pub_arr['Citation']) {
        $sql = "
          UPDATE {pubprop} SET value = :value
          WHERE pub_id = :pub_id  AND type_id = :type_id AND rank = :rank
        ";
        chado_query($sql, array(':value' => $citation, ':pub_id' => $pub->pub_id,
          ':type_id' => $citation_type_id, ':rank' => 0));
        $counter_updated ++;
        // Generate a new citation
      } else {
        $sql = "
          INSERT INTO {pubprop} (pub_id, type_id, value, rank)
          VALUES (:pub_id, :type_id, :value, :rank)
        ";
        chado_query($sql, array(':pub_id' => $pub->pub_id, ':type_id' => $citation_type_id,
          ':value' => $citation, ':rank' => 0));
        $counter_generated ++;
      }
    }
  }
  print "$counter_generated citations generated. $counter_updated citations updated.\n";
}


/**
 * This function generates citations for publications.  It requires
 * an array structure with keys being the terms in the Tripal
 * publication ontology.  This function is intended to be used
 * for any function that needs to generate a citation.
 *
 * @param $pub
 *   An array structure containing publication details where the keys
 *   are the publication ontology term names and values are the
 *   corresponding details.  The pub array can contain the following
 *   keys with corresponding values:
 *     - Publication Type:  an array of publication types. a publication can have more than one type
 *     - Authors: a  string containing all of the authors of a publication
 *     - Journal Name:  a string containing the journal name
 *     - Journal Abbreviation: a string containing the journal name abbreviation
 *     - Series Name: a string containing the series (e.g. conference proceedings) name
 *     - Series Abbreviation: a string containing the series name abbreviation
 *     - Volume: the serives volume number
 *     - Issue: the series issue number
 *     - Pages: the page numbers for the publication
 *     - Publication Date:  A date in the format "Year Month Day"
 *
 * @return
 *   A text string containing the citation
 *
 * @ingroup tripal_pub
 */
function tripal_pub_create_citation($pub) {
  $citation = '';
  $pub_type = '';

  // An article may have more than one publication type. For example,
  // a publication type can be 'Journal Article' but also a 'Clinical Trial'.
  // Therefore, we need to select the type that makes most sense for
  // construction of the citation. Here we'll iterate through them all
  // and select the one that matches best.
  if (is_array($pub['Publication Type'])) {
    foreach ($pub['Publication Type'] as $ptype) {
      if ($ptype == 'Journal Article' ) {
        $pub_type = $ptype;
        break;
      }
      else if ($ptype == 'Conference Proceedings'){
        $pub_type = $ptype;
        break;
      }
      else if ($ptype == 'Review') {
        $pub_type = $ptype;
        break;
      }
      else if ($ptype == 'Book') {
        $pub_type = $ptype;
        break;
      }
      else if ($ptype == 'Letter') {
        $pub_type = $ptype;
        break;
      }
      else if ($ptype == 'Book Chapter') {
        $pub_type = $ptype;
        break;
      }
      else if ($ptype == "Research Support, Non-U.S. Gov't") {
        $pub_type = $ptype;
        // we don't break because if the article is also a Journal Article
        // we prefer that type
      }
    }
    // If we don't have a recognized publication type, then just use the
    // first one in the list.
    if (!$pub_type) {
      $pub_type = $pub['Publication Type'][0];
    }
  }
  else {
    $pub_type = $pub['Publication Type'];
  }
  //----------------------
  // Journal Article
  //----------------------
  if ($pub_type == 'Journal Article') {
    if (array_key_exists('Authors', $pub)) {
      $citation = $pub['Authors'] . '. ';
    }

    $citation .= $pub['Title'] .  '. ';

    if (array_key_exists('Journal Name', $pub)) {
      $citation .= $pub['Journal Name'] . '. ';
    }
    elseif (array_key_exists('Journal Abbreviation', $pub)) {
      $citation .= $pub['Journal Abbreviation'] . '. ';
    }
    elseif (array_key_exists('Series Name', $pub)) {
      $citation .= $pub['Series Name'] . '. ';
    }
    elseif (array_key_exists('Series Abbreviation', $pub)) {
      $citation .= $pub['Series Abbreviation'] . '. ';
    }
    if (array_key_exists('Publication Date', $pub)) {
      $citation .= $pub['Publication Date'];
    }
    elseif (array_key_exists('Year', $pub)) {
      $citation .= $pub['Year'];
    }
    if (array_key_exists('Volume', $pub) or array_key_exists('Issue', $pub) or array_key_exists('Pages',$pub)) {
      $citation .= '; ';
    }
    if (array_key_exists('Volume', $pub)) {
      $citation .= $pub['Volume'];
    }
    if (array_key_exists('Issue', $pub)) {
      $citation .= '(' . $pub['Issue'] . ')';
    }
    if (array_key_exists('Pages', $pub)) {
      if (array_key_exists('Volume', $pub)) {
        $citation .= ':';
      }
      $citation .= $pub['Pages'];
    }
    $citation .= '.';
  }
  //----------------------
  // Review
  //----------------------
  if ($pub_type == 'Review') {
    if (array_key_exists('Authors', $pub)) {
      $citation = $pub['Authors'] . '. ';
    }

