crossref_unixref_parser.inc in Bibliography Module 6

File

crossref_unixref_parser.inc

<?php

function unixref_startElement($parser, $name, $attrs) {
  global $node, $element, $attribute, $auth_category, $contrib_count, $org_count;
  switch ($name) {
    case 'doi_record':
      $node = array();
      $element = $name;
      break;
    case 'book':
    case 'journal':
    case 'standard':
    case 'conference':
    case 'report-paper':
    case 'dissertation':
    case 'database':
    case 'sa_component':
      $node['biblio_type'] = _unixref_type_map($name);
      $element = $name;
      break;
    case 'journal_article':
    case 'conference_paper':
    case 'content_item':
    case 'report-paper_metadata':
    case 'standard_metadata':
    case 'database_date':
    case 'component':
      $node['year'] = '';
      $node['doi'] = '';
      $element = $name;
      break;
    case 'person_name':
      $auth_category = _unixref_get_contributor_category($attrs['contributor_role']);
      if (!isset($contrib_count)) {
        $contrib_count = 0;
      }
      $element = $name;
      break;
    case 'organization':
      if (!isset($org_count)) {
        $org_count = 0;
      }
      $element = $name;
      break;
    case 'issn':
      if ($attrs['media_type'] == 'print') {
        $attribute = 'issn_print';
      }
      $element = $name;
      break;
    case 'isbn':
      if ($attrs['media_type'] == 'print') {
        $attribute = 'isbn_print';
      }
      $element = $name;
      break;
    case 'i':

    // HTML font style tags
    case 'b':
    case 'u':
    case 'sub':
    case 'sup':
      unixref_characterData(NULL, ' <' . $name . '>');
      break;
    default:
      $element = $name;
  }
}
function unixref_decode(&$item, $key) {
  $item = html_entity_decode($item, NULL, 'UTF-8');
}
function unixref_endElement($parser, $name) {
  global $node, $nids, $element, $attribute, $terms, $batch_proc, $session_id, $save_node, $auth_category, $contrib_count, $org_count;
  switch ($name) {
    case 'doi_record':
      array_walk_recursive($node, 'unixref_decode');
      $nid = biblio_save_node($node, $batch, $session_id, $save_node);
      if (isset($nid)) {
        $nids[] = $nid;
      }
      break;
    case 'person_name':
      $node['biblio_contributors'][$auth_category][$contrib_count]['auth_type'] = _biblio_get_auth_type($auth_category, $node['biblio_type']);
      $node['biblio_contributors'][$auth_category][$contrib_count]['name'] = $node['biblio_contributors'][$auth_category][$contrib_count]['lastname'];
      if (isset($node['biblio_contributors'][$auth_category][$contrib_count]['firstname'])) {
        $node['biblio_contributors'][$auth_category][$contrib_count]['name'] .= ', ' . $node['biblio_contributors'][$auth_category][$contrib_count]['firstname'];
      }
      $auth_category = '';
      $contrib_count++;
      break;
    case 'organization':
      $node['biblio_contributors'][5][$org_count]['auth_type'] = _biblio_get_auth_type(5, $node['biblio_type']);
      $org_count++;
      break;
    case 'pages':
      if (isset($node['biblio_first_page'])) {
        $node['biblio_pages'] = $node['biblio_first_page'];
      }
      if (isset($node['biblio_last_page'])) {
        $node['biblio_pages'] .= ' - ' . $node['biblio_last_page'];
      }
      break;
    case 'publication_date':
      break;
    case 'journal_issue':
      $node['biblio_date'] = (!empty($node['month']) ? $node['month'] . '/' : '') . $node['year'];
      break;
    case 'journal_article':
    case 'conference_paper':
    case 'content_item':
    case 'report-paper_metadata':
    case 'standard_metadata':
    case 'database_date':
    case 'component':
      $node['biblio_year'] = $node['year'];
      $node['biblio_doi'] = $node['doi'];
      break;
    case 'issn':
      if ($attribute == 'issn_print') {
        $node['biblio_issn'] = $node['issn'];
      }
      $node['issn'] = '';
      break;
    case 'isbn':
      if ($attribute == 'isbn_print') {
        $node['biblio_isbn'] = $node['isbn'];
      }
      $node['isbn'] = '';
      break;
    case 'i':

    // HTML font style tags
    case 'b':
    case 'u':
    case 'sub':
    case 'sup':
      unixref_characterData(NULL, '</' . $name . '> ');
      break;
    default:
  }
}
function unixref_characterData($parser, $data) {
  global $node, $element, $auth_category, $contrib_count, $org_count;
  $data = htmlspecialchars_decode($data);
  if (trim($data)) {
    switch ($element) {
      case 'surname':
        $node['biblio_contributors'][$auth_category][$contrib_count]['lastname'] .= $data;
        break;
      case 'given_name':
        $node['biblio_contributors'][$auth_category][$contrib_count]['firstname'] .= $data;
        break;
      case 'suffix':
        $node['biblio_contributors'][$auth_category][$contrib_count]['suffix'] .= $data;
        break;
      case 'affiliation':
        $node['biblio_contributors'][$auth_category][$contrib_count]['affiliation'] .= $data;
        break;
      case 'organization':
        $node['biblio_contributors'][5][$org_count]['name'] .= $data;
        break;
      case 'year':
        $node['year'] = $data;
        break;
      case 'month':
        $node['month'] = $data;
        break;
      case 'day':
        $node['day'] = $data;
        break;
      default:
        if ($field = _unixref_field_map(trim($element))) {
          $node[$field] .= $data;
        }
    }
  }
}

/*
 * map a unixref XML field to a biblio field
 */
function _unixref_field_map($field) {
  static $field_map = array();
  if (empty($field_map)) {
    module_load_include('inc', 'biblio', 'biblio.type.mapper');
    $field_map = biblio_get_field_map('unixref');
  }
  return isset($field_map[$field]) ? $field_map[$field] : FALSE;
}
function _unixref_type_map($type) {
  static $map = array();
  if (empty($map)) {
    module_load_include('inc', 'biblio', 'biblio.type.mapper');
    $map = biblio_get_type_map('unixref');
  }
  return isset($map[$type]) ? $map[$type] : 129;

  //return the biblio type or 129 (Misc) if type not found
}
function _unixref_get_contributor_category($role) {
  if ($role == 'author') {
    return 1;
  }
  if ($role == 'editor') {
    return 2;
  }
  if ($role == 'chair') {
    return 3;
  }
  if ($role == 'translator') {
    return 4;
  }
  return NULL;
}

Functions

Name	Description
unixref_characterData
unixref_decode
unixref_endElement
unixref_startElement
_unixref_field_map
_unixref_get_contributor_category
_unixref_type_map