udm_citeproc/udm_citeproc.module

<?php

/**
 * @file
 * Implementation of some citeproc alters.
 */

/**
 * Implements citeproc_convert_mods_to_citeproc_jsons_alter
 *
 * @param $output
 * @param $mods
 */
function udm_citeproc_convert_mods_to_citeproc_jsons_alter(&$output, $mods) {
  module_load_include('inc', 'citeproc', 'includes/converter');  
  $output['type'] = udm_citeproc_local_to_csl_type($mods);
   if(empty($output['author'])){
     $names = udm_convert_mods_to_citeproc_json_names($mods);
     $output = array_merge($output, $names);
  }   
}

/**
 * Gets the type property for the Citation.
 *
 * There are a number of name-type vars which may be populated.
 * We will concern ourselves with the following:
 *  1. Author
 *  2. Editor
 *  3. Translator
 * There is no CSL var for "contributor", we will treat them as additional
 * authors.
 *
 * Each name may be either "corporate" or "given name / family name"
 *
 * As it's unlikely we'll have participles, suffixes, etc properly parsed
 * out, we will always pass the ("parse-names" : "true") flag with personal
 * names.
 *
 * mods namepart types (given, family) correspond to citeproc elements,
 * however, more precise mods elements (nonsort, etc.) do not.
 * @todo Make all name handling better.
 *
 * NAME(s) of RELATED ITEMS (host, series)
 * Zotero vs Bibutils do this a bit differently, but in bibutils it's common for
 * the editor of a book (containing a chapter, which is our bibliographic item)
 * to be associated with the relatedItem(host).
 * Also It'S A Shot In The Dark, But
 * Relateditem(series)->name->role->roleTerm=editor is plausible.
 *
 * Note also this section is *highly* repetitive of the section above and this
 * should probably be generalized, but for now a strict procedural reckoning
 * will have to suffice. The difference is in the very last section, where the
 * appropriate cs:names type is specified.
 *
 * @param SimpleXMLElement $mods
 *   A MODS document.
 *
 * @return string
 *   The type property for the Citation.
 */
function udm_convert_mods_to_citeproc_json_names(SimpleXMLElement $mods) {
  $queries = array(
    0 => array(
      // Path.
      variable_get('islandora_scholar_xpaths_local_author', '//mods:mods[1]/mods:name[normalize-space(mods:namePart)]'),
      // Default Role.
      'author',
      // Valid Roles.
      array(
        'editor' => 'editor',
        'translator' => 'translator',
        'interviewer' => 'interviewer',
        'composer' => 'composer',
        'original' => 'original-author',
        'recipient' => 'recipient',
        'author' => 'author',
      ),
    ),
    1 => array(
      variable_get('islandora_scholar_xpaths_local_container-author', '//mods:mods[1]/mods:relatedItem[@type="host"]/mods:name'),
      'container-author',
      array(
        'editor' => 'editor',
        'translator' => 'translator',
        'author' => 'container-author',
      ),
    ),
    2 => array(
      variable_get('islandora_scholar_xpaths_local_collection-editor', '//mods:mods[1]/mods:relatedItem[@type="series"]/mods:name'),
      NULL,
      array(
        'editor' => 'collection-editor',
      ),
    ),
  );
  $output = array();
  foreach ($queries as $query) {
    list($path, $default_role, $valid_roles) = $query;
    $names = $mods->xpath($path);
    if (!empty($names)) {
      foreach ($names as $name) {

        add_mods_namespace($name);
        $role = udm_convert_mods_to_citeproc_json_name_role($name, $valid_roles, $default_role);
        if ($role !== FALSE) {
          $parsed_name = udm_convert_mods_to_citeproc_json_name($name);
         
          if ($parsed_name) {
            $output[$role][] = $parsed_name;
          }
        }
      }
    }
  }
  return $output;
}

/**
 * Gets the array repersentation of the javascript Citation's name properties.
 *
 * @param SimpleXMLElement $name
 *   A name element.
 *
 * @return array
 *   An array that embodies the name properties of a Citation javascript object.
 */
function udm_convert_mods_to_citeproc_json_name(SimpleXMLElement $name) {
  $type = (string) $name->attributes()->type;
  $output = ($type == 'personal') ?
      udm_convert_mods_to_citeproc_json_name_personal($name) :
      udm_convert_mods_to_citeproc_json_name_corporate($name);
  
  $output = array_map('trim', $output);
  $output = array_map('ucfirst', $output);
  $output = array_filter($output);
  return $output;
}