    $citation .= $pub['Title'] .  '. ';

    if (array_key_exists('Journal Name', $pub)) {
      $citation .= $pub['Journal Name'] . '. ';
    }
    elseif (array_key_exists('Journal Abbreviation', $pub)) {
      $citation .= $pub['Journal Abbreviation'] . '. ';
    }
    elseif (array_key_exists('Series Name', $pub)) {
      $citation .= $pub['Series Name'] . '. ';
    }
    elseif (array_key_exists('Series Abbreviation', $pub)) {
      $citation .= $pub['Series Abbreviation'] . '. ';
    }
    if (array_key_exists('Publication Date', $pub)) {
      $citation .= $pub['Publication Date'];
    }
    elseif (array_key_exists('Year', $pub)) {
      $citation .= $pub['Year'];
    }
    if (array_key_exists('Volume', $pub) or array_key_exists('Issue', $pub) or array_key_exists('Pages',$pub)) {
      $citation .= '; ';
    }
    if (array_key_exists('Volume', $pub)) {
      $citation .= $pub['Volume'];
    }
    if (array_key_exists('Issue', $pub)) {
      $citation .= '(' . $pub['Issue'] . ')';
    }
    if (array_key_exists('Pages', $pub)) {
      if (array_key_exists('Volume', $pub)) {
        $citation .= ':';
      }
      $citation .= $pub['Pages'];
    }
    $citation .= '.';
  }
  //----------------------
  // Research Support, Non-U.S. Gov't
  //----------------------
  elseif ($pub_type == "Research Support, Non-U.S. Gov't") {
    if (array_key_exists('Authors', $pub)) {
      $citation = $pub['Authors'] . '. ';
    }

    $citation .= $pub['Title'] .  '. ';

    if (array_key_exists('Journal Name', $pub)) {
      $citation .= $pub['Journal Name'] . '. ';
    }
    if (array_key_exists('Publication Date', $pub)) {
      $citation .= $pub['Publication Date'];
    }
    elseif (array_key_exists('Year', $pub)) {
      $citation .= $pub['Year'];
    }
    $citation .= '.';
  }
  //----------------------
  // Letter
  //----------------------
  elseif ($pub_type == 'Letter') {
    if (array_key_exists('Authors', $pub)) {
      $citation = $pub['Authors'] . '. ';
    }

    $citation .= $pub['Title'] .  '. ';
    if (array_key_exists('Journal Name', $pub)) {
      $citation .= $pub['Journal Name'] . '. ';
    }
    elseif (array_key_exists('Journal Abbreviation', $pub)) {
      $citation .= $pub['Journal Abbreviation'] . '. ';
    }
    elseif (array_key_exists('Series Name', $pub)) {
      $citation .= $pub['Series Name'] . '. ';
    }
    elseif (array_key_exists('Series Abbreviation', $pub)) {
      $citation .= $pub['Series Abbreviation'] . '. ';
    }
    if (array_key_exists('Publication Date', $pub)) {
      $citation .= $pub['Publication Date'];
    }
    elseif (array_key_exists('Year', $pub)) {
      $citation .= $pub['Year'];
    }
    if (array_key_exists('Volume', $pub) or array_key_exists('Issue', $pub) or array_key_exists('Pages',$pub)) {
      $citation .= '; ';
    }
    if (array_key_exists('Volume', $pub)) {
      $citation .= $pub['Volume'];
    }
    if (array_key_exists('Issue', $pub)) {
      $citation .= '(' . $pub['Issue'] . ')';
    }
    if (array_key_exists('Pages', $pub)) {
      if (array_key_exists('Volume', $pub)) {
        $citation .= ':';
      }
      $citation .= $pub['Pages'];
    }
    $citation .= '.';
  }
  //-----------------------
  // Conference Proceedings
  //-----------------------
  elseif ($pub_type == 'Conference Proceedings') {
    if (array_key_exists('Authors', $pub)) {
      $citation = $pub['Authors'] . '. ';
    }