/**
 * Gets the array repersentation of the Citation's personal name properties.
 *
 * @param SimpleXMLElement $name
 *   A name element.
 *
 * @return array
 *   An array that embodies the name properties of a Citation javascript object.
 */
function udm_convert_mods_to_citeproc_json_name_personal(SimpleXMLElement $name) {
  $output = array();
  $name_parts = $name->xpath('mods:namePart');

  foreach ($name_parts as $name_part) {
    $type = (string) $name_part->attributes()->type;
    $content = (string) $name_part;

    // If the type is empty, it is a combined name last, first.
    if (empty($type)) {
      // Filter empty things out of the output, so we may potentially replace
      // them.
      $output = array_filter(array_map('trim', $output));

      $names = array_map('trim', explode(',', $content));
      $exploded = array();
      $exploded['family'] = array_shift($names);
      $exploded['given'] = array_shift($names);

      // If we appear to have a family (and potentially given) name part in our
      // combined field and are not currently outputting them, make them get
      // output.
      // XXX: May combine unexpectedly in the case a "given" typed part name is
      // specified, and our untyped field only contains a "family" name.
      $non_empty = array_filter($exploded);
      $parts_not_present = array_diff_key($non_empty, $output);
      if (!empty($parts_not_present)) {
        $output = $non_empty + $output;
      }
    }
    else {
      // Throw a period after single character values, as they likely represent
      // initials.
      $content .= (strlen($content) == 1) ? '. ' : ' ';
      $output[$type] = isset($output[$type]) ?
          $output[$type] . $content :
          $content;
    }
  }
  return $output;
}

/**
 * Gets the array repersentation of the Citation's corporate name properties.
 *
 * @param SimpleXMLElement $name
 *   A name element.
 *
 * @return array
 *   An array that embodies the name properties of a Citation javascript object.
 */
function udm_convert_mods_to_citeproc_json_name_corporate(SimpleXMLElement $name) {
  $output = array();
  $name_parts = $name->xpath(variable_get('islandora_scholar_xpaths_local_author', 'mods:namePart'));
  foreach ($name_parts as $name_part) {
    $content = (string) $name_part . ' ';
    $output['literal'] = isset($output['literal']) ? $output['literal'] . $content : $content;
  }
  return $output;
}



/**
 * Gets the role for the given name element.
 *
 * If no role can be determined it returns 'author' as a default.
 *
 * @param SimpleXMLElement $name
 *   A MODS name element.
 * @param array $valid_roles
 *   A map of mods role names to their citeproc equivalents.
 * @param string $default_role
 *   The role to use if a valid role is not found.
 *
 * @return string
 *   Gets the role of the given name.
 */
function udm_convert_mods_to_citeproc_json_name_role(SimpleXMLElement $name, array $valid_roles, $default_role) {
  module_load_include('inc', 'citeproc', 'includes/marcrelator_conversion');
  $role_term = $name->xpath(variable_get('islandora_scholar_xpaths_role_term', 'mods:role/mods:roleTerm'));
  
  if (isset($role_term[0])) {
    $role = strtolower((string) $role_term[0]);
    $role_term = $role_term[0];
  }
  else {
    $role = NULL;
  }

  if ($role) {
    $role_authority = (string) $role_term->attributes()->authority;
    $role_type = (string) $role_term->attributes()->type;

    if ($role_authority == 'marcrelator' && $role_type == 'code') {
      $role = marcrelator_code_to_term($role);
    }
    $role = ($role == 'auteur') ? 'author' : $role;

    return array_key_exists($role, $valid_roles) ? $valid_roles[$role] : FALSE;
  }
  return $default_role;
}

/**
 * Converts our local types that mostly came from refworks to CSL types.
 *
 * @param SimpleXML $mods
 * @return string
 *   The CSL type.
 */
function udm_citeproc_local_to_csl_type($mods) {
  $genre = $mods->xpath("//mods:mods[1]/mods:genre");

  if(empty($genre)){
    return convert_mods_to_citeproc_json_type($mods);
  }
  $term = strtolower(trim((string)$genre[0]));
  switch ($term) {
    case 'article scientifique':
    case '':
      return 'article-journal';
    case 'book chapter':
    case 'book section':
    case 'book, section':
    case 'chapitre de livre':
      return 'chapter';
    case 'book, whole':
    case 'livre':
      return 'book';
    case 'conference proceedings':
    case 'conference abstract':
    case 'Contribution à un congrès, un colloque, une conférence':
      return 'paper-conference';
    default:
      return convert_mods_to_citeproc_json_type($mods);
  }
}