    $citation .= $pub['Title'] .  '. ';
    if (array_key_exists('Conference Name', $pub)) {
      $citation .= $pub['Conference Name'] . '. ';
    }
    elseif (array_key_exists('Series Name', $pub)) {
      $citation .= $pub['Series Name'] . '. ';
    }
    elseif (array_key_exists('Series Abbreviation', $pub)) {
      $citation .= $pub['Series Abbreviation'] . '. ';
    }
    if (array_key_exists('Publication Date', $pub)) {
      $citation .= $pub['Publication Date'];
    }
    elseif (array_key_exists('Year', $pub)) {
      $citation .= $pub['Year'];
    }
    if (array_key_exists('Volume', $pub) or array_key_exists('Issue', $pub) or array_key_exists('Pages',$pub)) {
      $citation .= '; ';
    }
    if (array_key_exists('Volume', $pub)) {
      $citation .= $pub['Volume'];
    }
    if (array_key_exists('Issue', $pub)) {
      $citation .= '(' . $pub['Issue'] . ')';
    }
    if (array_key_exists('Pages', $pub)) {
      if (array_key_exists('Volume', $pub)) {
        $citation .= ':';
      }
      $citation .= $pub['Pages'];
    }
    $citation .= '.';
  }
  //-----------------------
  // Default
  //-----------------------
  else {
    if (array_key_exists('Authors', $pub)) {
      $citation = $pub['Authors'] . '. ';
    }
    $citation .= $pub['Title'] .  '. ';
    if (array_key_exists('Series Name', $pub)) {
      $citation .= $pub['Series Name'] . '. ';
    }
    elseif (array_key_exists('Series Abbreviation', $pub)) {
      $citation .= $pub['Series Abbreviation'] . '. ';
    }
    if (array_key_exists('Publication Date', $pub)) {
      $citation .= $pub['Publication Date'];
    }
    elseif (array_key_exists('Year', $pub)) {
      $citation .= $pub['Year'];
    }
    if (array_key_exists('Volume', $pub) or array_key_exists('Issue', $pub) or array_key_exists('Pages',$pub)) {
      $citation .= '; ';
    }
    if (array_key_exists('Volume', $pub)) {
      $citation .= $pub['Volume'];
    }
    if (array_key_exists('Issue', $pub)) {
      $citation .= '(' . $pub['Issue'] . ')';
    }
    if (array_key_exists('Pages', $pub)) {
      if (array_key_exists('Volume', $pub)) {
        $citation .= ':';
      }
      $citation .= $pub['Pages'];
    }
    $citation .= '.';
  }

  return $citation;
}

/**
 * Retrieves the minimal information to uniquely describe any publication.
 *
 * @param $pub
 *   A publication object as created by chado_generate_var()
 *
 * @return
 *   An array with the following keys:  'Citation', 'Abstract', 'Authors',
 *   'URL'. All keys are term names in the Tripal Publication Ontology :TPUB.
 */
function tripal_get_minimal_pub_info($pub) {
  // Chado has a null pub as default.  We don't return anything for this.
  if ($pub->uniquename == 'null') {
    return array();
  }

  // expand the title
  $pub = chado_expand_var($pub, 'field', 'pub.title');
  $pub = chado_expand_var($pub, 'field', 'pub.volumetitle');

  // get the citation
  $values = array(
    'pub_id' => $pub->pub_id,
    'type_id' => array(
      'name' => 'Citation',
    ),
  );
  $options = array(
    'include_fk' => array(
    ),
  );
  $citation = chado_generate_var('pubprop', $values, $options);
  $citation = chado_expand_var($citation, 'field', 'pubprop.value');
  $citation = $citation->value;

  // get the abstract
  $values = array(
    'pub_id' => $pub->pub_id,
    'type_id' => array(
      'name' => 'Abstract',
    ),
  );
  $options = array(
    'include_fk' => array(
    ),
  );
  $abstract = chado_generate_var('pubprop', $values, $options);
  $abstract = chado_expand_var($abstract, 'field', 'pubprop.value');
  $abstract_text = '';
  if ($abstract) {
    $abstract_text = htmlspecialchars($abstract->value);
  }

  // get the author list
  $values = array(
    'pub_id' => $pub->pub_id,
    'type_id' => array(
      'name' => 'Authors',
    ),
  );
  $options = array(
    'include_fk' => array(
    ),
  );
  $authors = chado_generate_var('pubprop', $values, $options);
  $authors = chado_expand_var($authors, 'field', 'pubprop.value');
  $authors_list = 'N/A';
  if ($authors) {
    $authors_list = $authors->value;
  }

  // get the first database cross-reference with a url
  $options = array('return_array' => 1);
  $pub = chado_expand_var($pub, 'table', 'pub_dbxref', $options);
  $dbxref = NULL;
  if ($pub->pub_dbxref) {
    foreach ($pub->pub_dbxref as $index => $pub_dbxref) {
      if ($pub_dbxref->dbxref_id->db_id->urlprefix) {
        $dbxref = $pub_dbxref->dbxref_id;
      }
    }
  }

  // get the URL
  // get the author list
  $values = array(
    'pub_id' => $pub->pub_id,
    'type_id' => array(
      'name' => 'URL',
    ),
  );

  $options = array(
    'return_array' => 1,
    'include_fk' => array(),
  );
  $urls = chado_generate_var('pubprop', $values, $options);
  $urls = chado_expand_var($urls, 'field', 'pubprop.value');
  $url = '';
  if (count($urls) > 0) {
    $url = $urls[0]->value;
  }
  return array(
    'Title' => $pub->title,
    'Citation' => $citation,
    'Abstract' => $abstract_text,
    'Authors' => $authors_list,
    'URL' => $url,
  );
}